Updates
0 new
1
Recommendations
0 new
0
Followers
0 new
4
Reads
3 new
164

Project log

Dalton meitei Thounaojam
added 2 research items
This paper proposes a novel shot boundary detection method which combines the Advantages of Particle Swarm Optimization (PSO) and Gravitational Search Algorithm (GSA) to optimize the weights of the Feed-Forward Neural Network (FNN). To increase the performance of the system, the output of the hybrid technique is again analyzed by forming a Continuity matrix (ϕ). Then an Outlier along with a Continuity matrix is used for extracting a possible set of transition frames. A set of thresholds δ1 and δ2 is selected for classifying abrupt and gradual transitions from the available set of possible transition frames. Experimental results using TRECVid 2001 depicts that PSOGSA outperforms GSA and PSO in-terms of training the feed forward neural network and generating a higher overall F1 score. The proposed system also gives better performance when compared with other latest techniques in-terms of F1 score.
There has been a lot of research on supervised learning in spiking neural network (SNN) for a couple of decades to improve computational efficiency. However, evolutionary algorithm based supervised learning for SNN has not been investigated thoroughly which is still in embryo stage. This paper introduce an efficient algorithm (SpiFoG) to train multilayer feed forward SNN in supervised manner that uses elitist floating point genetic algorithm with hybrid crossover. The evidence from neuroscience claims that the brain uses spike times with random synaptic delays for information processing. Therefore, leaky-integrate-and-fire spiking neuron is used in this research introducing random synaptic delays. The SpiFoG allows both excitatory and inhibitory neurons by allowing a mixture of positive and negative synaptic weights. In addition, random synaptic delays are also trained with synaptic weights in an efficient manner. Moreover, computational efficiency of SpiFoG was increased by reducing the total simulation time and increasing the time step since increasing time step within the total simulation time takes less iteration. The SpiFoG is benchmarked on Iris and WBC dataset drawn from the UCI machine learning repository and found better performance than state-of-the-art techniques.
Dalton meitei Thounaojam
added 2 research items
Due to the development of image handling tool or software, copy– move is increasingly becoming a common and on the other hand, the detection of such type Biswas, Debalina attack from digital images has become the challenging and active research area. This paper presents the recent block and keypoints-based Copy–Move Forgery Detection (CMFD) techniques. In this paper, we cover the critical discussions of different blocks and keypoints-based CMFD techniques with their pros and cons. The paper also describes the different publicly available databases and performance evaluation measures. Some unsolved research issues in the field of copy–move forgery detection is identified and present in this paper.
Many researches have been done on shot boundary detection, but the performance of shot boundary detection approaches is yet to be addressed for the videos having sudden illumination and object/camera motion effects efficiently. In this paper, a novel dual-stage approach for an abrupt transition detection is proposed which is able to withstand under certain illumination and motion effects. Firstly, an adaptive Wiener filter is applied to the lightness component of the frame to retain some important information on both frequencies and LBP-HF is extracted to reduce the illumination effect. From the experimentation, it is also confirmed that the motion effect is also reduced in the first stage. Secondly, Canny edge difference is used to further remove the illumination and motion effects which are not handled in the first stage. TRECVid 2001 and TRECVid 2007 datasets are applied to analyze and validate our proposed algorithm. Experimental results manifest that the proposed system outperforms the state-of-the-art shot boundary detection techniques.
Badal Soni
added 2 research items
Underwater scenarios cause the quality of images to be degraded by absorption, reflection, bending and scattering of light. This causes dominance of blue and green color in underwater images. In order to improve the visual quality of underwater images, we proposed a fusion based technique by which combines the Contrast Limited Adaptive Histogram Equalization(CLAHE) and Guided filter approaches. Initially, the Contrast Limited Adaptive Histogram algorithm is applied on R, G and B components of the input image to equalize the colour contrast in images. Secondly, the Guided filter approach is applied on the result of first step to improve the colour contrast and solve the issue of lighting. Experiments indicate that the proposed technique shows promising results in terms of visual quality as well as for quantitative measures.
Presence of haze components in the atmosphere, degrade the contrast as well as the true color of the real images. As a result of this, images are difficult to comprehend and to be used in several applications like surveillance, navigation, detection etc. Therefore, it is important to remove hazy effect from the real world images to get clear or haze free image. In the present paper, a technique to remove haze is implemented by combining the modified Adaptive Histogram Equalization and Guided Filter approaches. Experiments are carried out using various hazy images collected from Internet and the result reveals that the this method can dehaze the haze affected images and restore the original contrast of the images as well.
Amit Trivedi
added 2 research items
Fingerprint biometric has emerged as a highly successful biometric recognition system and find its application in the diverse domain. The fingerprint template database needs to be protected from unauthorized access. If an adversary has acquired the unprotected fingerprint template, then he/she may reconstruct original fingerprint which can be used for unauthorized access of all the systems where the same fingerprint has been used. Hence, securing the fingerprint template is required. In this paper, a non-invertible cancellable fingerprint template based on information extracted from the Delaunay triangulation of minutiae points is proposed. For making the template cancellable a user-specific key of a random binary string is used. The proposed template is revocable, diverse, secure and also gives a good recognition accuracy. The proposed template is tested with a standard FVC2002 database and gives good results as compared with the latest methods.
The shear number of fingerprints in a modern database makes exhaustive search, a computationally an expensive process. It is in this context a new indexing mechanism is proposed to speed up the process of identification of fingerprints. In this model, concentric circles are made around the core of a fingerprint, and by grouping the fingerprints according to the number of minutiae points in each of these rings, we can select the best prospective fingerprints for a particular query fingerprint, thus greatly reducing the number of fingerprints in which exhaustive search is to be employed and increasing the speed of Automated Fingerprint Identification Systems (AFIS). The proposed model was tested on different datasets of FVC2000 database, and the results show that the model achieves high CIP with a low penetration rate and our model was able to significantly speed up the process.
Dalton meitei Thounaojam
added 2 research items
Copy-move forgery is a popular image tampering technique. In this paper, we propose two efficient block-based systems for detection of copy-move forgeries present in images. The first system is based on the extraction of Local Binary Pattern Histogram Fourier features from each overlapping block and forgery decision based on the matching of these block features using Euclidean similarity measure. The second proposed system is based on the extraction of Fast Walsh Hadamard Transform features from each overlapping block and forgery decision based on the matching of these block features using shift vectors analysis. Both systems are tested using tampered images of the CoMoFoD dataset. Experimental results show that both systems are not only able to accurately detect tampered regions but also are invariant to various post-processing operations such as: blur movement, contrast adjustment, brightness and color reduction. Proposed system-I is computationally more efficient than system-II. However, system-II is more robust to blurring post-processing operation. © 2018, International Association of Engineers. All rights reserved.
Perceptual image hashing technique uses the appearance of the digital media object as human eye and generates a fixed size hash value. This hash value works as digital signature for the media object and it is robust against various digital manipulation done on the media object. This technique have been constantly in use in various application areas like content-based image retrieval, image authentication, digital watermarking, image copy detection, tamper detection, image indexing, etc., but it is difficult to generate a perfect perceptual image hash function due to the inverse relationship between its main properties i.e. perceptual robustness and discriminative capability. In this paper, a robust and desirable discrimination capable dual perceptual image hash functions are proposed which use fuzzy color histogram for hash generation. The fuzzy engine needs stable color representation to generate a robust fuzzy color histogram feature which is invariant to various content preserving attacks like gaussian low pass filtering, jpeg compression, etc. To satisfy this, \(CIEL^{*}a^{*}b^{*}\) color space forms an good basis as it approximates the human visual system and it is also uniform and device independent color space. The robustness of the fuzzy color histogram is further increased by selecting the most significant bins using an experimentally selected tuning factor and the same is furthermore normalized to make it scale invariant. Our experimentation shows that hash generated with this feature is more stable and able to handle various content preserving attacks and performs better as compared to the latest techniques. Both the proposed systems able to maintain good balance between perceptual robustness with optimal TPR when the FPR \(\simeq \)0 is 0.8115 and 0.8264 and discrimination capability with the optimal FPR when TPR\(\simeq \)1 is 0.0618 and 0.0208 respectively.
Amit Trivedi
added a research item
Fingerprint Recognition System is widely deployed in variety of application domain, ranging from forensic to mobile phones. Its widespread deployment in various applications has caused concern that a compromised fingerprint template may be used to reconstruct the original fingerprint and the reconstructed image can be used to deceive all the applications a person is enrolled in. In this paper, we proposed a fingerprint template based on relative position of minutiae points. The proposed template is non-invertible and immune to fingerprint reconstruction algorithm. The proposed system is evaluated on the standard fingerprint database (FVC2000) and yields better results.
Dalton meitei Thounaojam
added a research item
Fingerprint Recognition System is widely deployed in variety of application domain, ranging from forensic to mobile phones. Its widespread deployment in various applications were person authentication are required, has caused concern that a leaked fingerprint template may be used to reconstruct the original fingerprint and the reconstructed fingerprint can be used to circumvent all the applications the person is enrolled. In this paper, a non-invertible fingerprint template that stores only the relative geometric information about the minutiae points is proposed. The spatial location of the minutiae points in original fingerprint and its orientations are not available in the proposed template which makes it impossible to estimate the orientation of fingerprint from the template. The proposed template is invariant to rotation, translation and distortion and immune to reconstruction algorithm. The proposed system is experimented using standard FVC2000 database and yields better results in terms of EER and FMR as compared with latest techniques.
Badal Soni
added a research item
Due to excess uses of digital contents for communication and using image handling software and tools manipulation of these contents, detection of copy-move manipulation has become a prominent and interesting research area. The proposed copy-move forgery detection system is based on SIFT key-points extraction and density-based clustering algorithm. Extracted SIFT descriptors are matched using the generalized 2NN procedure. Thereafter, a density-based clustering algorithm is utilized to reduce or remove the false matches for improving detection accuracy. The proposed system is tested using MICC-F220, MICC-F2000 and MICC-F8multi, standard datasets. Due to the generalized 2NN matching procedure, the proposed system is able to detect multiple forgeries present in the image. Experimental results show the detection accuracy of the proposed system is more in comparison to existing systems and it is computationally efficient.
Dalton meitei Thounaojam
added a research item
Copy–move forgery is the most basic technique to alter an image. In this method, one region of an image is copied and pasted into another location of the same image, with an attempt to cover a potentially important feature or duplicate some features. As the copied part resides in the same image, its important properties, such as noise, brightness, texture, are compatible with rest of the image making its detection very difficult. The existing techniques for detecting copy–move forgery suffer from the computational time problem. In this paper, an efficient block-based copy–move forgery detection algorithm is present that reduces the processing time in identifying the duplicated regions in an image. Proposed method is tested on CoMoFoD dataset. Experimental results show the ability of the proposed method to accurately detect the tampered regions as well as reducing the time complexity (15) (PDF) An Efficient Block Phase Correlation Approach for CMFD System. Available from: https://www.researchgate.net/publication/324422402_An_Efficient_Block_Phase_Correlation_Approach_for_CMFD_System [accessed Oct 03 2021].
Dalton meitei Thounaojam
added 4 research items
With the advancement of image editing tools in today's world, the manipulation of images like cropping, cloning, resizing, etc., becomes an easy proposition and on the other end, checking or determining whether an image has been manipulated or not, becomes a great challenge. Copy-move forgery in images is the most popular tampering method in which a portion of an image is copied and pasted in some other location of the same image. The detection of copy-move forgery has become a prominent research area. This study presents a detailed review and critical discussions with pros and cons of each of copy-move forgery detection techniques from 2007 to 2017. This study also addresses the variation in databases, issues, challenges, future directions and references in this domain.
Copy-move forgery detection has become a prominent research area. In this paper, an efficient block based copy-move forgery detection algorithm based on Fast Walsh Hadamard Transform is presented with an objective to reduce processing time in identifying the duplicated regions in an image. Forged regions are detected using lexicographical sorting and efficient shift vector mechanism. Proposed system is tested on different attacked images of the CoMoFoD dataset. Attacks are blur movement, brightness changes, color reduction and contrast adjustment, etc. Performance of proposed system is quite good across all the attacks. Also, it is more robust to blur movement. Experimental results show the ability of the proposed method to accurately detect the tampered regions as well as reducing the time complexity.
Due to the advancement of image manipulation tool or techniques, the copy-move attack detection from digital images has become the challenging and active research area.This paper proposes an improved block-based technique forcopy-move attack detection using Speeded Up Robust Features(SURF) and Features from Accelerated Segment Test (FAST)keypoint matching. In the first phase of this technique, theimage is divided into non-overlapping blocks and SURF de-scriptors are extracted from each block. These descriptorsare matched using 2NN procedure and match blocks areidentified. In the second phase, large blocks are constituted byconcatenating the neighboring blocks of each matching block.Thereafter, from each large block FAST features points areextracted and matched using 2NN. Finally, the affine transformis applied to remove the outliers if any. The proposed techniqueis tested using MICC-F220 and MICC-F2000 standard datasetsand it yields better performance in comparison with state ofthe art techniques.
Dalton meitei Thounaojam
added 4 research items
This paper presents a short survey on video segmentation. Due to the growth in multimedia information, an effective video indexing and video retrieval is necessary. This can be achieved when an effective video segmentation tools and algorithms are available. MPEG-compressed videos are mostly used by researchers for video segmentation. Shot boundary detection, color histogram characteristics, DC-images, motion vector and motion compensation, threshold-based detector, etc. are mostly used for video segmentation.
Objectives: The objective of this paper is to find out the abrupt transitions between consecutive shots in a video with less false detection and high F1 score. Method/Analysis: This paper presents a video shot boundary detection approach using Gray Level Cooccurrence Matrix (GLCM). The proposed system can roughly be divided into feature extraction using GLCM and the application of the abrupt shot boundary detection. In the first step, the frames are converted into gray level and GLCM is calculated from each frame in the video. Secondly, correlation coefficient is calculated from the GLCM of two consecutive frames of the video. A threshold is set to identify the shot boundaries of the video. The proposed system can detect abrupt transitions effectively with less false detection in the uncompressed domain. Findings: The proposed system can able to achieve an average F1 score of 93.51%, which is achieve due to the reduced false detection. Novelty/Improvement: The proposed system uses the GLCM matrix directly instead of calculating the contrast, entropy,etc, i.e., the proposed system is purely based on the correlation of the pixel's co-occurrence. The proposed system also reduces the false detection thereby increasing the precision and F1 score.
The paper proposes a shot boundary detection system using Gist and local descriptor. Gist provides the perceptual and conceptual information of a scene. The proposed system can be roughly divided into three steps. The first step consists of forming of groups of frames by calculating the correlation of the Gist features between consecutive frames of the video. Secondly, abrupt transitions are found out using the group (G), MSER and a threshold (for abrupt separately, \(th_{cut}\)). And lastly, gradual transitions of the video are found using triangular pattern matching. We have performed the experiment on TRECVid 2001 and 2007 dataset. The novel contribution of this paper is that the proposed system shows an activity-based shot boundary detection where only the possible transition regions of a video are considered for shot detection. This approach reduces the computational complexity by processing the transition regions only. We have achieved better results in terms of F1, precision and recall, when compared to previously published approaches.