Search

Senior Editor Daniel LeMaster explains the nuances of OE's signal- and image-processing paper policy. [ABSTRACT FROM AUTHOR]

The traditional fractional order total variational model has better results in denoising and maintaining texture details in infrared images. However, it is difficult to determine the order of fractional order differentiation in image processing so that the model has the best denoising effect. To solve this problem, a fractional order total variational infrared image denoising model incorporating a flower pollination particle swarm optimization (PSO) algorithm is proposed in this paper. The model combines the search advantages of the flower pollination optimization algorithm and the PSO algorithm. The maximization multiobjective equation is designed as the fitness function of the optimization algorithm. The optimal order of the fractional order total variational model is found adaptively according to different features in different regions of the infrared image. The experimental results show that the improved model not only achieves the adaptivity of the adaption of the fractional order of total variational model order but also effectively removes the noise and retains the texture structure of infrared images to the maximum extent. [ABSTRACT FROM AUTHOR]

This article presents a comprehensive analysis of the dispersion characteristics of virtually imaged phased array (VIPA) based on the principle of paraxial dispersion. It proposes a calculation equation for the Brillouin shift specifically designed for a single-stage VIPA-based Brillouin scattering measurement system. To capture Brillouin scattering images of water and anhydrous glycerol, a Brillouin scattering measurement system is constructed. The Brillouin shift of both substances is calculated using the proposed method in this article as well as the methods proposed by Giuseppe Antonacci and Pei-Jun Wu. The comparison of the three calculation results with the theoretical values confirms the accuracy and reliability of the proposed method. Moreover, this study addresses practical challenges in Brillouin image processing and introduces a precise method for extracting Brillouin spectra from the images. Additionally, a spectral calibration program is developed using MATLAB, guaranteeing accurate spectral extraction while greatly streamlining the spectral processing workflow. [ABSTRACT FROM AUTHOR]

This paper presents an effective target binarization method for a linear timed address-event (TAE) vision system. In the preprocessing phase, TAE data are processed by denoising, thinning, and edge connection methods sequentially to obtain the denoised- and clear-event contours. Then, the object region will be confirmed by an event-pair matching method. Finally, the image open and close operations of morphology methods are introduced to remove the artifacts generated by event-pair mismatching. Several degraded images were processed by our method and some traditional binarization methods, and the experimental results are provided. As compared with other methods, the proposed method performs efficiently on extracting the target region and gets satisfactory binarization results from object images with low-contrast and nonuniform illumination. [ABSTRACT FROM AUTHOR]

This paper presents a simple, efficient, yet robust approach, named joint orthogonal combination of local ternary pattern, for automatic forward-looking infrared target recognition. It gives more advantages to describe the macroscopic textures and microscopic textures by fusing variety of scales than the traditional LBP-based methods. In addition, it can effectively reduce the feature dimensionality. Further, the rotation invariant and uniform scheme, the robust LTP, and soft concave-convex partition are introduced to enhance its discriminative power. Experimental results demonstrate that the proposed method can achieve competitive results compared with the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Multicamera arrays are increasingly employed in both consumer and industrial applications, and various passive techniques are documented to estimate depth from such camera arrays. Current depth estimation methods provide useful estimations of depth in an imaged scene but are often impractical due to significant computational requirements. This paper presents a framework that generates a high-quality continuous depth map from multicamera array/light-field cameras. The proposed framework utilizes analysis of the local epipolar plane image to initiate the depth estimation process. The estimated depth map is then refined using total variation minimization based on the Fenchel–Rockafellar duality. Evaluation of this method based on a well-known benchmark indicates that the proposed framework performs well in terms of accuracy when compared with the top-ranked depth estimation methods and a baseline algorithm. The test dataset includes both photorealistic and nonphotorealistic scenes. Notably, the computational requirements required to achieve an equivalent accuracy are significantly reduced when compared with the top algorithms. As a consequence, the proposed framework is suitable for deployment in consumer and industrial applications. [ABSTRACT FROM AUTHOR]

An algorithm based on differential evolution (DE) with adaptive strategy selection for remote sensing image registration is designed. The paper has two major contributions. One contribution is the reduction of the computation time for a better result, i.e., fitness function approximate, which is based on multi-resolution processing and histogram matching. The second contribution is an adaptive strategy known as probability matching integration. The probability matching technique is employed in DE to autonomously select the most suitable strategy while solving the problem. To the best of our knowledge, the proposed method is the first adaptive strategy selection-based approach for image registration. In our experiments, the proposed method significantly outperforms previously proposed image registration method based on DE. [ABSTRACT FROM AUTHOR]

The focusing evaluation method based on wavelet transform and adaptive genetic algorithm (AGA) is developed in this paper. First, db7 wavelet transform is performed and the corresponding images are evaluated. Afterward, the optimal evaluation region is searched automatically in 2D wavelet coefficients by AGA. Finally, the criteria function based on the weighted MWT2 operator is used to evaluate the focusing condition. The wavelet base, order number of vanishing moments, and wavelet decomposition level are the key factors in constructing the operator. The parameters of new focus measure operator are analyzed and verified through experimental methods. The comparative experiments with other methods show that the AGA method not only could find the approximate optimal evaluation window quickly and accurately, but it has better generality as well. The weighted evaluation function of the MWT2 operator based on the db7 wavelet has higher sensitivity and better real-time performance. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of finding small and low-contrast moving targets in infrared (IR) video sequences produced by sensors with inconsistent parameters, such as intensity offset and gain as well as bad pixels. This sensor variability makes it difficult to apply methods based on frame registration using simple pixel differences. Our proposed algorithm uses regression to normalize the variations of intensity offset and gain between compared registered frames. A statistical criterion is used to calculate the threshold for the difference between normalized intensities of two frames. The algorithm for finding the differences between frames is also used to create a bad pixel mask either on- or offline. This mask is essential for the reduction of false detection rates. Our experiments show that this approach produces good results and can be used for detection of small, low-contrast targets in high dynamic range IR data. The proposed algorithm also produces good results for detecting moving targets in cases when objects are occluded by sparse vegetation. [ABSTRACT FROM AUTHOR]

The main aim of this paper is to study image enhancement by using sparse and redundant representations of the reflectance component in the Retinex model over a learned dictionary. This approach is different from existing variational methods, and the advantage of this approach is that the reflectance component in the Retinex model can be represented with more details by the dictionary. A variational method based on the dynamic dictionaries is adopted here, where it changes with respect to iterations of the enhancement algorithm. Numerical examples are also reported to demonstrate that the proposed methods can provide better visual quality of the enhanced high-contrast images than the other variational methods, i.e., revealing more details in the low-light part. [ABSTRACT FROM AUTHOR]

This paper presents a pose estimation method for planar mirror based on a one-dimensional (1-D) target. First, a 1-D target is maneuvered in front of a camera so as to allow the camera to capture several images of both the original and virtual feature points on the target. Next, the normal vector of the mirror and the distance between the mirror and the camera are calculated by a closed-form method. Last, a nonlinear optimization is used to refine the closed-form solution by minimizing the reprojection error of reflection projection. Synthetic data and real experiments show that the proposed method is accurate and reliable. The proposed method is simple and flexible. No prior information about the target motion and camera pose is assumed. A minimum of two nonparallel target positions are required. It is suitable for the pose estimation of planar mirrors, especially when they are in fixed positions. We can find the posture of the fixed camera according to the normal vector of the plane mirror. [ABSTRACT FROM AUTHOR]

Spot estimation accuracy of Shack-Hartmann images and its impact on Airborne Aero-Optic Laboratory (AAOL) wavefront statistics are addressed. A study is conducted of an individual spot simulated using a double sinc function under varying degrees of additive nonzero mean Gaussian noise within a 15 x 15 pixel area-of-interest. The focus of this paper is two-fold. First, the accuracy of four existing centroiding methods including first moment, convolution, Gaussian, and weighted first moment are compared. It is found that the weighted first moment centroid most accurately estimates spot centers but requires significantly more computational time with respect to the first moment method. Second, three image-processing techniques, including gamma correction, thresholding, and windowing, are analyzed to determine their influence on each centroiding method's spot estimation accuracy. A fourth order gamma correction significantly reduces spot estimation error for three centroiding methods. The key result is that the accuracy of the first moment centroid with an applied gamma correction is comparable to the weighted first moment without the computational burden. Finally, the first moment centroid with gamma correction and weighted first moment centroid are applied to AAOL flight data. Wavefront statistics are computed and compared to the commonly used first moment centroid. [ABSTRACT FROM AUTHOR]

Fast and precise iris localization is a vital technique for face recognition, eye tracking, and gaze estimation. Low-resolution images bring about great difficulties for locating the iris precisely by traditional methods. In this paper, a fast and robust method to precisely detect the position and contour of the irises in low-resolution facial images is presented. A three-step coarse-to-fine strategy is employed. First, a gradient integral projection function is proposed to roughly detect the eye region, and the vertical integral projection function is adopted to select several possible vertical boundaries of the irises. Second, we have proposed a novel rectangular integro-variance operator to precisely locate both of the irises. Finally, the localization results are verified by two simple heuristic rules. A novel and more rigorous criterion is also proposed to evaluate the performance of the algorithm. Comparison experiments on images from the FERET and the Extended YaleB databases demonstrate that our method is more robust than traditional methods to scale variation, illumination changes, part occlusion, and limited changes of head poses in low-resolution facial images. [ABSTRACT FROM AUTHOR]

This paper describes a new anisotropic diffusion method for removing ringing artifacts from deconvolution images. The characteristics of the deconvolution image ringing artifact are also analyzed. Unlike the previous anisotropic diffusion methods that use the global statistical value as the scale function, the proposed anisotropic diffusion defines a new scale function based on image local statistical characteristics to recognize the ringing artifacts. The proposed anisotropic diffusion performs well in terms of the images' detail preservation, error reduction, and improvement in visual quality in the presence of various levels of image ringing artifacts. [ABSTRACT FROM AUTHOR]

In this paper, we present a universal deblurring method for real images without prior knowledge of the blur source. The proposed method uses the transition region of the blurred image to estimate the point spread function (PSF). It determines the main edges of the blurred image with high edge measures based on the difference of Gaussians (DoG) operator. Those edge measures are used to predict the transition region of the sharp image. By using the transition region, we select the pixels of the blurred image to form a series of equations for calculating the PSF. In order to overcome noise disturbance, the optimal method based on the anisotropic adaptive regularization is used to estimate the PSF, in which the constraints of non-negative and spatial correlations are incorporated. Once the PSF is estimated, the blurred image is effectively recovered by employing nonblind restoration. Experimental results show that the proposed method performs effectively for real images with different blur sources. [ABSTRACT FROM AUTHOR]

This paper proposes a novel local color correction algorithm that uses three dimensional (3D) point set registration especially for underwater stereo images. Due to the limited visibility in an underwater environment, underwater stereo images are usually captured by a half-mirror-based stereo camera that helps to prevent the unwanted visual fatigues induced by excessive disparities. However, the interaction between the half mirror and the scattered light components in the water may produce strong local color discrepancies. Thus, the proposed algorithm extracts sufficient information on the entire image area between the left and right images using 3D point set registration to counteract these effects. Since the proposed algorithm simultaneously processes all three color channels in the local color space, it is robust to the strong local color discrepancies in underwater stereo images. The proposed algorithm is both subjectively and objectively evaluated by comparing the histograms, the color similarities, and the disparity images before and after color correction. After color correction, the proposed algorithm achieves a color similarity up to 96.67%, while other conventional algorithms show color similarities only up to 90.66%. Although the proposed algorithm is designed for underwater stereo images, it can be used for various stereo images and applications. [ABSTRACT FROM AUTHOR]

Skin detection plays an important role in many applications, including face detection, human motion analysis, and objectionable image filtering. We propose a novel skin detection approach named multiple Gaussian models (MGMs). This approach combines multiple single Gaussian models and determines each model in order to maximize the true positive rate (TPR) of skin detection subject to a fixed predefined false positive rate (FPR). We derive the discrete and continuous forms of MGM approaches in the paper. The proposed approach has almost optimal performance for a broad range of FPRs in the Gaussian framework. Moreover, it has low computational costs in skin detection for new image instances. Experimental results show that the MGM approach has better skin detection performance than previous methods within the Gaussian framework. [ABSTRACT FROM AUTHOR]

Most conventional fingerprint verifications use minutiae information as features. However, some fingerprints have many scratches that can make minutiae extraction quite difficult to capture with enough accuracy. This paper presents a new fingerprint verification system that can cover normal and scratched fingerprints and can be implemented on personal portable devices such as smartphones. The system separates those fingerprints with serious scratches and extracts the scratches as an additional matching feature, thereby expanding the number and type of fingerprints accepted by the application. A simplified gabor filtering technique enables embedded systems to provide real-time verification for scratched fingerprints. In an evaluation of a fingerprint test database, in which 25% of the fingerprints have scratches, the new verification system reduced the zero false acceptance rate (FARzero) by approximately 71% compared with the simple minutiae method. Furthermore, the new system expanded coverage and increased performance. [ABSTRACT FROM AUTHOR]

To extract circular array targets from high-spatial resolution optical satellite images, we propose an efficient two-step detection framework. In the first stage, we propose a multiscale histogram contrast (MHC) saliency evaluation model to detect salient objects. This saliency model is able to highlight all salient areas with well-defined boundaries and effectively suppress the obvious background textures. Moreover, the generated saliency map can be used to find the locations of potential targets most likely causing visual attention, which greatly decreases the computation cost in later processes. In the second stage, we extract task-related targets from these salient objects by removing the irrelevant salient areas according to the special properties of the targets. In this paper, we present a circular array targets-detection method based on the saliency map produced by the proposed saliency detector. First we extract the circular objects from the saliency map using simple shape features, which are then clustered through a graph-based approach to locate the target areas. To evaluate the performance of our saliency model and targets extraction scheme, we carried out many experiments on large-field, high-spatial resolution remote-sensing images and achieved promising results both in efficiency and precision. [ABSTRACT FROM AUTHOR]

Removing shadows from single color images is an important problem in computer vision. In this paper, we propose a novel shadow removal approach, which could effectively remove shadows from textured surfaces, yielding high quality shadow-free images. Our approach aims at calculating scale factors to cancel the effect of shadows. Based on the regional gray edge hypothesis, which assumes the average of the reflectance differences in a region is achromatic, the scale factors can be computed without the restrictions that former algorithms need. The experimental results show that the proposed algorithm is effective. [ABSTRACT FROM AUTHOR]

As novel sensors, rotational symmetric triangulation sensors (RSTSs) have broad application potential because of their robustness and accuracy. However, existing RSTSs have the disadvantages of complicated structure and high cost. The current paper presents a novel RSTS based on an object space mirror. The proposed sensor uses a cone mirror and a commercial camera, which significantly simplifies the structure and reduces its cost. Mathematical models are discussed to reveal the measurement range of the sensor and the relation between the displacement of the spot and the radius of the ring in the image. An improved image processing method is also introduced for a more efficient processing. Finally, two experimental systems with mirror angles at π/6 and π/4 are set up. Experimental results showed that the proposed sensor can achieve a comparable accuracy of 1.6 μm compared with that of existing methods, as well as a higher efficiency of image processing, while remaining at a low cost. [ABSTRACT FROM AUTHOR]

Due to light being absorbed by water and scattered by suspended particles, color distortion, low image contrast, and detail loss are the common problems found in underwater images. We present an underwater image processing method, which based on an underwater imaging model, makes initial image restoration with the improved red channel prior combining transmission reduction, background light correction, and color compensation to acquire clearer underwater images, and then combines multiscale retinex (MSR) enhancement and hue saturation value to enhance image color contrast and saturation. Results indicate that, compared with the original images, the method proposed in this paper is capable of significantly enhancing image definition, saturation, contrast, and luminance. In comparison with dark channel prior, single-scale retinex (SSR), MSR, and multiscale retinex with color restoration (MSRCR), this method greatly improved the average gradient of images while keeping the information entropy of images unchanged. The average gradient calculated with the proposed method increased by 239%, 199%, 200%, and 220% for the three given images, respectively, compared with the other methods above. [ABSTRACT FROM AUTHOR]

Due to the inaccurate and unreliable moving distance measurement of the hydraulic support in mines, a method based on the random circle detection (RCD) algorithm and the fruit fly optimization algorithm (FOA) is proposed. According to the changing center and radium of the circle on the support, the relative position of adjacent supports is acquired by the camera. The noise of the collected image is moved, and the edge feature is protected using a bilateral filter. A local adaptive threshold algorithm is used for binary processing of the image. Then, RCD is used to detect the contour, which is similar to the circle. A method to detect the circle based on FOA is used to accurately detect the circle. Subsequently, the relative distance is calculated according to the change of the circle. Finally, the accuracy and reliability of the proposed method are verified though the experiment. [ABSTRACT FROM AUTHOR]

Digital adaptive optics (DAO) created using homodyne encoding can mitigate atmospheric turbulence in passive imaging systems. This work demonstrates a self-referencing homodyne interferometry technique that combines the passive imaging utility of multiframe algorithmic procedures with the single-frame correction capability of the Shack–Hartmann adaptive optics technique. This paper presents image reconstruction improvements through the addition of (1) phase diversity modulation techniques within the interferometry reconstruction algorithm and (2) temporal image processing techniques applied after the interferometry reconstruction algorithm. By imaging quick response codes through a turbulent air chamber in the laboratory, it was possible to quantify the machine-readable performance gain provided by DAO when compared with a standard imaging camera. Results from this research verify that DAO from homodyne encoding provides turbulence mitigation for single frames of data, paving the way for environmentally robust, high-speed, self-contained imaging systems. [ABSTRACT FROM AUTHOR]

This paper describes a Fourier propagator for computing the impulse response of an optical system, while including terms ignored in Fresnel and Fraunhofer calculations. The propagator includes a Rayleigh-Sommerfeld diffraction formula calculation from a distant point through the optical system to its image point predicted by geometric optics. The propagator then approximates the neighboring field points via the traditional binomial approximation of the Taylor series expansion around that field point. This technique results in a propagator that combines the speed of a Fourier transform operation with the accuracy of the Rayleigh-Sommerfeld diffraction formula calculation and extends Fourier optics to cases that are nonparaxial. The proposed propagator facilitates direct calculation of aberration coefficients, making it more versatile than the angular spectrum propagator. Bounds on the phase error introduced by the approximations are derived, which show that it should be more widely applicable than the Fresnel propagator. Guidance on how to sample the pupil and detector planes of a simulated imaging system is provided. This report concludes by showing examples of diffraction calculations for a laboratory setup and comparing them to measured diffraction patterns to demonstrate the utility of the propagator. [ABSTRACT FROM AUTHOR]

Direct signal injection (DSI) simulation has been widely used in the performance testing and evaluation of an infrared imaging system (IRIS). A method to generate a high realism infrared sensor signal (HRIRSS) for DSI simulation is proposed. The IR sensor signal generation system hardware is based on a computer and field-programmable gate array processor. Mathematical models are adopted to quantitatively characterize the imaging physical effects (IPES) of the IRIS. Spatial template convolution and pixel processing approaches are used to realize high-speed real-time computation of the IPES. Taking advantages from low-noise circuit design methodologies, we designed a low-noise digital–analog converter to convert digital images generated by the system to analog signals for DSI simulation. The noise voltage of the converter within the passband is less than 244 μV, which ensures the system a 14-bit precision. For simulation of IR sensors with a resolution of 320 × 256 pixels, the processing speed is up to 240 frames per second (fps) and that for the 128 × 128 pixel sensor is 480 fps. The test results also show that the generated IR sensor signals have high confidence. In addition, the system has the advantages of simple structure, strong expansibility, low cost, and it has been used in practical applications. [ABSTRACT FROM AUTHOR]

The processing of structured light images in real time is a challenging task for the development of three-dimensional (3-D) shape measurement methods. This paper presents a high speed and low-cost optical profiler implemented using a projection method based on the use of a digital light-processing device to illuminate the object to be measured. The image processing of the reflected structured light pattern allows potential realtime capabilities. The proposed method of absolute phase retrieval for unwrapping the relative phase uses a single additional staircase intensity pattern to determine and correct 2π discontinuities in the phase. Good results are obtained when the method is compared to another which uses three additional fringe patterns to determine the stair phase and then the absolute phase. Since the proposed technique uses only one extra pattern instead of three, it is less costly in terms of computation complexity and is thus faster. The hardware of the developed fringe projection system for a 3-D macroscopic reconstruction is presented and the performance of the method is evaluated. Simulated and experimental results are presented and compared to the other absolute phase- retrieval method. The proposed method is suitable for measuring 3-D object surfaces for a possible implementation in real time. [ABSTRACT FROM AUTHOR]

Several photogrammetry-based methods have been proposed that the derive three-dimensional (3-D) information from digital images from different perspectives, and lidar-based methods have been proposed that merge lidar point clouds and texture the merged point clouds with digital imagery. Image registration alone has difficulty with smooth regions with low contrast, whereas point cloud merging alone has difficulty with outliers and a lack of proper convergence in the merging process. This paper presents a method to create 3-D images that uses the unique properties of texel images (pixel-fused lidar and digital imagery) to improve the quality and robustness of fused 3-D images. The proposed method uses both image processing and point-cloud merging to combine texel images in an iterative technique. Since the digital image pixels and the lidar 3-D points are fused at the sensor level, more accurate 3-D images are generated because registration of image data automatically improves the merging of the point clouds, and vice versa. Examples illustrate the value of this method over other methods. The proposed method also includes modifications for the situation where an estimate of position and attitude of the sensor is known, when obtained from low-cost global positioning systems and inertial measurement units sensors. [ABSTRACT FROM AUTHOR]

The problem of viewpoint variations is a challenging issue in vision-based human action recognition. With the richer information provided by three-dimensional (3-D) point clouds thanks to the advent of 3-D depth cameras, we can effectively analyze spatial variations in human actions. In this paper, we propose a volumetric spatial feature representation (VSfR) that measures the density of 3-D point clouds for view-invariant human action recognition from depth sequence images. Using VSFR, we construct a self-similarity matrix (SSM) that can graphically represent temporal variations in the depth sequence. To obtain an SSM, we compute the squared Euclidean distance of VSFRs between a pair of frames in a video sequence. In this manner, an SSM represents the dissimilarity between a pair of frames in terms of spatial information in a video sequence captured at an arbitrary viewpoint. Furthermore, due to the use of a bag-of-features method for feature representations, the proposed method efficiently handles the variations of action speed or length. Hence, our method is robust to both variations in viewpoints and lengths of action sequences. We evaluated the proposed method by comparing with state-of-the-art methods in the literature on three public datasets of ACT42, MSRAction3D, and MSRDailyActivity3D, validating the superiority of our method by achieving the highest accuracies. [ABSTRACT FROM AUTHOR]

Yarn hairiness has been an important indication of yarn quality that affects weaving production and fabric appearance. In addition to many dedicated instruments, various image analysis systems have been adopted to measure yarn hairiness for potential values of high accuracy and low cost. However, there is a common problemin acquiring yarn images; that is, hairy fibers protruding beyond the depth of field of the imaging system cannot be fully focused. Fuzzy fibers in the image inevitably introduce errors to the hairiness data. This paper presents a project that attempts to solve the off-focus problem of hairy fibers by applying a new imaging scheme--multifocus image fusion. This new scheme uses compensatory information in sequential images taken at the same position but different depths to construct a new image whose pixels have the highest sharpness among the sequential images. The fused image possesses clearer fiber edges, permitting more complete fiber segmentation and tracing. In the experiments, we used six yarns of different fiber contents and spinning methods to compare the hairiness measurements from the fused images with those from unfused images and from the Uster tester. [ABSTRACT FROM AUTHOR]

Underwater images often suffer from problems such as low contrast, color distortion, and blurred details, which have a negative impact on subsequent image processing tasks. To mitigate such problems, we propose an algorithm that combines an underwater physical imaging model with a convolutional neural network. The physical imaging model has two important types of parameters: background scattering parameters and direct transmission parameters. For the background scattering parameters, we divide them into three levels, which are primary information, secondary information, and advanced information, and design three subnetworks for feature extraction to represent them. For the direct transmission parameters, we decompose them into two levels, which are shallow transmission information and deep transmission information, and also design two subnetworks to indicate them. Experimental results show that, compared with other enhancement algorithms, the proposed algorithm can not only effectively correct the color deviation and enhance the object edges and texture details but also can obtain superior values of objective evaluation metrics in terms of peak signal-to-noise ratio, structural similarity, underwater color image quality evaluation, and underwater image quality measure. [ABSTRACT FROM AUTHOR]

Decoding is a challenging and complex problem in a coded structured light system. In this paper, the decoding problem is modeled as matching two floating number sequences. Fuzzy window matching instead of hard window matching is proposed to determine the correspondence. In the proposed decoding method, the stripe edges are extracted using color edge detection after color correction has been conducted. Then, the stripe colors are identified through the use of a fuzzy c-means clustering algorithm on a color feature, which is only related to the spectral sensitivity of red, green, and blue sensors and the albedo of the surface. Finally, a fuzzy multistep matching method is proposed to find the correspondence, which aims to match the subsequences with the length less than the window width. In the experiments, fuzzy window matching is compared with hard window matching. No matter what local matching method is adopted, fuzzy window matching can achieve higher accuracies than hard window matching. The obtained point clouds and surfaces are also larger. When the multistep idea is added, the performance is enhanced. [ABSTRACT FROM AUTHOR]

We propose a texture saliency classifier to detect people in a video frame by identifying salient texture regions. The image is classified into foreground and background in real time. No temporal image information is used during the classification. The system is used for the task of detecting people entering a sterile zone, which is a common scenario for visual surveillance. Testing is performed on the Imagery Library for Intelligent Detection Systems sterile zone benchmark dataset of the United Kingdom's Home Office. The basic classifier is extended by fusing its output with simple motion information, which significantly outperforms standard motion tracking. A lower detection time can be achieved by combining texture classification with Kalman filtering. The fusion approach running at 10 fps gives the highest result of F1 = 0.92 for the 24-h test dataset. The paper concludes with a detailed analysis of the computation time required for the different parts of the algorithm. [ABSTRACT FROM AUTHOR]

To tackle robust visual tracking in complex environment, an online algorithm based on generative model is proposed. The target is represented with overlapped and selected local patches based on key point proportion ranking, and its location is estimated by spatiotemporal analysis. Temporally, a propagated affine warping dynamical model is newly introduced. Spatially, an observation model based on weighted sparse representation and geometric confidence inference is newly established. Both selection pattern and templates are periodically updated to adapt the target's appearance variation. Experiments demonstrate that the proposed approach achieves more favorable performance compared with classical works on challenging image sequences. [ABSTRACT FROM AUTHOR]

Uneven illumination, vignetting, and stray light lead to the undesirable decrease of response uniformity of high-throughput fluorescence imaging systems, which brings difficulties to subsequent image processing and data analysis. Here, we propose a flat-field correction method using array targets specially designed for fluorescence microscopy. Based on several dedicated correction sources and the response model, the proposed algorithm performs step-by-step correction to eliminate nonuniformities caused by stray light, high-frequency inhomogeneity, and low-frequency response inhomogeneity. For verification, the correction method is applied to a high-throughput microscopic imaging system with several time-delay-integration detectors. Experimental results show that the proposed flat-field correction method is effective and practical. [ABSTRACT FROM AUTHOR]

In this paper, we propose a method for the numerical simulation of optical double random phase encoding (DRPE). This method is based on a discrete model of the DRPE system that retains the properties of its optical counterpart and takes into account the capacity of DRPE to spread a signal's energy in both the space and the spatial frequency domains. Discrete versions of two narrowband lossless diffusers are created to simulate their analogue optical counterparts and their bandwidths control how much the energy of the input signal is spread in both the space and spatial frequency domains as it passes through the system. The limiting case, when complete aliasing (maximum overlap) takes place, is also discussed in relation to our method. In this case, we show our method reduces to a purely numerical encryption procedure which can, nonetheless, be decrypted perfectly. Simulation results are presented to demonstrate the feasibility of the proposed technique. [ABSTRACT FROM AUTHOR]

When the object is similar to the background, it is difficult to segment the completed human body object from video images. To solve the problem, this paper proposes an object segmentation algorithm based on guided layering from video images. This algorithm adopts the structure of advance by degrees, including three parts altogether. Each part constructs the different energy function in terms of the spatiotemporal information to maximize the posterior probability of segmentation label. In part one, the energy functions are established, respectively, with the frame difference information in the first layer and second layer. By optimization, the initial segmentation is solved in the first layer, and then the amended segmentation is obtained in the second layer. In part two, the energy function is built in the interframe with the shape feature as the prior guiding to eliminate the interframe difference of the segmentation result. In art three, the segmentation results in the previous two parts are fused to suppress or inhibit the over-repairing segmentation and the object shape variations in the adjacent two-frame. The results from the compared experiment indicate that this algorithm can obtain the completed human body object in the case of the video image with similarity between object and background. [ABSTRACT FROM AUTHOR]

This paper proposes an efficient intra-field deinterlacing algorithm which implements a filter switching interpolation method (FSID) according to different local region features, which produces outstanding visual results. In this method, a bilateral filter is used for smooth regions to interpolate the missing line due to its accurate interpolation performance through use of the closeness and similarity among the neighboring pixels. On the other hand, a trilateral filter is used for complex regions to interpolate the missing line since it has high edge-preserving ability due to taking into account the closeness and similarity among neighboring pixels, but also the pixel gradient. Experimental results show that the proposed FSID algorithm provides satisfactory performances in terms of both objective and subjective image quality. In addition, the algorithm exploits the local spatial similarity among the neighboring pixels without complex preset-conditions, which is easier to implement than many other existing algorithms. [ABSTRACT FROM AUTHOR]

Studies have shown that stereo images improve surgeons' visuomotor tasks and therefore constructively affect the outcome of a minimally invasive surgery. Stereo images are captured by a stereo endoscope, which consists commonly of duplicate lens systems. However, stereo images can also be captured by a single lens system following a dual aperture scheme (DAS). DAS creates two spatially separated optical channels by placing a dual aperture plate in the limiting aperture of a single lens system. This paper describes efforts to miniaturize the DAS-based imaging system for use in minimally invasive surgery. To demonstrate feasibility, a prototype was fabricated using lens elements 3 mm in diameter and was tested for its stereo depth effect (SDE). The SDE of the prototype was then compared to a duplicate lens system that was constructed theoretically in the same diameter as the 3-mm prototype. The results show that the prototype yields 4/7 of the SDE of the theoretical model. However, the SDE of the prototype provides sufficient SDE, in a viewing range of 1 to 2.5 cm from the lens, for minimally invasive surgery. [ABSTRACT FROM AUTHOR]

The presence of turbulence over horizontal imaging paths severely reduces the resolution available to imaging systems and introduces anisoplanatic distortions in the image frame. A variety of image processing techniques are being developed which can mitigate these effects, and sequences of high-fidelity images created via simulation are useful in their development. In this paper, we describe a simplification of the split-step wave propagation method that employs a series of uniformly spaced phase screens to accurately simulate turbulence effects on imaging over horizontal paths. Employing this method, a series of 1000 image frames were generated for each of three turbulence conditions. The mean squared error in intensity per pixel is also evaluated for each frame in comparison to a diffraction-limited reference. Examination of the per-frame intensity error statistics indicate these errors are log-normally distributed about a mean value that increases approximately linearly with turbulence strength. [ABSTRACT FROM AUTHOR]

We address video data-hiding on the context-adaptive variable-length coding domain of the H.264/AVC standard. It is challenging to maintain a minimum variation of rate-distortion performance when hiding data in compressed videos by slightly modifying one or more transform coefficients of prediction residual data. The fluctuant variation of rate-distortion performance is especially serious in the case of hiding data in H.264/AVC compressed videos due to the thorough use of various prediction modes. Most current video data-hiding schemes concentrate on hiding messages in video intra (I)-frames to mitigate performance degradation at the expense of limited hiding capacity and applicability. Unlike most previous video data-hiding schemes, this paper considers data-hiding schemes on both I- and predicted (P)-frames of videos. We first investigate the possible ways used to suppress the significant degradation of the rate-distortion performance for hiding data in H.264/AVC videos. Based on the investigation, a hybrid video data-hiding scheme is designed to perform the classification of 4 x 4 residual blocks according to the characteristics of each block and to hide message data differently in each classified block in order to reduce the performance degradation caused by the data-hiding process. The block classification is based on the magnitude of the last non-zero coefficient and the coefficient difference energy of each block to select suitable blocks that are capable of contributing less degradation of coding performance for hiding additional messages. Moreover, the proposed data-hiding algorithm is incorporated with the rate-distortion optimization loop of the H.264/AVC encoder to provide further degradation suppression on the output bit-rate and decoded quality. From experiments on several well-known test videos, it can be seen that the proposed data-hiding scheme demonstrates improved efficiency on video quality and output bit-rate compared to other similar methods [ABSTRACT FROM AUTHOR]

Video foreground object detection faces the problems of moving backgrounds, illumination changes, chaotic motion in real word applications. This paper presents a hybrid pixel-based background (HPB) model, which is constructed by single stable record and multi-layer astable records after initial learning. This HPB model can be used for background subtraction to extract objects precisely in various complex scenes. Using the multi-layer astable records, we also propose a homogeneous background subtraction that can detect the foreground object with less memory load. Based on the benchmark videos, the experimental results show that single stable and 3-layer multi-layer astable records can be enough for background model construction and are updated quickly to overcome the background variation. The proposed approach can improve the average error rates of foreground object detection up to 86% when comparing with the latest works. Furthermore, our method can achieve real-time analysis for complex scenes on personal computers and embedded platforms. [ABSTRACT FROM AUTHOR]

Clustering is one of the image-processing methods used in nondestructive testing (NDT). As one of the initializing parameters, most clustering algorithms, like fuzzy C means (FCM), Iterative self-organization data analysis (ISODATA), K-means, and their derivatives, require the number of clusters. This paper proposes an algorithm for clustering the pixels in C-scan images without any initializing parameters. In this state-of-the-art method, an image is sampled based on the rosette pattern and according to the pattern characteristics, and extracted samples are clustered and then the number of clusters is determined. The centroids of the classes are computed by means of a method used to calculate the distribution function. Based on different data sets, the results show that the algorithm improves the clustering capability by 92.93% and 91.93% in comparison with FCM and K-means algorithms, respectively. Moreover, when dealing with high-resolution data sets, the efficiency of the algorithm in terms of cluster detection and run time improves considerably. [ABSTRACT FROM AUTHOR]

Multiclass geospatial objects detection within complex environments is a challenging problem in remote-sensing areas. In this paper we propose a novel, generalized kernel-based learning framework for the purpose of enhanced object detection. There are two novel areas. (1) Multisource information, including shape, feature points, and appearance, was extracted to give a comprehensive representation of the objects. We improved a shape descriptor and introduced a two-level spatial pyramid to represent appearance, both global and local. Therefore, basis kernels were formed, one for each feature. (2) In order to illustrate the effect of each kind of feature on each pyramid level, a generalized and weighted combination method was first used to combine all of the levels and then the features. The weights and the classifier model are based on the support vector machine framework for obtaining balance between all basis kernels. This classifier was transformed into a powerful detector by using a sliding window. The reported results are for the detection on highresolution remote-sensing images. This study demonstrates that the proposed generalized and weighted combination of kernels can yield better performance compared with traditional single-kernel classifier and other combination methods. [ABSTRACT FROM AUTHOR]

A practical omnidirectional vision sensor, consisting of a curved mirror, a mirror-supporting structure, and a megapixel digital imaging system, can view a field of 360 deg horizontally and 135 deg vertically. The authors theoretically analyzed and evaluated several curved mirrors, namely, a spherical mirror, an equidistant mirror, and a single viewpoint mirror (hyperboloidal mirror). The focus of their study was mainly on the image-forming characteristics, position of the virtual images, and size of blur spot images. The authors propose here a practical design method that satisfies the required characteristics. They developed image-processing software for converting circular images to images of the desired characteristics in real time. They also developed several prototype vision sensors using spherical mirrors. Reports dealing with virtual images and blur-spot size of curved mirrors are few; therefore, this paper will be very useful for the development of omnidirectional vision sensors. [ABSTRACT FROM AUTHOR]

Dynamic range reduction and detail enhancement are two important issues for effectively displaying high-dynamic-range images acquired by thermal camera systems. They must be performed in such a way that the high dynamic range image signal output from sensors is compressed in a pleasing manner for display on lower dynamic range monitors without reducing the perceptibility of small details. In this paper, a new method of display and detail enhancement for high dynamic range infrared images is presented. This method effectively maps the raw acquired infrared image to 8-bit domain based on the same architecture of bilateral filter and dynamic range partitioning approach. It includes three main steps: First, a bilateral filter is applied to separate the input image into the base component and detail component. Second, refine the base and detail layer using an adaptive Gaussian filter to avoid unwanted artifacts. Then the base layer is projected to the display range and the detail layer is enhanced using an adaptive gain control approach. Finally, the two parts are recombined and quantized to 8-bit domain. The strength of the proposed method lies in its ability to avoid unwanted artifacts and adaptability in different scenarios. Its great performance is validated by the experimental results tested with two real infrared imagers. [ABSTRACT FROM AUTHOR]

In this paper, we present an approach to learn latent semantic analysis models from loosely annotated images for automatic image annotation and indexing. The given annotation in training images is loose due to: 1. ambiguous correspondences between visual features and annotated keywords; 2. incomplete lists of annotated keywords. The second reason motivates us to enrich the incomplete annotation in a simple way before learning a topic model. In particular, some "imagined" keywords are poured into the incomplete annotation through measuring similarity between keywords in terms of their co-occurrence. Then, both given and imagined annotations are employed to learn probabilistic topic models for automatically annotating new images. We conduct experiments on two image databases (i.e., Corel and ESP) coupled with their loose annotations, and compare the proposed method with state-of-the-art discrete annotation methods. The proposed method improves word-driven probability latent semantic analysis (PLSA-words) up to a comparable performance with the best discrete annotation method, while a merit of PLSA-words is still kept, i.e., a wider semantic range. [ABSTRACT FROM AUTHOR]