The Journal of Electronic Imaging would like to sincerely thank the following individuals who served as reviewers in 2012. The success of our publication hinges on the voluntary contributions of time and energy put forth by these professionals.

Compressive sensing (CS) is an emerging field of research with far-reaching impact on a variety of applications. For signals admitting sparse representations, CS permits collection of a significantly reduced number of measurements than required by the Shannon-Nyquist sampling theorem and provides efficient superresolution signal reconstruction. In essence, CS provides means for reducing data volume and efficient sensor deployment without a degradation in system performance. Such capabilities are highly desirable for a range of civil and military applications.

We propose a multiscale, iterative algorithm for reconstructing video signals from streaming compressive measurements. Our algorithm is based on the observation that, at the imaging sensor, many videos should have limited temporal bandwidth due to the spatial lowpass filtering that is inherent in typical imaging systems. Under modest assumptions about the motion of objects in the scene, this spatial filtering prevents the temporal complexity of the video from being arbitrarily high. Thus, even though streaming measurement systems may measure a video thousands of times per second, we propose an algorithm that only involves reconstructing a much lower rate stream of “anchor frames.” Our analysis of the temporal complexity of videos reveals an interesting tradeoff between the spatial resolution of the camera, the speed of any moving objects, and the temporal bandwidth of the video. We leverage this tradeoff in proposing a multiscale reconstruction algorithm that alternates between video reconstruction and motion estimation as it produces finer resolution estimates of the video.

We consider sparsity-driven joint localization of stationary and moving targets inside enclosed structures using a reduced set of spatial-fast time-slow time observations in ultra-wideband (UWB) pulsed radar platforms. We exploit the compact temporal support of the UWB signal to suppress the front wall clutter through time gating. The resulting enhancement in the signal-to-clutter ratio enables application of compressive sensing (CS) for scene reconstruction. We establish an appropriate signal model that permits formulation of linear modeling with sensing matrices so as to achieve efficient CS-based localization of stationary and moving targets in the downrange-crossrange-velocity space. We demonstrate the effectiveness of the proposed scheme using real data collected in a laboratory environment.

We consider imaging of the building interior structures using compressive sensing (CS) with applications to through-the-wall imaging and urban sensing. We consider a monostatic synthetic aperture radar imaging system employing stepped frequency waveform. The proposed approach exploits prior information of building construction practices to form an appropriate sparse representation of the building interior layout. We devise a dictionary of possible wall locations, which is consistent with the fact that interior walls are typically parallel or perpendicular to the front wall. The dictionary accounts for the dominant normal angle reflections from exterior and interior walls for the monostatic imaging system. CS is applied to a reduced set of observations to recover the true positions of the walls. Additional information about interior walls can be obtained using a dictionary of possible corner reflectors, which is the response of the junction of two walls. Supporting results based on simulation and laboratory experiments are provided. It is shown that the proposed sparsifying basis outperforms the conventional through-the-wall CS model, the wavelet sparsifying basis, and the block sparse model for building interior layout detection.

The quality control of composite multilayered materials and structures using nondestructive tests is of high interest for numerous applications in the aerospace and aeronautics industry. One of the established nondestructive methods uses microwaves to reveal defects inside a three-dimensional (3-D) object. Recently, there has been a tendency to extrapolate this method to higher frequencies (going to the subterahertz spectrum) which could lead to higher resolutions in the obtained 3-D images. Working at higher frequencies reveals challenges to deal with the increased data rate and to efficiently and effectively process and evaluate the obtained 3-D imagery for defect detection and recognition. To deal with these two challenges, we combine compressive sensing (for data rate reduction) with a dedicated image processing methodology for a fast, accurate, and robust quality evaluation of the object under test. We describe in detail the used methodology and evaluate the obtained results using subterahertz data acquired of two calibration samples with a frequency modulated continuous wave system. The applicability of compressive sensing within this context is discussed as well as the quality of the image processing methodology dealing with the reconstructed images.

Compressed sensing (CS) has great potential for use in video data acquisition and storage because it makes it unnecessary to collect an enormous amount of data and to perform the computationally demanding compression process. We propose an effective CS algorithm for video that consists of two iterative stages. In the first stage, frames containing the dominant structure are estimated. These frames are obtained by thresholding the coefficients of similar blocks. In the second stage, refined residual frames are reconstructed from the original measurements and the measurements corresponding to the frames estimated in the first stage. These two stages are iterated until convergence. The proposed algorithm exhibits superior subjective image quality and significantly improves the peak-signal-to-noise ratio and the structural similarity index measure compared to other state-of-the-art CS algorithms.

We introduce a robust image-formation approach for through-the-wall radar imaging (TWRI). The proposed approach consists of two stages involving compressive sensing (CS) followed by delay-and-sum (DS) beamforming. In the first stage, CS is used to reconstruct a complete set of measurements from a small subset collected with a reduced number of transceivers and frequencies. DS beamforming is then applied to form the image using the reconstructed measurements. To promote sparsity of the CS solution, an overcomplete Gabor dictionary is employed to sparsely represent the imaged scene. The new approach requires far fewer measurement samples than the conventional DS beamforming and CS-based TWRI methods to reconstruct a high-quality image of the scene. Experimental results based on simulated and real data demonstrate the effectiveness and robustness of the proposed two-stage image formation technique, especially when the measurement set is drastically reduced.

Spatial sparsity of the target space in subsurface or through-the-wall imaging applications has been successfully used within the compressive-sensing framework to decrease the data acquisition load in practical systems, while also generating high-resolution images. The developed techniques in this area mainly discretize the continuous target space into grid points and generate a dictionary of model data that is used in image-reconstructing optimization problems. However, for targets that do not coincide with the computation grid, imaging performance degrades considerably. This phenomenon is known as the off-grid problem. This paper presents a novel sparse ground-penetrating radar imaging method that is robust for off-grid targets. The proposed technique is an iterative orthogonal matching pursuit-based method that uses gradient-based steepest ascent-type iterations to locate the off-grid target. Simulations show that robust results with much smaller reconstruction errors are obtained for multiple off-grid targets compared to standard sparse reconstruction techniques.

We propose a sample selection method for multiple-input, multiple-output ultra-wideband noise radar imaging using compressive sensing. The proposed sample selection is based on comparing the norm values of candidates among the potential received signal and selecting the largest M samples among N per antenna to obtain selection diversity. Moreover, we propose an adaptive weighting allocation that improves reconstruction accuracy of compressive sensing by maximizing the mutual information between target echoes and transmitted signals. This weighting scheme is applicable to both sample selection schemes, a conventional random sampling and the proposed selection. Further, the weighting allocation with the knowledge of recovery error is proposed for more practical scenarios. Simulations show that the proposed selection and weighting allocation enhance multiple target detection probability and reduce normalized mean square error.

Randomly encoded compressive sensing (CS) has potential in fast acquisition of magnetic resonance imaging (MRI) data in most naturally compressible images. However, there is no guaranteed good performance for general applications by any of the traditional CS-MRI theoretical schemes developed so far. On the other hand, recent research demonstrates that adaptive sampling exploiting the tree structure of nonzero wavelet coefficients of signals allows more control over the sensing procedure in the form of feedback and improve the CS performance significantly. Following recent research strategies in CS-MRI, well-known adaptive sampling strategies in the wavelet domain, as used in image compression, to encode the MRI data yielding good reconstruction quality are introduced. Based on this underlying characteristic, adaptive k -space trajectories are designed with tailored spatially selective RF excitation pulses created by Battle-Lemarie wavelet functions. The input vectors formed from these significant samples of multilevel wavelet decomposed images are used in a CS framework for reconstruction of MR images. This MR image reconstruction uses a CS algorithm based on the minimization of total-variation regularized signal to provide stable results. The simulated results show that this approach can reduce almost 70% of MR image acquisition time and achieve good reconstructed image quality.

Compressive sensing (CS) theory has drawn great interest in recent years and has led to new image-acquisition techniques in many different fields. This research investigates a CS-based active underwater laser serial imaging system, which employs a spatial light modulator (SLM) at the source. A multiscale polarity-flipping measurement matrix and a model-assisted image reconstruction concept are proposed to address limitations imposed by a scattering medium. These concepts are also applicable to CS-based imaging in atmospheric environments characterized by fog, rain, or clouds. Simulation results comparing the performance of the proposed technique with that of traditional laser line scan (LLS) sensors and other structured illumination-based imager are analyzed. Experimental results from over-the-air and underwater tests are also presented. The potential for extending the proposed frame-based imaging technique to the traditional line-by-line scanning mode is discussed.

Compressive sensing makes it possible to recover sparse target scenes from under-sampled measurements when uncorrelated random-noise waveforms are used as probing signals. The mathematical theory behind this assertion is based on the fact that Toeplitz and circulant random matrices generated from independent identically distributed (i.i.d) Gaussian random sequences satisfy the restricted isometry property. In real systems, waveforms have smooth, nonideal autocorrelation functions, thereby degrading the performance of compressive sensing algorithms. Compressive sensing requires the system matrix to have particular properties. Incorporating prior information into the target scene either to enhance imaging or to mitigate nonidealities can result in system matrices that are not suitable for compressive sensing. We can overcome this problem by designing appropriate transmit waveforms. We extend the existing theory to incorporate such nonidealities into the analysis of compressive recovery. As an example we consider the problem of tailoring waveforms to image extended targets. Extended targets make the target scene denser, causing random transmit waveforms to be suboptimal for recovery. We propose to incorporate extended targets by considering them to be sparsely representable in redundant dictionaries. We demonstrate that a low complexity algorithm to optimize the transmit waveform leads to improved performance.

Compressed sensing is the theory and practice of sub-Nyquist sampling of sparse signals of interest. Perfect reconstruction may then be possible with significantly fewer than the Nyquist required number of data. In this work, we consider a video system where acquisition is performed via framewise pure compressed sensing. The burden of quality video sequence reconstruction falls, then, solely on the decoder side. We show that effective decoding can be carried out at the receiver/decoder side in the form of interframe total variation minimization. Experimental results demonstrate these developments.

We propose a method to acquire compressed measurements for efficient video reconstruction using a single-pixel camera. The method is suitable for implementation using a single-pixel detector, along with a digital micromirror device or other types of spatial light modulators. Conventional implementations of single-pixel cameras are able to spatially compress the signal, but the compressed measurements make it difficult to exploit temporal redundancies directly. Moreover, a single-pixel camera needs to make measurements in a sequential manner before the scene changes, making it inefficient for video imaging. We discuss a measurement scheme that exploits sparsity along the time axis for video imaging. After acquiring all measurements required for the first frame, measurements are acquired only from the areas that change in subsequent frames. We segment the first frame, detect the magnitude and direction of change for each segment, and acquire compressed measurements for the changing segments in the predicted direction. Next, we compare the reconstruction results for a few test sequences with existing techniques and demonstrate the practical utility of the scheme.

Here, a new low-complexity block motion estimation method is presented. The proposed method significantly reduces the computational complexity by checking only one of two search areas in the conventional hybrid method when the two search centers are close to each other. It also improves the performance by efficiently making use of the motion vector information of the co-located macroblock. The proposed techniques can be easily applied to many hybrid motion estimation methods.

Recently, tracking is regarded as a binary classification problem by discriminative tracking methods. However, such binary classification may not fully handle the outliers, which may cause drifting. We argue that tracking may be regarded as one-class problem, which avoids gathering limited negative samples for background description. Inspired by the fact the positive feature space generated by one-class support vector machine (SVM) is bounded by a closed hyper sphere, we propose a tracking method utilizing one-class SVMs that adopt histograms of oriented gradient and 2bit binary patterns as features. Thus, it is called the one-class SVM tracker (OCST). Simultaneously, an efficient initialization and online updating scheme is proposed. Extensive experimental results prove that OCST outperforms some state-of-the-art discriminative tracking methods that tackle the problem using binary classifiers on providing accurate tracking and alleviating serious drifting.

Our purpose is to present an intrafield deinterlacing method using the Bayes classifier. The conventional intrafield deinterlacing methods interpolate the pixel along the local edge direction, but they yield interpolation errors when the local edge direction is determined to be wrong. On the basis of the Bayes classifier, the proposed algorithm performs region-based deinterlacing. The proposed algorithm utilizes an input feature vector that includes five directional correlations, which are used to extract the characteristics of the local region, to classify the local region. After the classification of the local region, one of the three simple interpolation methods, which possesses the highest probability to be used among the three, is chosen for the corresponding local region. In addition, we categorized the range of the feature vector to reduce the computational complexity. Simulation results show that the proposed Bayes classifier-based deinterlacing method minimizes interpolation errors. Compared to the traditional deinterlacing methods and Wiener filter-based interpolation method, the proposed method improves the subjective quality of the reconstructed image, and maintains a higher peak signal-to-noise ratio level.

A digital halftone of blue noise characteristics is preferred as dots in the halftone of a constant input should be isotropically and homogeneously distributed. In practice, the placement of dots is constrained by a sampling grid and hence aliasing happens when the input gray level is in the middle range. To solve this problem, Lau et al. suggested replacing isolated dots by dot clusters to maintain the principal frequency of the output to be 1/2 when this happens. However, this model does not take into account that due to the stochastic nature of the dot distribution there is a considerable amount of energy distributed around the principal frequency and it causes aliasing problems even when the principal frequency of the output is 1/2 . Here, we present a new noise model that takes this factor into account. A halftoning algorithm is then proposed to generate halftones that satisfy the new noise model. By comparing its performance with that of some other algorithms that are proposed based on the traditional blue noise model and Lau et al.’s noise model, one can see that the proposed noise model can be a better model to describe the noise characteristics of a high-quality halftone.

This paper proposes an automatic video matting method that uses hybrid video object segmentation and closed-form matting. Hybrid video object segmentation, based on background construction-based video object segmentation and foreground extraction-based video object segmentation, is used to obtain reliable foreground objects. Next, the trimaps are automatically generated using two different scanning windows from the hybrid video object segmentation results. Finally, closed-form matting is used, with the automatically generated trimaps, to yield the alpha mattes and the foreground objects with an opacity estimate. Experimental results show that the proposed method outperforms state-of-the-art automatic video matting methods based on closed-form matting.

Retinex algorithms have been widely applied in many aspects of image processing. Based on the iterative Retinex algorithm, we propose an edge-preserving illumination estimation method. Inspired by the anisotropic diffusion, an edge-stopping function is introduced in the iterative computation. This modification enables the preservation of abrupt edges when computing the upper envelope of a given image. Based on the illumination-reflectance decomposition, a high-dynamic-range (HDR) radiance map can be easily tone-mapped to be a low-dynamic-range image by compressing the range of the estimated illumination. Artifacts are effectively suppressed using the proposed method. Meanwhile, we also propose a jumping-spiral iteration manner to improve the symmetry of the edge response. Experimental results show that the proposed tone mapping algorithm is very effective in reproducing HDR scenes, and has a better performance compared with some similar operators.

The local binary pattern (LBP) operator is a very effective texture descriptor that describes images using texture patterns. However, existing LBP operators discard the texture contrast information by definition and, in addition, they are sensitive to noise. This paper presents a local contrast descriptor (LCD) to represent images using texture contrast by measuring how far neighbors are spread out in a given neighborhood. LCD and LBPs can be combined to improve texture description performance. Furthermore, the LCD is extended to the local ternary contrast descriptor (LTCD) to describe textures using both contrast and magnitude features. Experiments on the UIUC texture database show that the LCD and LBP combinations are more accurate than either descriptor alone and LCD is an easy and efficient complement to LBPs. Experiments on the FERET face database and the database in which the probe sets contain added noise show that LCD, especially LTCD, has promising discriminative power and strong robustness against noise.

Electrical impedance tomography (EIT) is an imaging modality in which the conductivity distribution inside a target is reconstructed based on voltage measurements from the target’s surface. Reconstructing the conductivity distribution is known to be an ill-posed inverse problem whose solutions are highly intolerant to modeling errors. To achieve sufficient accuracy, very dense meshes are usually needed in a finite element approximation of the EIT forward model. This leads to very high-dimensional problems and often unacceptably tedious computations for real-time applications. We consider the model reduction in EIT within the Bayesian inversion framework. We construct the reduced-order model by proper orthogonal decompositions (POD) of the electrical conductivity and the potential distributions. The associated POD modes are computed based on a priori information on the conductivity. The feasibility of the reduced-order model is tested both numerically and with experimental data. The proposed approach is shown to speed up EIT reconstruction considerably without significantly decreasing image quality. In the selected test cases, high-quality reconstructions are obtained with the reduced-order model in 3.5% to 5% of the time required for conventional reconstructions.

Here, we propose a confidence shape metric for level set segmentation. First, the confidence shape metric, which encodes local confidence into the matching result, is used in matching shapes and producing confidence maps. Then, based on the confidence shape prior, the level set function evolves to a desired contour. The proposed shape metric allows representation of shape variations beyond the coverage of training shapes under the level set framework, which is suitable for segmenting strongly deformed and cluttered images, especially when the set of training shapes is sparse compared with numerous intracategory variations. We evaluated the proposed approach on the challenging Weizmann dataset and computed tomography images. Experimental results indicate the advantage of confidence shape prior over shape prior without confidence under the Dice-coefficient metric.

A new image enhancement technique based on a self-tunable transformation function to improve the visual quality of images captured with low dynamic range devices in extreme lighting conditions is presented. This technique consists of four processes: histogram adjustment, dynamic range compression, contrast enhancement, and nonlinear color restoration. Histogram adjustment on each spectral band is performed to minimize the effect of illumination. Dynamic range compression is accomplished by a newly designed inverse sine nonlinear function that provides various nonlinear curvatures with an image dependent parameter. A nonlinear curve generated by this parameter is used to modify the intensity of each pixel in the luminance image. A nonlinear color restoration process based on the chromatic information and luminance of the original image is employed. The effectiveness of this technique is evaluated on various natural images and aerial images, and compared with other state-of the art techniques. A quantitative evaluation is performed by estimating the number of Harris corners and speeded up robust features on wide area motion imagery data. The application of the proposed algorithm on face detection is also demonstrated. The evaluation results demonstrate that the proposed method holds significant benefits for surveillance and security applications and also as a preprocessing technique for object detection and tracking applications.

We approach the problem around several main directions of video temporal segmentation and propose an intensity-based dissolve detection approach that is able to perform on animated video contents. It uses the hypothesis that during a dissolve the amount of fading-out and fading-in pixels should be significant compared with other visual transitions. We use this information as a visual discontinuity function. Instead of just applying a global threshold to filter these values, as most of the existing approaches do, we use a twin-thresholding approach and the shape analysis of the discontinuity measure. This allows us to reduce false detections caused by steep intensity fluctuations as well as to retrieve dissolves caught up in other visual transitions (e.g., caused by movement, color effects, etc.). Experimental tests conducted on more than 452 dissolve transitions show that whether classic approaches tend to fail, the proposed method is able to provide good performance achieving average precision and recall ratios above 94% and 79.6%, respectively.

With the rapid development of multispectral imaging technique, it is desired that the spectral color can be accurately reproduced using desktop color printers. However, due to the specific spectral gamuts determined by printer inks, it is almost impossible to exactly replicate the reflectance spectra in other media. In addition, as ink densities can not be individually controlled, desktop printers can only be regarded as red-green-blue devices, making physical models unfeasible. We propose a locally adaptive method, which consists of both forward and inverse models, for desktop printer characterization. In the forward model, we establish the adaptive transform between control values and reflectance spectrum on individual cellular subsets by using weighted polynomial regression. In the inverse model, we first determine the candidate space of the control values based on global inverse regression and then compute the optimal control values by minimizing the color difference between the actual spectrum and the predicted spectrum via forward transform. Experimental results show that the proposed method can reproduce colors accurately for different media under multiple illuminants.

In recent years, verification based on thermal face images has been extensively studied because of its invariance to illumination and immunity to forgery. However, most of them have not given full consideration to high-verification performance and singular within-class scatter matrix problems. We propose a novel thermal face verification algorithm, which is named two-directional two-dimensional modified Fisher principal component analysis. First, two-dimensional principal component analysis (2-DPCA) is utilized to extract the optimal projective vector in the row direction. Then, 2-D modified Fisher linear discriminant analysis is implemented to overcome the singular within-class scatter matrix problem of the 2-DPCA space in the column direction. Comparative experiments on the natural visible and infrared facial expression thermal face subdatabase demonstrate that the proposed approach outperforms state-of-the-art methods in terms of verification performance.

The processing pipeline of a digital camera converts the RAW image acquired by the sensor to a representation of the original scene that should be as faithful as possible. There are mainly two modules responsible for the color-rendering accuracy of a digital camera: the former is the illuminant estimation and correction module, and the latter is the color matrix transformation aimed to adapt the color response of the sensor to a standard color space. These two modules together form what may be called the color correction pipeline. We design and test new color correction pipelines that exploit different illuminant estimation and correction algorithms that are tuned and automatically selected on the basis of the image content. Since the illuminant estimation is an ill-posed problem, illuminant correction is not error-free. An adaptive color matrix transformation module is optimized, taking into account the behavior of the first module in order to alleviate the amplification of color errors. The proposed pipelines are tested on a publicly available dataset of RAW images. Experimental results show that exploiting the cross-talks between the modules of the pipeline can lead to a higher color-rendition accuracy.

Dimensional mismatch between a low-resolution (LR) surveillance face image and its high-resolution (HR) template makes recognition very difficult. A novel method called coupled cross-regression (CCR) is proposed to deal with this problem. Instead of processing in the original observing space directly, CCR projects LR and HR face images into a unified low-embedding feature space. Spectral regression is employed to improve generalization performance and reduce computational complexity. Meanwhile, cross-regression is developed to utilize HR embedding to increase the information of the LR space, thus improving the recognition performance. Experiments on the FERET and CMU PIE face database show that CCR outperforms existing structure-based methods in terms of recognition rate as well as computational complexity.

Block-matching motion estimation (ME) is a principal process for most existing video compression standards, but it has a significant impact on computation overhead. An efficient block-matching algorithm is proposed for ME. The proposed algorithm consists of three effective steps: (1) apply the full search (FS) algorithm to construct a polynomial interpolation model from the group with four skipping frames to determine near-optimal global motion vector (MV) for the initial search point, (2) perform an adaptive search pattern in the interpolation step to refine the MV and (3) use the half-way stop technique to reduce search points. Experimental results show that the proposed algorithm achieves a maximum speed-up factor of 366.47 with only 0.52% peak signal-to-noise ratio (PSNR) degradation in comparison with the FS algorithm.

For local smooth regions in multifocus images, it is difficult to judge whether they are in focus or not, whether using human eyes or special focus measures. We propose to classify the images into smooth and nonsmooth regions based on structural similarity index. Quaternion wavelet transform (QWT), as a novel tool of image analysis, has some superior properties compared to discrete wavelet transform, such as nearly shift-invariant wavelet coefficients and phase-based texture representation. We use the local variance of the QWT phases to detect the focus position for the pixels belonging to the nonsmooth image regions. Thus, binary images of the left-focus, right-focus, and smooth region, e.g., there are two different focuses, are obtained. Then, the connected components labeling algorithm is exploited to label the two binary images containing the focus position information, and the regions with focus measure errors are transferred between the two binary images. The fusion result is finally acquired through three weighted binary images combined with the original multifocus images. Furthermore, we conduct several experiments to verify the feasibility of the proposed fusion method. The performance is demonstrated to be superior to current methods.

A fast, low-bit-depth representation–based block motion estimation (ME) approach is presented. It basically utilizes truncated gray-coded bit planes for matching and introduces an efficient adaptive search range and early termination approaches to speed up ME processes. All the additional operations used decide adaptive search range, and early termination is computationally lightweight. Experimental results show that the proposed approach provides similar quality in terms of objective assessment by significantly reducing computational load compared to full-search based ME.

We propse the one-line search (OLS) method for estimating the three-dimensional (3-D) model of an object from a sequence of images where the object is visible from different points of view. The OLS method estimates each 3-D point through a OLS along the ray defined by the camera center and its two-dimensional (2-D) image projection in one view by maximizing a photo-consistency index built with the other views. The key advantage of the proposed method is that only one variable is introduced for estimating the sought 3-D point, whereas approaches for multiview stereo typically exploit multiple epipolar lines and hence require multiple variables. Results with both synthetic and real data illustrate and validate the proposed method.

One of the most recent techniques developed for face recognition is diagonal principal component analysis (DiaPCA). In contrast to two-dimensional PCA (2DPCA), DiaPCA is based on diagonal representation of face images. Therefore, the correlations between variations of rows and those of columns can be kept. This paper proposes a novel projection scheme called bilateral DiaPCA (BDiaPCA) which simultaneously performs left and right projections of original face images in a new subspace derived from diagonal representation. Therefore BDiaPCA reduces the dimension of the original image matrix in both column and row directions. Experiments were carried out under varying illumination and facial expression using three well-known face databases: Yale, PF01, and a subset of FERET containing 200 people. Comparisons with 2DPCA and DiaPCA show that BDiaPCA outperforms these methods and achieves higher recognition rates using the Yang, Frobenius, or assembled matrix distance metrics.

Here, we focus on the calibration of a multiview shooting system designed for the production of three-dimensional content for auto-stereoscopic visualization. The particularities of this multiview camera are the coplanarity of the image sensors and the shift that each sensor represents with regard to the corresponding optical axis. For the calibration of such cameras, we propose an approach based on Faugéras and Toscani’s calibration method, which herein is extended to a multiview camera with parallel and decentered image sensors. First, the geometrical model of the shooting system is presented along with some industrial prototypes based on a geometrical analysis. Next, the development of the proposed calibration method is described. Last, some simulation results are presented, and some final concluding remarks about this research are then given.

The morphological connected slope filters (MCSFs) are studied as gray level transformations, and two contributions are made on these operators with the purpose of modifying the gradient criterion performance. The proposals consist of: (a) the introduction of three weighting functions and (b) the application of a displacement parameter. The displacement parameter will permit the image segmentation in a certain intensity interval and the contrast improvement at the same time. This characteristic is an important difference among the MCSFs introduced previously, together with the other transformations defined in the current literature utilized uniquely to enhance contrast. Also, an application example of the weighted morphological slope filters is provided. In such an example, white matter is separated from brain magnetic resonance images T1.

Temporal coherence of retargeted video is important. Even a mild temporal inconsistency among subsequent frames can result in visually annoying artifacts, such as jittering or flickering. To reduce these artifacts, we assign a common seam for a group of consecutive frames (GoF). The video sequence is divided into GoFs, and a representative frame for each GoF is formed. Then, both the spatial saliency in each representative frame and the temporal coherence among the consecutive representative frames are considered to determine a common seam for each GoF. Since a fixed seam of the representative frame is applied to the whole frames in a GoF, there will be no jittering and flickering artifacts within the GoF. Also, since the temporal coherence between the two consecutive representative frames (i.e., two consecutive GoFs) was taken into account, the transition between two consecutive GoFs will be smooth.

An iterative algorithm and a denoising method for accuracy enhancement of three-dimensional (3-D) reconstruction processes by shadow moiré is proposed. Using traditional shadow moiré for 3-D reconstruction leads to a nonuniform phase shift error. The proposed iterative algorithm eliminates this error by using three phase-shifted interferograms and obtains precise phase information. Moreover the proposed denoising method, which is based on undecimated wavelet transform, effectively eliminates noise and grating patterns while retaining useful information. The proposed phase shifting shadow moiré method is compared with the structured lighting method, which is a common method for 3-D reconstruction, and it is further compared with traditional shadow moiré. The methods are simulated in a 3ds-Max environment. The simulation results show that the proposed shadow moiré technique achieves greater accuracy in comparison with the traditional shadow moiré and structured lighting techniques and also it has higher accuracy than existing typical phase-shifting algorithms.

High-efficiency video coding (HEVC) introduces a flexible hierarchy of three block structures: coding unit (CU), prediction unit (PU), and transform unit (TU), which have brought about higher coding efficiency than the current national video coding standard H.264/advanced video coding (AVC). HEVC, however, simultaneously requires higher computational complexity than H.264/AVC, although several fast inter-mode decisions were proposed in its development. To further reduce this complexity, a fast inter-mode decision algorithm is proposed based on temporal correlation. Because of the distinct difference of inter-prediction block between HEVC and H.264/AVC, in order to use the temporal correlation to speed up the inter prediction, the correlation of inter-prediction between two adjacent frames needs to be analyzed according to the structure of CU and PU in HEVC. The probabilities of all the partition modes in all sizes of CU and the similarity of CU segmentation and partition modes between two adjacent frames are tested. The correlation of partition modes between two CUs with different sizes in two adjacent frames is tested and analyzed. Based on the characteristics tested and analyzed, at most, two prior partition modes are evaluated for each level of CU, which reduces the number of rate distortion cost calculations. The simulation results show that the proposed algorithm further reduces coding time by 33.0% to 43.3%, with negligible loss in bitrate and peak signal-to-noise ratio, on the basis of the fast inter-mode decision algorithms in current HEVC reference software HM7.0.

Recent advances in semiconductor technology have made it possible to integrate an entire system including processors, memory, and other system units into a single programmable chip [a field programmable gate array (FPGA)]. These configurations, called system on a programmable chip (SOPC), have the advantage of being designed and implemented quicker than traditional technologies and are relatively cheap to produce for low volume applications. One processor-intensive application, which is ideal for SOPC technology, is that of image processing where there is a repeated application of operations on the two-dimensional data. We present the results of research that investigate the use of SOPC technology in building a real-time image-processing system with the capability of performing video acquisition, image processing, and image display. In order to solve the complex on-chip data communication, while not degrading the transfer speed of large amounts of video data, a novel and effective bus arbitration solution called “simultaneously multimastering Avalon streaming transfer with peripheral-controlled waitrequest” is developed. Rather than using the software approach to initialize direct memory access (DMA) like transfers, this solution takes advantage of the FPGA hardware resources to perform bus arbitration and hence increase system efficiency.

A new spectral signature analysis method for tumor segmentation in breast magnetic resonance images is presented. The proposed method is called an independent component texture analysis (ICTA), which consists of three techniques including independent component analysis (ICA), entropy-based thresholding, and texture feature registration (TFR). ICTA was mainly developed to resolve the inconsistency in the results of independent components (ICs) due to the random initial projection vector of ICA and then accordingly determine the most likely IC. A series of experiments were conducted to compare and evaluate ICTA with principal component texture analysis, traditional ICA, traditional principal component analysis (PCA), fuzzy c-means, constrained energy minimization, and orthogonal subspace projection methods. The experimental results showed that ICTA had higher efficiency than existing methods.

Many preprocessing techniques have been proposed for isolated word recognition. However, recently, recognition systems have dealt with text blocks and their compound text lines. In this paper, we propose a new preprocessing approach to efficiently correct baseline skew and fluctuations. Our approach is based on a sliding window within which the vertical position of the baseline is estimated. Segmentation of text lines into subparts is, thus, avoided. Experiments conducted on a large publicly available database (Rimes), with a BLSTM (bidirectional long short-term memory) recurrent neural network recognition system, show that our baseline correction approach highly improves performance.

An efficient face-recognition algorithm is proposed, which not only possesses the advantages of linear subspace analysis approaches—such as low computational complexity—but also has the advantage of a high recognition performance with the wavelet-based algorithms. Based on the linearity of Gabor-wavelet transformation and some basic assumptions on face images, we can extract pseudo-Gabor features from the face images without performing any complex Gabor-wavelet transformations. The computational complexity can therefore be reduced while a high recognition performance is still maintained by using the principal component analysis (PCA) method. The proposed algorithm is evaluated based on the Yale database, the Caltech database, the ORL database, the AR database, and the Facial Recognition Technology database, and is compared with several different face recognition methods such as PCA, Gabor wavelets plus PCA, kernel PCA, locality preserving projection, and dual-tree complex wavelet transformation plus PCA. Experiments show that consistent and promising results are obtained.

The development of an effective and dependable multiple description coding scheme over a lossy communication system is described. A general framework of an effective multiple description robust communication system with two-channel and four-channel cases is presented with a proposed block-based DC coefficient relocation and AC coefficient interpolation approach. A benefit of such system is that, if all the channels work properly, a very good quality, probably lossless, reconstruction can be obtained from the received descriptions. On the other hand, if some of the channels do not work properly, a lower but still quiet satisfactory quality of reconstruction can be obtained. The performance of the proposed scheme is measured with mean squared error, peak signal-to-noise ratio, entropy, and bit rate to analyze the reconstruction quality, and the computed results are matched with that of the other schemes. The system complexity is also computed and compared. Comparing the simulation results, it is observed that the proposed scheme, which uses a DC coefficient relocation and AC coefficient interpolation scheme, gives comprehensive enhancements over the other recently developed schemes.

The H.264/AVC video standard adopts a fixed search range (SR) and fixed reference frame (RF) for motion estimation. These fixed settings result in a heavy computational load in the video encoder. We propose a dynamic SR and multiframe selection algorithm to improve the computational efficiency of motion estimation. By exploiting the relationship between the predicted motion vector and the SR size, we develop an adaptive SR adjustment algorithm. We also design a RF selection scheme based on the correlation between the different block sizes of the macroblock. Experimental results show that our algorithm can significantly reduce the computational complexity of motion estimation compared with the JM15.1 reference software, with a negligible decrease in peak signal-to-noise ratio and a slight increase in bit rate. Our algorithm also outperforms existing methods in terms of its low complexity and high coding quality.

Accurate sensor spectral sensitivity (SSS) plays an important role in color reproduction, imaging, and computer vision. However, such data are not always available from manufacturers. We focus on recovering SSS from raw data, and propose a method to recover SSS. Using raw data, instead of sRGB data, to recover the SSS is due to the fact that the raw data are more closely related with actual sensor response values than sRGB data. With the raw data, the SSS can be recovered more accurately than with sRGB data that is widely used to recover SSS by almost all state-of-art recovery methods. Besides applying raw data in the recovery process, the other contribution is that we propose a new and simple cost function for the SSS recovery problem. This problem is then transformed into an optimization issue by minimizing the cost function. This minimization can be solved by a multiobject optimization method with a genetic algorithm. In contrast to the previous methods, our method is simple and has no parameter that needs to be determined by some prior knowledge. We validate our method through the experiments and comparison with the state-of-art methods under different illuminants.

Deconvolving Poissonian image has been a significant subject in various application areas such as astronomical, microscopic, and medical imaging. In this paper, a regularization-based approach is proposed to solve Poissonian image deconvolution by minimizing the regularization energy functional, which is composed of the generalized Kullback-Leibler divergence as the data-fidelity term and sparsity prior constraints as the regularization term, and a non-negativity constraint. We consider two sparsity prior constraints which include framelet-based analysis prior and combination of framelet and total variation analysis priors. Furthermore, we show that the resulting minimization problems can be efficiently solved by the split Bregman method. The comparative experimental results including quantitative and qualitative analysis manifest that our algorithm can effectively remove blur, suppress noise, and reduce artifacts.

In image processing applications, soft mathematical morphology (MM) can be employed for both binary and grayscale systems and is derived from set theory. Soft MM techniques have improved behavior over standard morphological operations in noisy environments, as they can preserve small details within an image. This makes them suitable for use in image processing applications on portable field programmable gate arrays for tasks such as robotics and security. We explain how the systems were developed using Altera’s DSP Builder in order to provide optimized code for the many different devices currently on the market. Also included is how the circuits can be inserted and combined with previously developed work in order to increase their functionality. The testing procedures involved loading different images into these systems and analyzing the outputs against matlab-generated validation images. A set of soft morphological operations are described, which can then be applied to various tasks and easily modified in size via altering the line buffer settings inside the system to accommodate a range of image attributes ranging from image sizes such as 320×240  pixels for basic webcam imagery up to high quality 4000×4000  pixel images for military applications.

This article [J. Electron. Imaging.. 22, (2 ), 023012 (2013)] was originally published online on 10 May 2013 with an error on page 4. In several instances, the symbol H was omitted from the following passage: