Article

Image Shadow Removal Using Pulse Coupled Neural Network

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

This paper introduces an approach for image shadow removal by using pulse coupled neural network (PCNN), based on the phenomena of synchronous pulse bursts in the animal visual cortexes. Two shadow-removing criteria are proposed. These two criteria decide how to choose the optimal parameter (the linking strength beta). The computer simulation results of shadow removal based on PCNN show that if these two criteria are satisfied, shadows are removed completely and the shadow-removed images are almost as the same as the original nonshadowed images. The shadow removal results are independent of changes of intensities of shadows in some range and variations of the places of shadows. When the first criterion is satisfied, even if the second criterion is not satisfied, as to natural grey images that have abundant grey levels, shadows also can be removed and PCNN shadow-removed images retain the shapes of the objects in original images. These two criteria also can be used for color images by dividing a color image into three channels (R, G, B). For shadows varying drastically, such as the noisy points in images, these two criteria are still right, but difficult to satisfy. Therefore, this approach can efficiently remove shadows that do not include the random noise.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Johnson et al. had emphasized the importance of their PCNN factoring model [27,[121][122][123], and even the corresponding pseudocode is given in [27,121]. The factoring model can be applied to shadow removal in an image [27,124], image enhancement [123] and multi-scale analysis [122,125]. The time matrix of PCNN is applied to image enhancement by using mechanism similar to the factoring model [90,109]. ...
... which is named as sigmoidal-linking in [27], quantizedlinking in [30], and unit-linking in [124]. The sigmoidal-linking renders the network be simpler than the classical PCNN [27,30,124] because it is usually given by, The intersecting cortical model only has the feeding wave, and the spiking cortical and sigmoidal-linking models only have the linking wave. ...
... which is named as sigmoidal-linking in [27], quantizedlinking in [30], and unit-linking in [124]. The sigmoidal-linking renders the network be simpler than the classical PCNN [27,30,124] because it is usually given by, The intersecting cortical model only has the feeding wave, and the spiking cortical and sigmoidal-linking models only have the linking wave. The feature-linking model not only Computational Mechanisms of Pulse-Coupled Neural Networks: A Comprehensive Review reduces the parameters but also two types of waves are contained in network. ...
Article
Pulse-coupled neural networks (PCNN) have an inherent ability to process the signals associated with the digital visual images because it is inspired from the neuronal activity in the primary visual area, V1, of the neocortex. This paper provides insight into the internal operations and behaviors of PCNN, and reveals the way how PCNN achieves good performance in digital image processing. The various properties of PCNN are categorized into a novel three-dimensional taxonomy for image processing mechanisms. The first dimension specifies the time matrix of PCNN, the second dimension captures the firing rate of PCNN, and the third dimension is the synchronization of PCNN. Many examples of processing mechanisms are provided to make it clear and concise.
... One is shadow-region location, and the other is illuminance compensation. To locate the shadow regions, interactive methods [11,12,13,14,15,16] and automatic methods [17,18,19,20,21] have been studied. In [11], an image was interactively partitioned into the shadow region, non-shadow region and penumbra, which were fed into a Bayesian network as priori probability, to seek an optimal compensation in illuminance. ...
... In [16], interactive graph cut was utilized to segment the shadow regions from background. While interactive methods often fall short in efficiency, machine-learning based automatic methods have been exploited in [19,17]. ...
... In the following year, Ranganath et al. demonstrated that PCNN exhibits outstanding performance for image smoothing, image segmentation and feature extraction [6]. Since then, the PCNN has found extensive use in image processing areas such as pattern recognition [3], feature extraction [5], image segmentation [6], image shadow removal [7], image encryption [8], and object recognition [9,10]. ...
Article
Full-text available
This paper surveys recent advances in pulse-coupled neural networks (PCNNs) and their applications in image processing. The PCNN is a neurology-inspired neural network model that aims to imitate the information analysis process of the biological cortex. In recent years, many PCNN-derived models have been developed. Research aims with respect to these models can be divided into three categories: (1) to reduce the number of manual parameters, (2) to achieve better real cortex imitation performance, and (3) to combine them with other methodologies. We provide a comprehensive and schematic review of these novel PCNN-derived models. Moreover, the PCNN has been widely used in the image processing field due to its outstanding information extraction ability. We review the recent applications of PCNN-derived models in image processing, providing a general framework for the state of the art and a better understanding of PCNNs with applications in image processing. In conclusion, PCNN models are developing rapidly, and it is projected that more applications of these novel emerging models will be seen in future.
... A few years later, in 1994, Johnson et al. designed a PCNN model based on Eckhorn's original cortical model for the imageprocessing area [34,35]. Since then, PCNN has been widely used in many image processing applications such as image shadow removal [36], feature extraction [34], pattern recognition [32], image segmentation [35], and object recognition [37,38]. ...
Article
Full-text available
In this study, the anti-noise performance of a pulse-coupled neural network (PCNN) was investigated in the neutron and gamma-ray ( n − γ\gamma γ ) discrimination field. The experiments were conducted in two groups. In the first group, radiation pulse signals were pre-processed using a Fourier filter to reduce the original noise in the signals, whereas in the second group, the original noise was left untouched to simulate an extremely high-noise scenario. For each part, artificial Gaussian noise with different intensity levels was added to the signals prior to the discrimination process. In the aforementioned conditions, the performance of the PCNN was evaluated and compared with five other commonly used methods of n − γ\gamma γ discrimination: (1) zero crossing, (2) charge comparison, (3) vector projection, (4) falling edge percentage slope, and (5) frequency gradient analysis. The experimental results showed that the PCNN method significantly outperforms other methods with outstanding FoM\mathrm{FoM} FoM -value at all noise levels. Furthermore, the fluctuations in FoM\mathrm{FoM} FoM -value of PCNN were significantly better than those obtained via other methods at most noise levels and only slightly worse than those obtained via the charge comparison and zero-crossing methods under extreme noise conditions. Additionally, the changing patterns and fluctuations of the FoM\mathrm{FoM} FoM -value were evaluated under different noise conditions. Hence, based on the results, the parameter selection strategy of the PCNN was presented. In conclusion, the PCNN method is suitable for use in high-noise application scenarios for n − γ\gamma γ discrimination because of its stability and remarkable discrimination performance. It does not rely on strict parameter settings and can realize satisfactory performance over a wide parameter range.
... The pulse-coupled neural networks (PCNN) are biology-inspired models derived from the mammalian visual cortex (Chen and Shibata, 2010). Owing to the simple calculation of the PCNN, it is widely applied in image segmentation , image shadow removal (Gu et al., 2005), object detection , and image fusion (Li and Zhao, 2020). ...
Article
Full-text available
The Izhikevich (IZH) spiking neuron model can display spiking and bursting behaviors of neurons. Based on the switching property and bio-plausibility of the memristor, the memristive Izhikevich (MIZH) spiking neuron model is built. Firstly, the MIZH spiking model is introduced and used to generate 23 spiking patterns. We compare the 23 spiking patterns produced by the IZH and MIZH spiking models. Secondly, the MIZH spiking model actively reproduces various neuronal behaviors, including the excitatory cortical neurons, the inhibitory cortical neurons, and other cortical neurons. Finally, the collective dynamic activities of the MIZH neuronal network are performed, and the MIZH oscillatory network is constructed. Experimental results illustrate that the constructed MIZH spiking neuron model performs high firing frequency and good frequency adaptation. The model can easily simulate various spiking and bursting patterns of distinct neurons in the brain. The MIZH neuronal network realizes the synchronous and asynchronous collective behaviors. The MIZH oscillatory network can memorize and retrieve the information patterns correctly and efficiently with high retrieval accuracy.
... The core mechanism of PCNN is that the fired neurons capture the synchronous emission of neurons in a certain region through spatial proximity and brightness similarity [3]. In the last decade, PCNN has developed rapidly in many aspects of image processing, such as image shadow removal [4], feature extraction [5], image segmentation [6,7], etc. ...
... Ha aumentado el interés en el uso de PCNN para diversas aplicaciones, como el reconocimiento de objetivos [8], el procesamiento de imágenes [9], el reconocimiento de patrones [10] y reducir el efecto de ruido gaussiano [11]. Algunas investigaciones recientes muestran que la dinámica espacio temporal de las PCNN proporciona una buena capacidad computacional para resolver algunos problemas de optimización. ...
... Intersecting cortical model [13,31] only has the feeding wave in its neural activity, with b ¼ 0. Spiking cortical model [29,66] simplifies the expression of PCNN membrane potential from three equations to one and sets the external stimulus S to the feeding input. Sigmoidal-linking model [21,29] defines L ij ðnÞ as a sigmoid function; meanwhile, it eliminates the feeding wave. Feature-linking model [68] is a synthesis of intersecting cortical model and sigmoidal-linking model. ...
Article
Full-text available
Pulse-coupled neural network (PCNN) is a powerful unsupervised learning model with many parameters to be determined empirically. In particular, the weight matrix is invariable in the iterative process, which is inconsistent with the actual biological system. Based on the existing research foundation of biology and neural network, we propose a spike-synchronization-dependent plasticity (SSDP) rule. In this paper, the mathematical model and algorithm of SSDP are presented. Furthermore, a novel memristor-based circuit model of SSDP is designed. Finally, experimental results demonstrate that SSDP has greatly improved the image processing capabilities of PCNN.
... The Unit-linking PCNN [14] simplifies both the feeding inputs and linking inputs to compact forms. ...
Article
In this paper, recent pulse coupled neural networks (PCNN) model’s development, especially in the fields related to the image processing, were surveyed. Our review aims to provide a comprehensive and systematic analysis of selected researches from past few decades, having powerful methods to infer the area of study. In this paper, all selected references are categorized in three groups, on the basis of neurons structure, parameters setting, and the inherent characteristics of PCNN. Various applications of these models were mentioned and underlying difficulties, limitations, merits and disadvantages were discussed in applying these models. The researchers will find it helpful to choose and use the appropriate model for a better application. Read online: http://rdcu.be/FvPZ
... integrated into the bhattacharyya coefficient.The proposed approach reaches almost 10percent retrieval quality comparing to other recently reported state-of-the-art methods. Trong -Thuc Hoang , Ngoc-Hung Nguyen , Xuan-Thuan Nguyen and Trong-Tu Bui [28] proposed two hardware architecture based on PCNN for image feature vector extraction . Based on these architecture the demonstration recognition system including a camera ,a feature vector generator ,a search engine and a DVI controller has been built and tested successfully on FPGA chips.M. Monica Subashini , Sarat Kumar Sahoo [16] proposed Pulse coupled neural networks and its applications thatsurveys the extensive usage of pulse coupled neural networks. ...
Article
Full-text available
We propose a pulse Coupled Neural Networks for identifying Near-Duplicate images. Temporal series of pulsed output are generated by using pulse coupled neural networks which carries the information about the input image which in turn used to detect NEAR-DUPLICATE images. Near-duplicate images are generated by taking photos of the same scene under different conditions in illuminations, resolutions and so on .Besides they can be formed by modifying the unique images using some transformations(i.e)image rotation ,image mapping, scaling etc. Identifying the near duplicate images plays an important role in many applications such as copyright protection, plagiarism in images, forged images.
... Taking advantage of the powerful segmentation ability of the PCNN model and combining with the shadow attributes, a novel shadow elimination method based on the improved PCNN model was put forward [20]. Based on the phenomena of synchronous pulse bursts in animal visual cortex, [21] introduced a PCNN approach for image shadow removal. [22] proposed a PCNN method improved by characters of lateral inhibition of human vision and coefficient of variation for shadow detection. ...
Conference Paper
A novel double-threshold pulse coupled neural networks (DT-PCNN) is proposed and applied to shadow detection. It attempts to reduce the false detection of shadows in a single image where the hue and brightness of some non-shadow regions are similar to or even lower than those of shadows. Shadows whose intensity and hue fall in between those of the scene and objectives are often viewed as non-shadows by the single dynamic threshold of PCNN. Moreover, entities with similar or darker hue and intensity may be wrongly classified as shadows. To solve this problem, two different dynamic thresholds that iteratively alter are designed. The upper and lower limits of detecting shadows are determined respectively by a higher threshold that decreases iteratively and a lower one that increases iteratively. The detection result is obtained by a fusion of two detection components. Experimental results demonstrate that compared to other tested methods, the misclassifications are significantly reduced and the shadows are more accurately extracted. Index Terms— Shadow detection, double-threshold pulse coupled neural networks (DTPCNN)
... (Eckhorn, 1999). Until now, PCNN has mainly been applied in two-dimensional image processing tasks (Yu and Zhang, 2004) and (Gu et al., 2005). Here, a novel 3-D PCNN model (Cao et al., 2007) is constructed to perform three-dimensional optical flow optimization. ...
Conference Paper
This paper presents the application of a novel optical flow optimization algorithm for a comprehensive on-road vehicle motion analysis. Optical flow, which contains abundant local motion information, has been extensively studied for vehicle motion estimation in the last decades. How to generate a reliable optical flow at a low computation cost is always a challenging task. The primary aim of this paper is to enhance the accuracy and efficiency of optical flow estimation for a reliable vehicle motion analysis. In the paper, an innovative optical flow optimization algorithm is proposed based on a 3-D Pulse-Coupled Neural Network (PCNN) model. Because of the excellent information clustering ability of PCNN, the proposed algorithm can significantly improve the quality of optical flow. Moreover, a sparse motion flow field is generated to boost the computation efficiency. We employ a preliminary processing to detect the Region of Interest (ROI) in the image, and optical flow is only calculated and optimized in the ROI to save computation resources. Finally, the improved sparse optical flow field is exploited for a systematic on-road vehicle motion analysis. The proposed methodology has been evaluated under various challenging traffic situations to demonstrate its excellent performance.
... Furthermore, there was an implicit assumption that shadows are cast on the same kind of surface, which may not hold true for a variety of outdoor scenes (Fung et al., 2002). Gu et al. (2005) implemented a biological approach to shadow removal. Noting synchronous pulse bursts in the visual cortex of cats, they implemented a Pulse Coupled ...
... Esta característica é importante em processamento de imagens, pois, neurônios representando pontos do fundo da cena podem ser ignorados por receberam entrada primária nula permanecendo inativos. Figura 2.6: O Neurônio PCN obtido em (Johnson, 1994) Diversos pesquisadores ao longo dos últimos anos têm ampliado o trabalho de Johnson aplicando o modelo original e suas modificações a diversas tarefas: Fusão de Imagens (Broussard et al., 1999), Segmentação de Imagens (Kuntimad & Ranganath, 1999), Visão Foveal (Kinser, 1999), Remoção de Sombras em Imagens (Gu et al., 2005), Casamento de Contorno e Movimento (Yu & Zhang, 2004), dentre outras aplicações. ...
... Pulse coupled neural network (PCNN), a single layer neural network, was derived from the Echorn's neuron model which had not been carried out until 1990's [8] . PCNN has a great difference with the traditional neural networks and has been proven to be highly effective when it is used in a diverse set of applications [9][10][11][12][13][14] . Compared with the classical artificial neural networks model, PCNN can be applied to image segmentation without the pre-practice process. ...
Article
The extraction of roads from high spatial resolution remote sensing images remains a problem though lots of efforts have been made in this area. High spatial resolution remote sensing images represent the surface of the earth in detail. As spatial resolution increases, spectral variability within the road cover units becomes complex and traditional remote sensing image processing methods on pixel basis are no longer suitable. This paper studies automatic road extraction from remote sensing images based on methods of Pulse-Coupled Neural Network and mathematical morphology. PCNN is a useful biologically inspired algorithm, and has the properties of linking field and dynamic threshold which make similar neurons generate pulses simultaneously. PCNN has the ability of a neuron to capture neighboring neurons which are in similar states and the independency of the pulses within unattached neuron regions. The method of mathematical morphology has the prime principle which is using a certain structure element to measure and extract the corresponding form in an image. In this paper, the simplified PCNN is applied as the image segmentation algorithm, and morphological transformation is used to purify the roads' information and to extract the road centerlines. Experimental results show that this method is efficient in road extraction from remote sensing images.
Article
This study proposes a ladder gradient method for neutron and gamma-ray discrimination. The proposed method exhibited state-of-the-art performance with low time consumption, which incorporates two parts: information extraction and discrimination factor calculation. A quasi-continuous spiking cortical model was proposed to extract information from the radiation pulse signals, thus generating an ignition map corresponding to each pulse signal. The ignition map can be used to calculate the discrimination factor. A ladder gradient calculation was introduced to obtain a discrimination factor with low computational complexity. The proposed method was compared with five other discrimination methods to evaluate its robustness and efficacy. Furthermore, the filter adaptability of the pulse-coupled neural network and ladder gradient methods was investigated. Possible reasons for adapting the conditions with different discrimination methods and filters were analyzed. Experiments were conducted in 20 filtering situations with 11 types of filters to determine the most suitable filters for discrimination methods. The experimental results revealed that the three most adaptive filters of the pulse-coupled neural networks and ladder gradient methods are the wavelet, elliptic, and median filters and the elliptic, moving average, and wavelet filters, respectively.
Article
Image denoising, a significant research area in the field of medical image processing, makes an effort to recover the original image from its noise corrupted image. The Pulse Coupled Neural Networks (PCNN) works well against denoising a noisy image. Generally, image denoising techniques are directly applied on the pixels. From the literature review, it is reported that denoising after frequency domain transformation is performing better since noise removal is applied over the coefficients. Motivated by this, in this paper, a new technique called the Static Thresholded Pulse Coupled Neural Network (ST-PCNN) is proposed by combining PCNN with traditional filtering or threshold shrinkage technique in Contourlet Transform domain. Four different existing PCNN architectures, such as Neuromime Structure, Intersecting Cortical Model, Unit-Linking Model and Multichannel Model are considered for comparative analysis. The filters such as Wiener, Median, Average, Gaussian and threshold shrinkage techniques such as Sure Shrink, HeurShrink, Neigh Shrink, BayesShrink are used. For noise removal, a mixture of Speckle and Gaussian noise is considered for a CT skull image. A mixture of Rician and Gaussian noise is considered for MRI brain image. A mixture of Speckle and Salt and Pepper noise is considered for a Mammogram image. The Performance Metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Image Quality Index (IQI), Universal Image Quality Index (UQI), Image Enhancement Filter (IEF), Structural Content (SC), Correlation Coefficient (CC), and Weighted Signal-to-Noise Ratio (WSNR) and Visual Signal-to-Noise Ratio (VSNR) are used to evaluate the performance of denoising.
Article
Inspired by multiple information processing mechanisms of the human nervous system, a fusion simplified pulse coupled neural network (FSPCNN) model for infrared (IR) image segmentation is proposed in this paper. In the method based on FSPCNN, the time decay factor is set adaptively based on Stevens’ power law, and the synaptic weight is generated adaptively based on Lateral Inhibition (LI), without manual intervention. Meanwhile, according to Fast linking mechanism, the similarity between adjacent iteration results is used to implement the automatic selection of optimal segmentation result and control iteration. Experimental results indicate that the proposed method can satisfactorily segment targets from complex backgrounds, and it has favorable robustness and segmentation performance.
Article
Hyperspectral Image (HSI) visualization, which aims at displaying as much material information of original images as possible on a trichromatic monitor with natural color, plays an important role in image interpretation and analysis. However, most of the HSI visualization methods only focus on presenting the detail information of a scene without providing natural colors and distinguishing land covers with similar colors. In order to address this problem, this article proposes a multichannel pulse-coupled neural network (MPCNN)-based HSI visualization method, which consists of the following steps. First, the MPCNN is proposed and explored to fuse the original HSI so as to obtain a fused band with rich spatial details. Then, a color mapping scheme is proposed to determine the weights of red, green, and blue (RGB) channels. Finally, the weighted RGB channels are stacked together for visualization. Experiments performed on four hyperspectral data sets demonstrate that the proposed method not only displays the HSI with nature colors but also improves the details in the image. The effectiveness of the proposed method is demonstrated in terms of both visual effect and objective indexes.
Conference Paper
Video contains a large number of motion information. The video– particularly video with moving camera – is segmented based on the relative motion occurring between moving targets and background. By using fusion ability of pulse coupled neural network (PCNN), the target regions and the background regions are fused respectively. Firstly using PCNN fuses the direction of the optical flow fusing, and extracts moving targets from video especially with moving camera. Meanwhile, using phase spectrums of topological property and color pairs (red/green, blue/yellow) generates attention information. Secondly, our video attention map is obtained by means of linear fusing the above features (direction fusion, phase spectrums and magnitude of velocity), which adds weight for each information channel. Experimental results shows that proposed method has better target tracking ability compared with three other methods– Frequency-tuned salient region detection (FT) [5], visual background extractor (Vibe) [6] and phase spectrum of quaternion Fourier transform (PQFT) [1].
Article
The aim of infrared and visible image fusion is to enhance the feature in infrared image and preserve abundant detail information in visible image. Based on the fact that the human sense system accepts external stimulation only when the stimulus intensity is greater than a certain value and the reaction of neuronal cells have obvious regional characters, an image fusion algorithm based on region dual-channel unit-linking pulse coupled neural networks (RDU-PCNN) and independent component analysis (ICA) bases in non-subsampled shearlet transform (NSST) domain for infrared and visible images is proposed. RDU-PCNN we constructed has obvious regional characters and much lower computational costs. We trained ICA-bases using a number of images that the content and statistical properties are similar with the fusion images but applied it as low-frequency ICA-bases, which can reduce calculation complexity. Experimental results demonstrate that the proposed method can significantly improved the fusion quality and need less computational costs.
Conference Paper
This paper presents an efficient face recognition method invariant to slight variation in pose, illumination and background where pulse coupled neural network (PCNN) is used to compute time signature used for feature extraction and generalized self organizing map is being used for classification. The efficiency has been significant improvement by combining PCNN which better resemble human brain neurons and possess better computation power in terms of performance than other neural networks. It is coupled with generalized self organizing map for recognition. The proposed method is robust as it provided better result for slight variation of background and pose and recognition time is greatly reduced.
Article
Using the image moment as the characteristics to recognize biological cells has gradually become the research hotspot of academia. The extraction of image features is to find a set of data which is as small as possible and which can represent the image. However, all the current researches cannot well balance its accuracy and robustness, especially cannot guarantee the multi-distortional invariants such as scaling, translation and rotation when choosing the features of the images. Therefore, this paper proposes an algorithm that possesses the multi-distortional invariants features for the extraction of squamous carcinoma cells, which is based on Mapping Exponential-Fourier Moments (MEFMs). At the same time, extending it to the general cell feature extraction mode and finally using the image moment to reconstruct the image. From the result of the experiment, we can see that the accuracy of the feature extraction of the target image is significantly improved, which thus ensures a high recognition efficiency.
Conference Paper
Visual tracking is a challenging problem in computer vision. Many visual trackers either rely on luminance information or other simple color representations for image description. This paper introduces a tracking algorithm using unit-linking PCNN (Pulse Coupled Neural Network) image icon and particle filter. This approach has the translation, rotation, and scale invariance for using unit-linking PCNN image icon as the features. The experimental results show the proposed approach is with 16.43 % higher median distance precision than the color gradient-based tracker. This unit-linking PCNN image icon-based particle filter tracker can better solve the problems caused by partial occlusions, or out-of-plane rotation, or scale variation, or non-rigid object deformation, or fast motion.
Chapter
The image captured by eyes is transmitted to brain by the optic nerve, and the image signal is transferred in the fiber pathways and finally processed by the primate visual cortex dominantly. The primate visual cortex is devoted to visual processing, and nearly all visual signals reach the cortex via the primary visual cortex. The primary visual cortex is the largest and most important visual cortical area, and does so when neurons in the cortex fire action potentials as stimuli appear within their receptive fields. Signal produced in neurons is transferred to their neighbors by means of localized contact of synapses, which are located on the dendrites and also on the neuron cell body. Electrical charges produced at the synapses propagate to the soma and produce a net postsynaptic potential. If the postsynaptic potential is large enough to exceed a threshold value, the neuron generates an action potential. Synchronized Gama oscillations (30 – 100 Hz) were found in the primary visual cortex of mammalian [1], [2]. In Ref. [2], the linking field model was proposed based on the hypothesis that neuronal pulse synchronizations can be partitioned into two types: stimulus-forced and stimulus-induced synchronizations. Stimulus-forced synchronizations are directly driven by stimulus transients and establish fast but crude sketches of association in the visual cortex, while stimulus-induced synchronizations are believed to be produced via process among local neural oscillations that are mutually connected. The feeding and the linking create the membrane potential.
Article
This paper proposed an improved Pulse-Coupled Neural Network (PCNN)-based object tracking algorithm aimed to the case that the object being tracked hardly in complex and dynamic scene. The principle of PCNN and its simplify procedure was discussed firstly and the relationship between PCNN and image was described. Then, the Mean-Shift theory was used to construct the complete object tracking process. In this process, the PCNN was employed as the classification tool to classify the specific data into object and scene. The moment theory was used to calculate the ellipse corresponding to the object so that to adapt the scale changing of the object. In the experiment, the real image sequences of the videos were employed, and the tracking results using proposed method showed that the method of this paper was efficient.
Article
Shadow is an integral part of many natural images, which can pose tough problems and limitations for further image processing tasks. By the analysis of shadow characteristics, a single image shadow detection method based on the intersecting cortical model (ICM) is proposed. Neurons in ICM possess dynamical spiking properties have the capability to segment the image naturally. We modify the linking matrix among neurons and combine the local texture features shown by local binary patterns (LBP) to make the Te-ICM for segment of shadow regions. The new model possesses the capability of taking adjacent pixel information into the firing matrix. The optimized parameters produced by the modified hierarchical clustering histogram partition method lead to the shadow detection sequences. We build an automatic stopping condition for umbra and penumbra iterations. Experimental results demonstrate that the output shadow mask keeps the size and shape of original objects well for typical image dataset, and that the proposed method can find wide applications to monochromatic or chromatic images containing one or more shadow regions, yielding high-quality results for further shadow removal operation.
Article
Though the mechanism called lateral inhibition exists among the visual cells, visual information can be selectively detected. By simulating this mechanism among visual cells through a mathematical method, a lateral inhibition neural network is proposed. A method of image segmentation based on this model with partial reset is presented. The simulation results show that partial reset may allow a fine control of the irregularity of the spike trains. In image segmentation, experimental results show that the lateral inhibition neural network with partial reset is more effective than the neural network with constant threshold by quantificationally evaluating their performance.
Article
In order to solve the problem that shadows pose difficult problems and limitations for various computer vision algorithms, a single image shadow removal method was proposed based on the analysis of shadow characteristics and modeling. An energy minimization framework was constructed to remove the umbra based on color constancy computation by regions. According to color proportion method, restitution coefficient was calculated by texture matching. Furthermore, a composite model composed of local color constancy and smoothing changing luminance profile gave the scale factor for penumbra region. Experimental results demonstrate that the output shadow mask keeps the size and shape of original objects well, and yielding high-quality results that those areas inside the shadow are compatible with lit areas by luminance, color and texture.
Conference Paper
This paper proposes a biologically inspired video attention saliency detector by combining optical flow and topological properties. In this paper, we expound how to utilize feature informations of video and how to use topological property and optical flow based on attention detection for target tracking. Visual attention used in our model is a consequence of tuning of some saliency features such as color, shape and motion. The model (OFTPA-Optical Flow Topological Properties Attention) we proposed for motion attention saliency detection includes two stages. First stage focuses on extracting saliency features, including optical-flow velocity field, topological properties and Intensity, from video. The second integrates the traditional saliency features and an extra position prediction calculated from optical-flow field, to form bottom-up saliency maps which indicate where the object candidates are located. Spatiotemporal saliency maps are obtained from the phase spectrum of a video's hypercomplex Fourier transform. Experimental results show that the OFTPA model takes advantage over other models such as PQFT in complex background.
Article
On the research of shadow essence and visual scheme, we propose a single image shadow removal method based on certain layered perceptual representation models with the flowchart from shadow detection to shadow removal. Firstly, a modified intersecting cortical model, the typically useful image factorization technique, is applied to extract umbra and penumbra masks. Then, under the energy minimization framework, scale factors for umbra are computed. Furthermore, transparency-coupled atmospheric transfer function is introduced for penumbra compensation by pixel-by-pixel transparency estimation. For aerial images, experimental results illustrate that shadow regions are matted well, and the proposed method yields vivid shadow-free images with smooth boundaries.
Article
Shadows are physical phenomena observed in most natural scenes. Shadows in an image can cause segmentation, tracking, or recognition algorithms to fail. In this paper, we give a research on the assumptions introduced in most researches about the complexity of shadow removal from a single image and the classification of shadow images. The purpose of this paper is twofold: in the first place, it provides a comprehensive survey of the problems and challenges which may occur when removing shadows from a single image. In the second place, an idea of shadow removal is presented, in which an attempt is made to overcome some of the fundamental problems such as illumination, color constancy, and type of reflection. Experimental results show that the proposed method can detect and eliminate the shadows effectively.
Article
A candidate set of support vectors is selected by using pulse-coupled neural networks to reduce computational cost in learning phase for support vector machines (SVMs). The size of the candidate set of support vectors selected this way is smaller than that of the original training samples so that the computation complexity in learning process for support vectors machines based on this candidate set is reduced and the learning process is accelerated. On the other hand, the candidate set of support vectors includes almost all support vectors, and the performance of the SVM based on this candidate set matches the performance when the full training samples are used.
Article
This article, by presenting a brief summary on the development of Pulse Coupled Neural Network (PCNN) and its application, describes major research area and application field of PCNN in China and abroad. The structure of this article is: first is a brief and systematic introduction to the origin, model and characteristics of PCNN, which is followed by a discussion and analysis concerning PCNN theoretical development and application studies, especially summarizing the research achievement of PCNN in the field of speech processing for the first time in recent years. The final section is a prospect of PCNN development in the future.
Article
Long-term autonomy in robotics requires perception systems that are resilient to unusual but realistic conditions that will eventually occur during extended missions. For example, unmanned ground vehicles (UGVs) need to be capable of operating safely in adverse and low-visibility conditions, such as at night or in the presence of smoke. The key to a resilient UGV perception system lies in the use of multiple sensor modalities, e.g., operating at different frequencies of the electromagnetic spectrum, to compensate for the limitations of a single sensor type. In this paper, visual and infrared imaging are combined in a Visual-SLAM algorithm to achieve localization. We propose to evaluate the quality of data provided by each sensor modality prior to data combination. This evaluation is used to discard low-quality data, i.e., data most likely to induce large localization errors. In this way, perceptual failures are anticipated and mitigated. An extensive experimental evaluation is conducted on data sets collected with a UGV in a range of environments and adverse conditions, including the presence of smoke (obstructing the visual camera), fire, extreme heat (saturating the infrared camera), low-light conditions (dusk), and at night with sudden variations of artificial light. A total of 240 trajectory estimates are obtained using five different variations of data sources and data combination strategies in the localization method. In particular, the proposed approach for selective data combination is compared to methods using a single sensor type or combining both modalities without preselection. We show that the proposed framework allows for camera-based localization resilient to a large range of low-visibility conditions.
Conference Paper
Full-text available
The article presents an original approach to optical character recognition (OCR) used in real environments, such as gas- and electricity-meters, where the quantity of noise is sometimes as large as the quantity of good signal. This approach uses two algorithms for better results. These are a neural network on one hand, respectively the k-nearest neighbor as the confirmation algorithm. Unlike other OCR systems, this one is based on the angles of the digits, rather than on pixels. This makes it insensitive to the possible rotations of the digits, respectively to the quantity of noise that may appear in an image. We will prove that the approach has several advantages, such as: insensitivity to the possible rotations of the digits, the possibility to work in different light and exposure conditions, the ability to deduct and use heuristics for character recognition.
Article
Topological properties are with invariance and take priority over other features, which play an important role in cognition. This paper introduces a new attention selection model called TPA (topological properties-based attention), which adopts topological properties and quaternion. In TPA, using Unit-linking PCNN (Pulse Coupled Neural Network) hole-filter expresses an important topological property (the connectivity) in visual attention selection. Meanwhile, using the quaternion Fourier transform based phase spectrum of an image or a frame in a video obtains the spatio-temporal saliency map, which shows the result of attention selection. Adjusting the weight of a topological channel can change its influence. The experimental results show that TPA reflects the real attention selection more accurately than PQFT (Phase spectrum of Quaternion Fourier Transform).
Conference Paper
In this paper, we consider the problem to recover a blurred and noisy binary images when the point spread function (PSF) is known. By explicitly using the a priori knowledge for binary image in the constrained minimization model, we present a fast algorithm for binary image restoration. The main computation at each iteration is one Fourier transform (FFT) and one inverse fast Fourier transform (IFFT). Experimental results show that the proposed method is feasible and effective for binary image restoration.
Conference Paper
X-ray imaging is extensively used in the NDT. In the conventional method, interpretation of the large number of radiographs for defect detection and evaluation is carried out manually by operator or expert, which makes the system subjective. Also interpretation of large number of images is tedious and may lead to misinterpretation. Automation of Non-Destructive evaluation techniques is gaining greater relevance but automatic analysis of X-Ray images is still a complex problem, as the images are noisy, low contrast with a number of artifacts. ANN's are systems which can be trained to analyze input data based on conditions provided to derive required output. This makes the system automatic reducing the subjective interference in analysis of data. Artificial neural network based systems are thus a feasible solution to this problem of X-Ray NDT. Due to complex nature of input images and noise present, Noise removal becomes a problem in X-Ray images. Preprocessing techniques based on statistical analysis have shown improvement in image noise reduction. Pixels/group of pixels, which deviate from the general structural pattern and grey scale distribution are located. The statistically processed pixel values are used to obtain the features vector from defective as well as from non-defective areas. Software for pre-processing and analyzing NDT images has been developed. Software allows user to train neural networks for defect detection. Once trained satisfactorily, the software scans the new input image and uses the trained ANN for defect detection. The final image with defect regions marked will be displayed. This system can be used to obtain the probable defective areas in a given input image. This paper presents performance of MLP and RBF for detection of defect. The effect of different types of input viz. template and moments on performance of ANN is discussed.
Article
Full-text available
In this paper, we propose a novel process to optical character recognition (OCR) used in real environments, such as gas-meters and electricity-meters, where the quantity of noise is sometimes as large as the quantity of good signal. Our method combines two algorithms an artificial neural network on one hand, and the k-nearest neighbor as the confirmation algorithm. Our approach, unlike other OCR systems, it is based on the angles of the digits rather than on pixels. Some of the advantages of the proposed system are: insensitivity to the possible rotations of the digits, the possibility to work in different light and exposure conditions, the ability to deduct and use heuristics for character recognition. The experimental results point out that our method with moderate level of training epochs can produce a high accuracy of 99.3 % in recognizing the digits, proving that our system is very successful.
Article
In this paper, we present a method for reducing the intensity of shadows cast on the ground in outdoor sports videos to provide TV viewers with a better viewing experience. In the case of soccer videos taken by a long-shot camera technique, it is difficult for viewers to discriminate the tiny objects (i.e., soccer ball and player) from the ground shadows. The algorithm proposed in this paper comprises three modules, such as long-shot detection, shadow region extraction and shadow intensity reduction. We detect the shadow region on the ground by using the relationship between Y and U values in YUV color space and then reduce the shadow components depending on the strength of the shadows. Experimental results show that the proposed scheme offers useful tools to provide a more comfortable viewing environment and is amenable to real-time performance even in a software based implementation.
Article
Topological properties have important function in human beings visual attention. TPQFT (Topological properties based Phase spectrum of Quaternion Fourier Transform) model is an attention selection model using topological properties expression we have introduced. A new quantification criterion to evaluate every channel's contribution and model's performance is proposed in this paper and used in TPQFT. This paper improves TPQFT model in several aspects and WTPQFT (Weight Adjusted Topological properties based Phase spectrum of Quaternion Fourier Transform) model is introduced. The experimental results show that WTPQFT model reflects the real attention selection more accurately than TPQFT and PQFT (Phase spectrum of Quaternion Fourier Transform) method. existed in TPQFT model. In order to solve these problems, WTPQFT model is introduced. Experiments and criteria show this model has better performance in attention distribution than TPQFT and PQFT method. The innovation in this paper has three aspects: Firstly, we introduce topological properties into attention and construct TPQFT model. Secondly, we propose a new quantification criterion to evaluate every channel's influence and the performance of attention models. Lastly, we construct WTPQFT model improved from TPQFT model. The paper involves five sections. In section 2, we introduce TPQFT model, which is an attention model using topological theory (8), (9). In section 3, a new quantification criterion is proposed to evaluate every channel's contribution and the performance of the model objectively without needing ground truth. In section 4, we present a new model---WTPQFT model, which is improved from TPQFT method. The last section is conclusion.
Article
A novel method is proposed to detect and Alter pepper and salt noise in an image based on pulse coupled neural network (PCNN) firing matrix, which contains the spatial information and time information of the original image. The PCNN structure is simplified to avoid selecting too many parameters, which a unidirectional decaying threshold is proposed. The pixels polluted by noise are detected through analyzing and processing the PCNN firing matrix then the noisy pixels are filtered by median filter, which is proved to be effective in removing pepper and salt noise. The window size of the filter and the filtering time for noisy pixels are adaptively determined by calculating the noise intensity of the contaminated image. The experiment result demonstrates that the proposed method illustrates better performance in removing noise while conserving image edges and details than traditional filter does.
Article
Full-text available
We recently discovered stimulus-specific interactions between cell assemblies in cat primary visual cortex that could constitute a global linking principle for feature associations in sensory and motor systems: stimulus-induced oscillatory activities (35-80 Hz) in remote cell assemblies of the same and of different visual cortex areas mutually synchronize, if common stimulus features drive the assemblies simultaneously. Based on our neurophysiological findings we simulated feature linking via synchronizations in networks of model neurons. The networks consisted of two one-dimensional layers of neurons, coupled in a forward direction via feeding connections and in lateral and backward directions via modulatory linking connections. The models' performance is demonstrated in examples of region linking with spatiotemporally varying inputs, where the rhythmic activities in response to an input, that initially are uncorrelated, become phase locked. We propose that synchronization is a general principle for the coding of associations in and among sensory systems and that at least two distinct types of synchronization do exist: stimulus-forced (event-locked) synchronizations support “crude instantaneous” associations and stimulus-induced (oscillatory) synchronizations support more complex iterative association processes. In order to bring neural linking mechanisms into correspondence with perceptual feature linking, we introduce the concept of the linking field (association field) of a local assembly of visual neurons. The linking field extends the concept of the invariant receptive field (RF) of single neurons to the flexible association of RFs in neural assemblies.
Article
Full-text available
A pulse-coupled neural network was implemented, for the first time to our knowledge, in a hybrid electro-optical laboratory demonstration system. Dynamic coherent traveling-wave patterns were observed that repeated their spatial patterns at each locality with a period that depended on the local input pattern and strength. Coherence and periodicity were maintained far beyond the physical limits of the linking receptive fields, suggesting a new mechanism for information transmission in a network with limited local connectivity. With no linking, the output became chaotic because the relative phases increased linearly in time.
Article
Full-text available
A fundamental step in visual pattern recognition is the establishment of relations between spatially separate features. Recently, we have shown that neurons in the cat visual cortex have oscillatory responses in the range 40-60 Hz (refs 1, 2) which occur in synchrony for cells in a functional column and are tightly correlated with a local oscillatory field potential. This led us to hypothesize that the synchronization of oscillatory responses of spatially distributed, feature selective cells might be a way to establish relations between features in different parts of the visual field. In support of this hypothesis, we demonstrate here that neurons in spatially separate columns can synchronize their oscillatory responses. The synchronization has, on average, no phase difference, depends on the spatial separation and the orientation preference of the cells and is influenced by global stimulus properties.
Article
Full-text available
Primary visual coding can be characterized by the receptive field (RF) properties of single neurons. Subject of this paper is our search for a global, second coding step beyond the RF-concept that links related features in a visual scene. In recent models of visual coding, oscillatory activities have been proposed to constitute such linking signals. We tested the neurophysiological relevance of this hypothesis for the visual system. Single and multiple spikes as well as local field potentials were recorded simultaneously from several locations in the primary visual cortex (A17 and A18) using 7 or 19 individually advanceable fiber-microelectrodes (250 or 330 microns apart). Stimulus-evoked (SE)-resonances of 35-85 Hz were found in these three types of signals throughout the visual cortex when the primary coding channels were activated by their specific stimuli. Stimulus position, orientation, movement direction and velocity, ocularity and stationary flicker caused specific SE-resonances. Coherent SE-resonances were found at distant cortical positions when at least one of the primary coding properties was similar. Coherence was found 1) within a vertical cortex column, 2) between neighbouring hypercolumns, and 3) between two different cortical areas. We assume that the coherence of SE-resonances is mediated by recurrent excitatory intra- and inter-areal connections via phase locking between assemblies that represent the linking features of the actual visual scene. Visually related activities are, thus, transiently labelled by a temporal code that signalizes their momentary association.
Article
Full-text available
Humans do not stare at an image, they foveate. Their eyes move about points of interest within the image collecting clues as to the content of the image. Object shape is one of the driving forces of foveation. These foveation points are generally corners and, to a lesser extent, the edges. The pulse-coupled neural network (PCNN) has the inherent ability to segment an image. The corners and edges of the PCNN segments are similar to the foveation points. Thus, it is a natural extension of PCNN technology to use it as a foveation engine. This paper will present theory and examples of foveation through the use of a PCNN, and it will also demonstrate that it can be quite useful in image recognition.
Article
Full-text available
This paper presents the first physiologically motivated pulse coupled neural network (PCNN)-based image fusion network for object detection. Primate vision processing principles, such as expectation driven filtering, state dependent modulation, temporal synchronization, and multiple processing paths are applied to create a physiologically motivated image fusion network. PCNN's are used to fuse the results of several object detection techniques to improve object detection accuracy. Image processing techniques (wavelets, morphological, etc.) are used to extract target features and PCNN's are used to focus attention by segmenting and fusing the information. The object detection property of the resulting image fusion network is demonstrated on mammograms and Forward Looking Infrared Radar (FLIR) images. The network removed 94% of the false detections without removing any true detections in the FLIR images and removed 46% of the false detections while removing only 7% of the true detections in the mammograms. The model exceeded the accuracy obtained by any individual filtering methods or by logical ANDing the individual object detection technique results.
Article
In this paper, researches are deep done on using PCNN-pulse coupled neural network, a new artificial neural network based on biology, to restore binary images and smooth images. Meanwhile, the image restoration algorithm based on PCNN is brought forward. The results of computer simulations show that noisy binary images can be restored efficiently by using PCNN and the SNRs of restored images by using PCNN is higher than by using one of two usual image restoration methods, the median filter and the mean filter.
Article
A pulse-coupled neuron (PCN) circuit is proposed. This circuit uses a positive feedback circuit with two capacitors. One corresponds to the sodium ion potential and the other corresponds to the potassium ion potential and can be easily implemented in CMOS technology. The circuit behaves like a real neuron, generating a pulse train in which the frequency increases with an increase of input excitation. The circuit threshold increases after pulse generation and then gradually returns to the initial level. These electronic neuron models are used for image processing. Images with sharp corners and those with narrow pathways as features were generated and corrupted with noise. Performance of mean and median filters was compared to that of the new PCN networks with several strengths of coupling. These tests of the capabilities of the new circuit demonstrate successful restoration of interesting images and their features.
Conference Paper
PCNN-Pulse Coupled Neural Network, a new artificial neural network based on biological experimental results, can be widely used for image processing. The complexities of the PCNN’s structure and its dynamical behaviors limit its application so simplification of PCNN is necessary. We have used simplified PCNNs to efficiently process image. In this paper dynamical behaviors of simplified PCNNs under certain conditions are analyzed in detail and we obtain the conclusion that under these conditions, simplified PCNNs have periodic solutions, i.e. their dynamical behaviors are periodical.
Article
PCNN-pulse coupled neural network, based on the phenomena of synchronous pulse bursts in the animal visual cortex, is different from traditional artificial neural networks. This paper first introduces a new approach for binary image thinning by using the pulse parallel transmission characteristic of PCNN. The thinning result obtains when pulses emitted by background meet. The criterion of pulse meeting and the criterion of thinning completion are proposed. The computer simulation results of applying the method to thin binary image are present. Comparisons of skeleton structure and execution time with results from other thinning methods are present too. The PCNN skeleton retains more information of original binary image, such as the size of a quadrate, than the result from Zhang and Suen method. The procedure is faster than Arcelli et al. thinning method when the image resolution is from 600 to1800 dpi. Combining with PCNN restoration algorithm (namely PCNN noise-reducing algorithm), the skeletons of the objects in a noisy binary image can be obtained with the accuracy. This paper also expands the application field of PCNN.
Article
It has been proposed that synchronized oscillations play a key role in perceptual feature linking and sensory integration. This idea was supported by the discovery of strongly synchronized stimulus-specific oscillations in the visual cortex of anaesthetized cats. The 'synchronization hypothesis' was controversial because in the visual cortex of awake monkeys either only weak or no oscillations were found. We have now recorded high amplitude synchronized oscillation at the level of spike activity and local field potential from the primary visual cortex of an awake monkey. The dominant frequencies (70-80 Hz) were considerably higher than those observed previously in cats and monkeys (30-50 Hz). However, stimulus specificities of the oscillations were comparable to and amplitudes even higher than those in cats.
Article
This paper describes an object detection system based on pulse coupled neural networks. The system is designed and implemented to illustrate the power, flexibility and potential the pulse coupled neural networks have in real-time image processing. In the preprocessing stage, a pulse coupled neural network suppresses noise by smoothing the input image. In the segmentation stage, a second pulse coupled neural-network iteratively segments the input image. During each iteration, with the help of a control module, the segmentation network deletes regions that do not satisfy the retention criteria from further processing and produces an improved segmentation of the retained image. In the final stage each group of connected regions that satisfies the detection criteria is identified as an instance of the object of interest.
Article
A pulse coupled neural network (PCNN) can run mazes nondeterministically (taking all possible paths) with constant time per step. Thus, when a signal emerges, it has taken the shortest path in the shortest time.
Article
This paper describes a method for segmenting digital images using pulse coupled neural networks (PCNN's). The pulse coupled neuron (PCN) model used in PCNN is a modification of Eckhorn's cortical neuron model. A single layered laterally connected PCNN is capable of perfectly segmenting digital images even when there is a considerable overlap in the intensity ranges of adjacent regions. Conditions for perfect image segmentation are derived. It is also shown that addition of an inhibition receptive field to the neuron model increases the possibility of perfect segmentation. The inhibition input reduces the overlap of intensity ranges of adjacent regions by effectively compressing the intensity range of each region.
Article
This paper presents a compact architecture for analog CMOS hardware implementation of voltage-mode pulse-coupled neural networks (PCNN's). The hardware implementation methods shows inherent fault tolerance specialties and high speed, which is usually more than an order of magnitude over the software counterpart. A computational style described in this article mimics a biological neural network using pulse-stream signaling and analog summation and multiplication. Pulse-stream encoding technique uses pulse streams to carry information and control analog circuitry, while storing further analog information on the time axis. The main feature of the proposed neuron circuit is that the structure is compact, yet exhibiting all the basic properties of natural biological neurons. Functional and structural forms of neural and synaptic functions are presented along with simulation results. Finally, the proposed design is applied to image processing to demonstrate successful restoration of images and their features.
Article
The pulse coupled neural network (PCNN) models are described. The linking field modulation term is shown to be a universal feature of any biologically grounded dendritic model. Applications and implementations of PCNN's are reviewed. Application based variations and simplifications are summarized. The PCNN image decomposition (factoring) model is described in new detail.
Conference Paper
The PCNN (pulse coupled neural network), an artificial neural network based on biology, can be efficiently applied to image segmentation. The performance of image segmentation based on PCNN depends on suitable PCNN parameters. However, it is difficult to get suitable PCNN parameters for different kinds of images because different kinds of images have different suitable PCNN parameters. So far, no paper has described how to get the suitable PCNN parameters to efficiently segment images. In this paper, we put forward a new approach for image segmentation based on a unit-linking PCNN, by which we can use the same PCNN parameter to efficiently segment different kinds of images. Therefore, using this new approach can automatically and efficiently segment images without choosing different parameters for different kinds of images.
Conference Paper
This paper focuses on the novel approaches to chemosensor signal analysis: (1) forming image patterns from the time sequences, (2) PCNN factor formation, and (3) factor classification using wavelets
binary image restoration using pulse coupled neural network
  • gu