[Show abstract][Hide abstract] ABSTRACT: We have designed, fabricated, and measured the performance of a linear and area-efficient implementation of the compressive sensing (CS) method in CMOS image sensors. The use of an active pixel sensor (APS) with an integrator and in-pixel current switches are exploited to develop a compact implementation of CS encoding in analog domain. The intrinsic linearity of APS with integrator circuit guarantees the linearity of the CS encoding structure. The CS measurement process is performed for different blocks of the imager separately. This block-based implementation provides individual access to all the pixels from outside the array, resulting in a scalable design with relatively high fill-factor using only two transistors in each pixel. The CS-CMOS image sensor is designed and fabricated in 130-nm technology for , , and arrays. The linearity of the extracted measurement is confirmed by the experimental results from and blocks. In addition, the block readout scheme and the scalability of the design is examined by fabricating a larger array of blocks.
[Show abstract][Hide abstract] ABSTRACT: In this paper, we propose a homography-aware semi-supervised formulation for the logo-based indoor localization problem using smartphone cameras. Our method labels unmatched feature points detected inside the logo parts of query images with their estimated 3D coordinates. The 3D coordinates are computed using the homography estimated from the matched features. We demonstrate the accuracy improvement and lower localization error variance resuLTEd from our semi-supervised approach via experiments in an indoor scenario.
[Show abstract][Hide abstract] ABSTRACT: Microarray technology is a powerful tool in molecular biology which is used for concurrent monitoring of large number of genes expressions. Each microarray experiment produces hundreds of images. Each digital image requires a large storage space. Hence, real-time processing of these images and transmission of them necessitates efficient and custom-made lossless compression schemes. In this paper, we present a hardware scheme for lossless compression scheme of 16-bit microarray images. The most significant part of the image pixels are compressed by classifying them into foreground and background regions. The foreground regions are packed together using a novel compaction unit. Same is performed for the background part. It is shown that a prediction process and a subsequent entropy coding of the packed pixels result in lower entropy. The segmentation map is also compressed with the same proposed compaction unit. The least significant part of the pixels is transmitted without compression. Real-time compression of these images is achieved while the bit-per-pixel values are comparable with the standard offline compression tools.
[Show abstract][Hide abstract] ABSTRACT: Multi-focus image fusion has emerged as a major topic in image processing to generate all-focus images with increased depth-of-field from multi-focus photographs. Different approaches have been used in spatial or transform domain for this purpose. But most of them are subject to one or more of image fusion quality degradations such as blocking artifacts, ringing effects, artificial edges, halo artifacts, contrast decrease, sharpness reduction, and misalignment of decision map with object boundaries. In this paper we present a novel multi-focus image fusion method in spatial domain that utilizes a dictionary which is learned from local patches of source images. Sparse representation of relative sharpness measure over this trained dictionary are pooled together to get the corresponding pooled features. Correlation of the pooled features with sparse representations of input images produces a pixel level score for decision map of fusion. Final regularized decision map is obtained using Markov Random Field (MRF) optimization. We also gathered a new color multi-focus image dataset which has more variety than traditional multi-focus image sets. Experimental results demonstrate that our proposed method outperforms existing state-of-the-art methods, in terms of visual and quantitative evaluations.
Information Fusion 11/2014; 25. DOI:10.1016/j.inffus.2014.10.004 · 3.68 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: We propose a method of using a support vector machine (SVM) to select between multiple well-performing contemporary denoising algorithms for each pixel of a noisy image. We describe a number of novel and pre-existing features based on seam energy, local colour, and saliency which are used as inputs to the SVM. Our SVM strategic image denoising (SVMSID) results demonstrate better image quality than either candidate denoising algorithm, as measured using the perceptually-based quaternion structural similarity image metric (QSSIM).
2014 IEEE International Conference on Image Processing (ICIP), Paris, France; 10/2014
[Show abstract][Hide abstract] ABSTRACT: Retargeting algorithms are needed to transfer an image from a device to another with different size and resolution. The goal is to preserve the best visual quality for important objects of the original image. In order to reduce image size, pixels should be removed from less important parts of the image. Therefore, we need an energy function to select less important pixels in seam carving. Various energy functions have been proposed in previous works to minimize the distortion in salient objects. In this paper we combine three different importance maps to form a new energy map. We first use both gradient and depth maps to highlight the values in the saliency map, eventually generates the final energy map. Experimental results using the proposed energy map show better visual appearance in comparison to previous algorithms even at high resizing percentage. The visual artifacts that cause shape deformation in salient objects and deteriorates geometrical consistency of the scene are considerably reduced in our proposed algorithm.
2014 IEEE International Conference on Image Processing, ICIP 2014; 09/2014
[Show abstract][Hide abstract] ABSTRACT: Three dimensional MRI images which are powerful tools for diagnosis of many diseases require large storage space. A number of lossless compression schemes exist for this purpose. In this paper we propose a new approach for lossless compression of these images which exploits the inherent symmetry that exists in 3D MRI images. First, an efficient pixel prediction scheme is used to remove correlation between pixel values in an MRI image. Then a block matching routine is employed to take advantage of the symmetry within the prediction error image. Inter-slice correlations are eliminated using another block matching. Results of the proposed approach are compared with the existing standard compression techniques.
Multimedia Tools and Applications 08/2014; DOI:10.1007/s11042-014-2214-9 · 1.35 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: An implementation of the compressive sensing (CS) method with a CMOS image sensor is presented. The conventional three-transistor active pixel sensor (APS) structure and switched capacitor circuits are exploited to develop an analog implementation of the CS encoding in a CMOS sensor. With the analog implementation, the sensing and encoding are performed in the same time interval and making a real-time encoding process to optimize the frame rate of the imager. A block readout strategy is proposed to capture the required CS measurements for different blocks of the image, rather than the common column-row readout method. All measurement circuits are placed outside the array by this readout strategy, and the imager becomes scalable for larger array sizes. Because there is no extra in-pixel element for the CS measurement process, the fill factor of the imager is the same as its corresponding APS imager without CS. The proposed structure is designed and fabricated in 0.13- (mu ) m CMOS technology for a (2times 2) array. The experimental results confirm the validity of the design in making monotonic and appropriate CS measurements. The functionality of the block readout method and the scalability of the imager are confirmed by fabrication of a (4times 4) block and a (16times 16) array.
[Show abstract][Hide abstract] ABSTRACT: With the increasing demand for image-based applications, the efficient and
reliable evaluation of image quality has increased in importance. Measuring the
image quality is of fundamental importance for numerous image processing
applications, where the goal of image quality assessment (IQA) methods is to
automatically evaluate the quality of images in agreement with human quality
judgments. Numerous IQA methods have been proposed over the past years to
fulfill this goal. In this paper, a survey of the quality assessment methods
for conventional image signals, as well as the newly emerged ones, which
includes the high dynamic range (HDR) and 3-D images, is presented. A
comprehensive explanation of the subjective and objective IQA and their
classification is provided. Six widely used subjective quality datasets, and
performance measures are reviewed. Emphasis is given to the full-reference
image quality assessment (FR-IQA) methods, and 9 often-used quality measures
(including mean squared error (MSE), structural similarity index (SSIM),
multi-scale structural similarity index (MS-SSIM), visual information fidelity
(VIF), most apparent distortion (MAD), feature similarity measure (FSIM),
feature similarity measure for color images (FSIMC), dynamic range independent
measure (DRIM), and tone-mapped images quality index (TMQI)) are carefully
described, and their performance and computation time on four subjective
quality datasets are evaluated. Furthermore, a brief introduction to 3-D IQA is
provided and the issues related to this area of research are reviewed.
[Show abstract][Hide abstract] ABSTRACT: We propose a weighted KNN Epipolar Geometry-based method for vision-based indoor localization using cellphone cameras. The proposed method is applicable for fine localization whenever a pose-tagged (position + rotation matrix) image database is available rather than just Geo-tagged one. To the best of our knowledge, this is the first that Epipolar geometry has been utilized for fine localization in indoor applications using smartphone images. We compare the performance of our method with two outstanding literature works. It will be also demonstrated that the proposed method can extrapolate the location of queries located outside of the database location set, as well as compensate for the small databases, where database location set is sparse as two additional new features of this method.
2014 IEEE 8th Sensor Array and Multichannel Signal Processing Workshop (SAM); 06/2014
[Show abstract][Hide abstract] ABSTRACT: Dual-mode (hybrid) cameras are able to simultaneously shoot two types of video streams: a high resolution with low frame rate stream and similarly a low resolution with high frame rate stream. There are some works on super-resolving the single camera video sequences. In this paper we propose a method for multi-view video super-resolution, utilizing sequences generated from a hybrid camera. In the proposed method we exploit the self-similarity in the spatial and temporal domains to reconstruct a pixel. We modify the nonlocal means method to be applied in our method. Through a combination of techniques including adaptive thresholds and specialized candidate pixel selection schemes, the proposed method reconstructs a high fidelity video stream with considerably improved performance.
International Conference on Image Processing (ICIP), Paris, France; 01/2014
[Show abstract][Hide abstract] ABSTRACT: Technologies are generally described for automatically optimizing an efficiency of camera placement, numbers, and resolution in multi-camera monitoring and surveillance applications. In some examples, a fraction of a total area may be monitored at a higher resolution than the rest. Employing techniques such as combinatorial state Viterbi technique or combinatorial state trellis technique, a minimum number of cameras that provide the coverage at the needed resolution may be selected. Similarly, a number of points may be covered with at least a predefined number of cameras. For example, a subject of interest may be tracked in a public area, where specific camera(s) may be used to image the subject's face at a higher resolution than the background.
[Show abstract][Hide abstract] ABSTRACT: In some applications such as digital video broadcasting, video is transmitted over a low capacity channel with lower frame rates. The lower the frame rate, the jerkier or unevener the video motion would be noticed. To solve this problem, frame rate up conversion (FRUC) is employed to increase the frame rate. In this paper, we propose a new FRUC method using the nonlocal-means estimator. In this method, a pixel is reconstructed as a weighted linear combination of pixel pairs in its adjacent frames. The pixels of each pair are temporally symmetric from the view point of the pixel being interpolated. The weights are calculated based on the self-similarity assumption. To reduce the computational complexity, we calculate the weights of linear combination for each super-pixel. Experimental results show the superior performance of our proposed method in comparison to the existing methods.
IEEE International Conference on Image Processing (ICIP), Paris, France; 01/2014
[Show abstract][Hide abstract] ABSTRACT: A novel formulation for de-interlacing problem is provided. De-interlacing problem can be reconsidered as estimating the best sequence of interpolators from a set of candidate interpolators output from a finite-state-machine. This way we formulate the problem with the help of suitable assumptions to finally exploit the structure of trellis diagrams with Viterbi and forward-backward algorithm to efficiently solve the estimation problem. At the Next step,we consider addition of a motion compensation scheme to the set of candidate interpolators for the second pass trellis processing.Finally we will reduce the complexity of the algorithm by using a motion detector to predict the blocks of the image frame in which motion compensation method will probably outperform the other candidates.This way we just insert the result from motion compensation in these blocks instead of adding the motion compensated method to the the candidate set for further trellis processing.We provide that the proposed algorithm will outperform many of the well-known deinterlacing algorithms and also benefits from a parameter-free structure.
[Show abstract][Hide abstract] ABSTRACT: Determining an object location in a specific region is an important task in many machine vision applications. Different parameters affect the accuracy of the localization process. The quantization process in charge-coupled device of a camera is one of the sources of error that causes estimation rather than identifying the exact position of the observed object. A cluster of points, in the field of view of a camera are mapped into a pixel. These points form an uncertainty region. In this paper, we present a geometrical model to analyze the volume of this uncertainty region as a criterion for object localization error. The proposed approach models the field of view of each pixel as an oblique cone. The uncertainty region is formed via the intersection of two cones, each emanating from one of the two cameras. Because of the complexity in modeling of two oblique cones' intersection, we propose three methods to simplify the problem. In the first two methods, only four lines are used. Each line goes through the camera's lens, modeled as a pinhole, and then passes one of the four vertices of a square that is fitted around the circular pixel. The first proposed method projects all points of these four lines into an image plane. In the second method, the line-cone intersection is used instead of intersection of two cones. Therefore, by applying line-cone intersection, the boundary points of the intersection of the two cones are determined. In the third approach, the extremum points of the intersection of two cones are determined by the Lagrangain method. The validity of our methods is verified through extensive simulations. In addition, we analyze effects of parameters, such as the baseline length, focal length, and pixel size, on the amount of the estimation error.
[Show abstract][Hide abstract] ABSTRACT: This paper introduces a new approach for video super resolution problem. To this end Compressive Sensing (CS) theory along with contourlet transform has been used. In CS framework the signal is assumed to be sparse in a transform domain. An approach has been suggested using this fact in which contourlet domain is used as the transform domain and a CS algorithm helps to find the high resolution frame. A post processing step is applied afterward to the estimated outputs to increase the quality. The post processing step consists of a deblurring term and a Bilateral Total Variation (BTV) filter for increasing the consistency. This method helps to relax the conditions on hardware and increase the quality of the video after capturing, in fact the quality of the video streams in consumer applications can be increased even the capturing device represents the scene in a low resolution format. Experimental results show significant improvement over existing super resolution methods in both objective and subjective quality.