This PDF file contains the front matter associated with SPIE Proceedings Volume 10020, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

Full-waveform LiDAR is a method that digitizes the complete waveform of backscattered pulses to obtain range information of multi-targets. To avoid expensive sensors in conventional full-waveform LiDAR system, a new system based on compressive sensing method is presented in this paper. The non-coherent continuous-wave laser is modulated by electro-optical modulator with pseudo-random sequences. A low-bandwidth detector and a low-bandwidth analog-digital converter are used to acquire the returned signal. OMP algorithm is employed to reconstruct the high resolution range information.

Ghost imaging (GI) is an important technique in single-pixel imaging. It has been demonstrated that GI has applications in various areas such as imaging through harsh environments and optical encryption. Correlation is widely used to reconstruct the object image in GI. But it only offers the signal-to-noise ratios (SNR) of the reconstructed image linearly proportional to the number of measurements. Here, we develop a kind of iterative deconvolution methods for GI. With the known image transmission matrix in GI, the first one uses an iterative algorithm to decrease the error between the reconstructed image and the ground-truth image. Ideally, the error converges to a minimum for speckle patterns when the number of measurements is larger than the number of resolution cells. The second technique, Gerchberg-Saxton (GS) like GI, takes the advantage of the integral property of the Fourier transform, and treats the captured data as constraints for image reconstruction. According to this property, we can regard the data recorded by the bucket detector as the Fourier transform of the object image evaluated at the origin. Each of the speckle patterns randomly selects certain spectral components of the object and shift them to the origin in the Fourier space. One can use these constraints to reconstruct the image with the GS algorithm. This deconvolution method is suitable for any single pixel imaging models. Compared to conventional GI, both techniques offer a nonlinear growth of the SNR value with respect to the number of measurements.

A high-resolution and large field-of-view micro-CT system is indispensable for the visualization of fine threedimensional (3-D) structures of a large specimen. Such a system drastically increases the overall number of effective sensor pixels. At SPring-8 over a decade ago, a micro-CT system based on a 10M-pixel CCD camera was developed for 3-D specimen imaging of centimeter-sized objects with approximately 7 μm spatial resolution. Subsequently, more recent studies have required systems with higher spatial resolution and a wider field-of-view. Detectors with spatial resolution of around 5 μm can visualize capillaries. However, such detectors make it extremely expensive to develop a new x-ray detector with several tens of megapixels in a conventional manner. Fortunately, dizzying advances in image sensor technology for consumer appliances have enabled the development of x-ray detectors with spatial resolution of around 5 μm using a commercial digital single-lens reflex camera fitted with a 36M-pixel CMOS image sensor for the visualization of fine 3-D structures of large human lung specimens. This paper describes a comparison of the performance offered by the new 36M-pixel micro-CT system and the 10M-pixel system.

Underwater 3D range-gated imaging can extend the detection range over underwater stereo cameras, and also has great potentials in real-time high-resolution imaging than 3D laser scanning. In this paper, a triangular-range-intensity profile spatial correlation method is used for underwater 3D range-gated imaging. Different from the traditional trapezoidal method, in our method gate images have triangular range-intensity profiles. Furthermore, inter-frame correlation is used for video-rate 3D imaging. In addition, multi-pulse time delay integration is introduced to shape range-intensify profiles and realize flexible 3D SRGI. Finally, in experiments, 3D images of fish net, seaweed and balls are obtained with mm-scaled spatial and range resolution.

Light field cameras have been rapidly developed and are likely to appear in mobile devices in near future. It is essential to develop efficient and robust depth estimation algorithm for mobile applications. However, existing methods are either slow or lack of adaptability to occlusion such that they are not suitable to mobile computing platform. In this paper, we present the generalized EPI representation in light field and formulate it using two linear functions. By combining it with the light field occlusion theory, a highly efficient and anti-occlusion depth estimation algorithm is proposed. Our algorithm outperforms the previous local method, especially in occlusion areas. Experimental results on public light field datasets have demonstrated the effectiveness and efficiency of the proposed algorithm.

Conventional multispectral imaging methods detect photons of a 3D hyperspectral data cube separately either in the spatial or spectral dimension using array detectors, and are thus photon inefficient and spectrum range limited. Besides, they are usually bulky and highly expensive. To address these issues, this paper presents single-pixel multispectral imaging techniques, which are of high sensitivity, wide spectrum range, low cost and light weight. Two mechanisms are proposed, and experimental validation are also reported.

Fourier ptychographic microscopy (FPM) is a new computational super-resolution approach, which can obtain not only the correct object function, but also the pupil aberration, the LED misalignment, and beyond. Although many state-mixed FPM techniques have been proposed to achieve higher data acquisition efficiency and recovery accuracy in the past few years, little is known that their reconstruction performance highly depends on the data redundancy in both object and frequency domains. In this paper, we explore the spatial and spectrum data redundancy requirements for the FPM recovery process to introduce sampling criteria for the conventional and state-mixed FPM techniques in both object and frequency space.

Spectral reflectance provides the most fundamental information of objects and is recognized as the “fingerprint” of them, since reflectance is independent of illumination and viewing conditions. However, reconstructing high-dimensional spectral reflectance from relatively low-dimensional camera outputs is an illposed problem and most of methods requaired camera’s spectral responsivity. We propose a method to reconstruct spectral reflectance from digital camera outputs without prior knowledge of camera’s spectral responsivity. This method respectively averages reflectances of selected subset from main training samples by prescribing a limit to tolerable color difference between the training samples and the camera outputs. Different tolerable color differences of training samples were investigated with Munsell chips under D65 light source. Experimental results show that the proposed method outperforms classic PI method in terms of multiple evaluation criteria between the actual and the reconstructed reflectances. Besides, the reconstructed spectral reflectances are between 0-1, which make them have actual physical meanings and better than traditional methods.

The multi-view light fields (MVLF) provide new solutions to the existing problems in monocular light field, such as the limited field of view. However as key steps in MVLF, the calibration and registration have been limited studied. In this paper, we propose a method to calibrate the camera and register different LFs without the checkboard at the same time, which we call the self-calibrating method. We model the LF structure as a 5-parameter two-parallel-plane (2PP) model, then represent the associations between rays and reconstructed points as a 3D projective transformation. With the constraints of ray-ray correspondences in different LFs, the parameters can be solved with a linear initialization and a nonlinear refinement. The result in real scene and 3D point clouds registration error of MVLF in simulated data verify the high performance of the proposed model.

The Lenselet-Based Plenoptic has recently drawn a lot of attention in the field of computational photography. The additional information inherent in light field allows a wide range of applications, but some preliminary processing of the raw image is necessary before further operations. In this paper, an effective method is presented for the rotation rectification of the raw image. The rotation is caused by imperfectly position of micro-lens array relative to the sensor plane in commercially available Lytro plenoptic cameras. The key to our method is locating the center of each microlens image, which is projected by a micro-lens. Because of vignetting, the pixel values at centers of the micro-lens image are higher than those at the peripheries. A mask is applied to probe the micro-lens image to locate the center area by finding the local maximum response. The error of the center coordinate estimate is corrected and the angle of rotation is computed via a subsequent line fitting. The algorithm is performed on two images captured by different Lytro cameras. The angles of rotation are -0.3600° and -0.0621° respectively and the rectified raw image is useful and reliable for further operations, such as extraction of the sub-aperture images. The experimental results demonstrate that our method is efficient and accurate.

With the advantages of high resolution, large field of view and compacted size, optoelectronic imaging sensors are widely used in many fields, such as robot’s navigation, industrial measurement and remote sensing. Many researchers pay more attention to improve the comprehensive performances of imaging sensors, including large field of view (FOV), high resolution, compact size and high imaging efficiency, etc. One challenge is the tradeoff between high resolution and large field of view simultaneously considering compacted size. In this paper, we propose an optoelectronic imaging system combining the lenses of short focal length and long focal length based on dual CMOS to simulate the characters of human eyes which observe object within large FOV in high resolution. We design and optimize the two lens, the lens of short focal length is used to search object in a wide field and the long one is responsible for high resolution imaging of the target area. Based on a micro-CMOS imaging sensor with low voltage differential transmission technology-MIPI (Mobile Industry Processor Interface), we design the corresponding circuits to realize collecting optical information with high speed. The advantage of the interface is to help decreasing power consumption, improving transmission efficiency and achieving compacted size of imaging sensor. Meanwhile, we carried out simulations and experiments to testify the optoelectronic imaging system. The results show that the proposed method is helpful to improve the comprehensive performances of optoelectronic imaging sensors.

This paper aims to achieve robust behavior recognition of video object in complicated background. Features of the video object are described and modeled according to the depth information of three-dimensional video. Multi-dimensional eigen vector are constructed and used to process high-dimensional data. Stable object tracing in complex scenes can be achieved with multi-feature based behavior analysis, so as to obtain the motion trail. Subsequently, effective behavior recognition of video object is obtained according to the decision criteria. What’s more, the real-time of algorithms and accuracy of analysis are both improved greatly. The theory and method on the behavior analysis of video object in reality scenes put forward by this project have broad application prospect and important practical significance in the security, terrorism, military and many other fields.

One of the challenges of augmented reality is a seamless combination of objects of the real and virtual worlds, for example light sources. We suggest a measurement and computation models for reconstruction of light source position. The model is based on the dependence of luminance of the small size diffuse surface directly illuminated by point like source placed at a short distance from the observer or camera. The advantage of the computational model is the ability to eliminate the effects of indirect illumination. The paper presents a number of examples to illustrate the efficiency and accuracy of the proposed method.

Seed size, interior abnormal and damage of the tomato seeds will affect the germination. The purpose of this paper was to study the relationship between the internal morphology, seed size and seed germination of tomato. The preprocessing algorithm of X-ray image of tomato seeds was studied, and the internal structure characteristics of tomato seeds were extracted by image processing algorithm. By developing the image processing software, the cavity area between embryo and endosperm and the whole seed zone were determined. According to the difference of area of embryo and endosperm and Internal structural condition, seeds were divided into six categories, Respectively for three kinds of tomato seed germination test, the relationship between seed vigor and seed size , internal free cavity was explored through germination experiment. Through seedling evaluation test found that X-ray image analysis provide a perfect view of the inside part of the seed and seed morphology research methods. The larger the area of the endosperm and the embryo, the greater the probability of healthy seedlings sprout from the same size seeds. Mechanical damage adversely effects on seed germination, deterioration of tissue prone to produce week seedlings and abnormal seedlings.

Currently we are developing a 10 cm silicon carbide (SiC) deformable mirror with 37 actuators operating at 500 Hz. The deformable mirror will be applied in a 1.5 m telescope. An adaptive optics system capability for the deformable mirror was simulated and performance was predicted based on the Kolmogorov atmospheric turbulence model. However, in order to confirm the predictions, a closed-loop adaptive optics system was constructed with the insertion of an atmospheric turbulence simulator consisting of two point sources, a Boston deformable mirror, and double random phase plates. In order to simulate a binary star, the two point sources are mounted on 3-axis micron meter stages and are optically merged into a single beam path by a beam splitter cube. The light intensity of each source is adjustable to a corresponding stellar magnitude, while the angular separation can be precisely adjusted by moving the 3-axis stages. The atmospheric disturbance is generated by shaping the Boston deformable mirror and also by rotating the double phase plates. The Fried parameter of the generated the atmospheric disturbance corresponds to an area from 7 to 15 cm at 500 nm at the telescope pupil plane, which represents typical seeing conditions at the Bohyun observatory, South Korea.

The resolution and observing width of optical remote sensing camera can be enhanced by increasing the detector length of focal plane. Optical butting is used to increase the imaging length owing to its simple structure, low cost and the simultaneously-imaging detectors. But butting process is influenced by many factors, which result in imaging gaps, so overlapping pixels between adjacent detectors is the key. In this paper, the factors causing imaging gap are analyzed, and calculation of overlapping pixel number is given out based on the analysis. Firstly an optical-butting focal plane system is built on optical imaging principles. Then the factors causing imaging gap are listed and analyzed, under the consideration of telecentric and non-telecentric optical system. Based on the imaging gap analysis, a formula for overlapping pixel number calculation is obtained, with the MTF assurance in vignetting districts resulted from optical butting. Finally a calculation example of a camera is given.

Purpose to find an efficient, non-destructive examining method for showing the disappearing words after writing with automatic disappearance pen. Method Using the imaging spectrometer to show the potential disappearance words on paper surface according to different properties of reflection absorbed by various substances in different bands. Results the disappeared words by using different disappearance pens to write on the same paper or the same disappearance pen to write on different papers, both can get good show results through the use of the spectral imaging examining methods. Conclusion Spectral imaging technology can show the disappearing words after writing by using the automatic disappearance pen.

Yacht has become a fashionable way for entertainment. However, to obtain the precise location of a yacht docked at a port has become one of the concerns of a yacht manager. To deal with this issue, we adopt a positioning method based on the principle of binocular parallax and background difference in this paper. Binocular parallax uses cameras to get multi-dimensional perspective of the yacht based on geometric principle of imaging. In order to simplify the yacht localization problem, we install LED light indicator as the key point on a yacht. And let it flash at a certain frequency during day time and night time. After getting the distance between the LED and the cameras, locating the yacht is easy. Compared with other traditional positioning methods, this method is simpler and easier to implement. In this paper, we study the yacht positioning method using the LED indicator. Simulation experiment is done for a yacht model in the distance of 3 meters. The experimental result shows that our method is feasible and easy to implement with a small 15% positioning error.

The propagation characteristics of the orbital angular momentum in vortex waves have been studied. The representation of electric multipole radiation filed is derived from the Laugerre-Gaussian beams to electromagnetic vortex. Simulation results show the capability of using the orbital angular momentum for remote imaging.

Due to the problem that distortion exist for panoramic image stitching in terms of vehicle panoramic system, the research on image preprocessing for panoramic system was carried out. Firstly, the principal of vehicle panoramic vision was analyzed and the fundamental role that image preprocessing procedure plays in panoramic vision is found out. And then, the camera was calibrated with Hyperboloid model and the correction for distorted image is realized. Oriented by panoramic system, the effect of mounting position was taken into consideration for image correction and the suggested angle of camera installation for different mounting position is given. Through analyzing the existing problem of bird-eye image, a method of transforming twice with calibration plates is proposed. Experiment results indicate that the proposed method can weaken the existing problem of bird-eye image effectively and it can contribute to reduce the distortion in terms of image stitching for panoramic system.

Localization algorithm based on machine vision is a hot topic in the field of intelligent mobile robot．A fast method for mobile robot 3D SLAM (simultaneous localization and mapping) was presented to address the problem of 3D modeling in complex indoor environment. After filtering, the SIFT feature is extracted and matched between every two frames of the sequence. According to the camera calibration model and the image feature extraction and matching procedure, the association between two 3D point clouds was established. On the basis of the RANSAC (random sample consensus) algorithm, the correspondence based iterative closest point arithmetic model was solved to realize the robot’s precise localization effectively. With the key frame-to-frame selection mechanism, the 3D map method and the unique normal characteristic of a spatial point were used for maintaining and updating the global map. Experimental results demonstrate the feasibility and effectiveness of the proposed algorithm in the indoor environment.

In this paper, the wavelet threshold denoising method was used into the filtered back-projection algorithm of imaging reconstruction. To overcome the drawbacks of the traditional soft- and hard-threshold functions, a modified wavelet threshold function was proposed. The modified wavelet threshold function has two threshold values and two variants. To verify the feasibility of the modified wavelet threshold function, the standard test experiments were performed by using the software platform of MATLAB. Experimental results show that the filtered back-projection reconstruction algorithm based on the modified wavelet threshold function has better reconstruction effect because of more flexible advantage.

A number of the modern applications such as medical imaging, remote sensing satellites imaging, virtual prototyping etc use the High Dynamic Range Image (HDRI). Generally to obtain HDRI from ordinary digital image the camera is calibrated. The article proposes the camera calibration method based on the clear sky as the standard light source and takes sky luminance from CIE sky model for the corresponding geographical coordinates and time. The article considers base algorithms for getting real luminance values from ordinary digital image and corresponding programmed implementation of the algorithms. Moreover, examples of HDRI reconstructed from ordinary images illustrate the article.

Automatic weapon platform is one of the important research directions at domestic and overseas, it needs to accomplish fast searching for the object to be shot under complex background. Therefore, fast detection for given target is the foundation of further task. Considering that chest-shape target is common target of shoot practice, this paper treats chestshape target as the target and studies target automatic detection method based on Deformable Part Models. The algorithm computes Histograms of Oriented Gradient(HOG) features of the target and trains a model using Latent variable Support Vector Machine(SVM); In this model, target image is divided into several parts then we can obtain foot filter and part filters; Finally, the algorithm detects the target at the HOG features pyramid with method of sliding window. The running time of extracting HOG pyramid with lookup table can be shorten by 36%. The result indicates that this algorithm can detect the chest-shape target in natural environments indoors or outdoors. The true positive rate of detection reaches 76% with many hard samples, and the false positive rate approaches 0. Running on a PC (Intel(R)Core(TM) i5-4200H CPU) with C++ language, the detection time of images with the resolution of 640 × 480 is 2.093s. According to TI company run library about image pyramid and convolution for DM642 and other hardware, our detection algorithm is expected to be implemented on hardware platform, and it has application prospect in actual system.

Image deblurring is a fundamental problem in image processing. Conventional methods often deal with the degraded image as a whole while ignoring that an image contains two different components: cartoon and texture. Recently, total variation (TV) based image decomposition methods are introduced into image deblurring problem. However, these methods often suffer from the well-known stair-casing effects of TV. In this paper, a new cartoon -texture based sparsity regularization method is proposed for non-blind image deblurring. Based on image decomposition, it respectively regularizes the cartoon with a combined term including framelet-domain-based sparse prior and a quadratic regularization and the texture with the sparsity of discrete cosine transform domain. Then an adaptive alternative split Bregman iteration is proposed to solve the new multi-term sparsity regularization model. Experimental results demonstrate that our method can recover both cartoon and texture of images simultaneously, and therefore can improve the visual effect, the PSNR and the SSIM of the deblurred image efficiently than TV and the undecomposed methods.

Aiming at the uncertainty of traditional 3D video capturing including camera focal lengths, distance and angle parameters between two cameras, a red-blue 3D video fusion system based on DM642 hardware processing platform is designed with the parallel optical axis. In view of the brightness reduction of traditional 3D video, the brightness enhancement algorithm based on human visual characteristics is proposed and the luminance component processing method based on YCbCr color space is also proposed. The BIOS real-time operating system is used to improve the real-time performance. The video processing circuit with the core of DM642 enhances the brightness of the images, then converts the video signals of YCbCr to RGB and extracts the R component from one camera, so does the other video and G, B component are extracted synchronously, outputs 3D fusion images finally. The real-time adjustments such as translation and scaling of the two color components are realized through the serial communication between the VC software and BIOS. The system with the method of adding red-blue components reduces the lost of the chrominance components and makes the picture color saturation reduce to more than 95% of the original. Enhancement algorithm after optimization to reduce the amount of data fusion in the processing of video is used to reduce the fusion time and watching effect is improved. Experimental results show that the system can capture images in near distance, output red-blue 3D video and presents the nice experiences to the audience wearing red-blue glasses.

LIDAR three-dimensional imaging technology have been used in many fields, such as military detection. However, LIDAR require extremely fast data acquisition speed. This makes the manufacture of detector array for LIDAR system is very difficult. To solve this problem, we consider using compress sensing which can greatly decrease the data acquisition and relax the requirement of a detection device. To use the compressive sensing idea, a spatial light modulator will be used to modulate the pulsed light source. Then a photodetector is used to receive the reflected light. A convex optimization problem is solved to reconstruct the 2D depth map of the object. To improve the resolution in transversal direction, we use multiframe image restoration technology. For each 2D piecewise-planar scene, we move the SLM half-pixel each time. Then the position where the modulated light illuminates will changed accordingly. We repeat moving the SLM to four different directions. Then we can get four low-resolution depth maps with different details of the same plane scene. If we use all of the measurements obtained by the subpixel movements, we can reconstruct a high-resolution depth map of the sense. A linear minimum-mean-square error algorithm is used for the reconstruction. By combining compress sensing and multiframe image restoration technology, we reduce the burden on data analyze and improve the efficiency of detection. More importantly, we obtain high-resolution depth maps of a 3D scene.

Underwater range-gated laser imaging (URGLI) still has some problems like un-uniform light, low brightness and contrast. To solve the problems, a variant of adaptive histogram equalization called contrast limited adaptive histogram equalization (CLAHE) is proposed in this paper. In experiment, using the CLAHE and HE to enhance the images, and evaluate the quality of enhanced images by peak signal to noise ratio (PSNR) and contrast. The result shows that the HE gets the images over-enhanced, while the CLAHE has a good enhancement with compressing the over-enhancement and the influence of un-uniform light. The experimental results demonstrate that the CLAHE has a good result of image enhancement for target detection by underwater range-gated laser imaging system.

Layered depth video (LDV) is a sparse representation of MVD, which is considered as a promising 3D video format for supporting 3D video services. This format consists of one full view and additional residual data that represents side views. However, the amount of residual data becomes larger when the distance between the central view and side views increases. To address this problem, a new inpainting-based residual data generation method is proposed in this paper. Then, the inpainting-induced artifacts is considered as new residual data and the residual data of two side views is merged into one buffer to further reduce the amount of data. On the other hand, the block wise alignment is used for higher coding efficiency. And in order to fit the shape or distribution of residual data, a new compression algorithm for coding residual data is proposed. The experiments show high compression efficiency of the proposed method. The proposed method allows reduction of required bitrate of at least 30% comparing to classical LDV method, while they have the similar quality of intermediate virtual view in the terminal’s display.

Intelligent video surveillance is to analysis video or image sequences captured by a fixed or mobile surveillance camera, including moving object detection, segmentation and recognition. By using it, we can be notified immediately in an abnormal situation. Pedestrian detection plays an important role in an intelligent video surveillance system, and it is also a key technology in the field of intelligent vehicle. So pedestrian detection has very vital significance in traffic management optimization, security early warn and abnormal behavior detection. Generally, pedestrian detection can be summarized as: first to estimate moving areas; then to extract features of region of interest; finally to classify using a classifier. Redundant wavelet transform (RWT) overcomes the deficiency of shift variant of discrete wavelet transform, and it has better performance in motion estimation when compared to discrete wavelet transform. Addressing the problem of the detection of multi-pedestrian with different speed, we present an algorithm of pedestrian detection based on motion estimation using RWT, combining histogram of oriented gradients (HOG) and support vector machine (SVM). Firstly, three intensities of movement (IoM) are estimated using RWT and the corresponding areas are segmented. According to the different IoM, a region proposal (RP) is generated. Then, the features of a RP is extracted using HOG. Finally, the features are fed into a SVM trained by pedestrian databases and the final detection results are gained. Experiments show that the proposed algorithm can detect pedestrians accurately and efficiently.

As the light travels through the wavefront coding (WFC) system, the modulation transfer function(MTF) of the WFC system was very low, consequently the intermediate blurred image has been received by the detector. However, there is no zero point in the passband of the MTF of the WFC imaging system, and the target information cannot be saved very well. An appropriate filter can be used to restore the sampled intermediate image. The noise of the system is enlarged in the restoration process where the signal be amplified by the filter, and the signal to noise ratio(SNR) of the image is reduced. In order to solve the above issues, an improved algorithm has been proposed in this paper. The noise is controlled by the wavelet in the reconstruction process, and the intermediate blurred image is restored by the wiener filter algorithm with a prior knowledge of the degradation function. Thus, the wavelet de-noising and wiener filter algorithm are combined to restore the middle blurred image of the WFC system. Finally, the restoration image with the diffraction limit level is acquired in image detail restoration and noise control.

In order to achieve automatic measurement function, laser scanning measurement system need to ensure that measurement system of the cross and the target center overlapped. To develop the obtaining accuracy of the cross central coordinate, a novel algorithm based on the combination of two-step Hough Transform and Sub-pixel curve fitting was presented. Firstly, Hough Transform was used to detect the straight line of long axis of the cross, and then the central coordinate value of the cross can be deduced roughly through straight line equations. Secondly, take this coordinate as a center, and a certain number of pixel values as the edge length, a region of interest (ROI) can be defined. In this ROI, the two short axes of the cross were obtained to detect the straight line roughly using Hough Transform again. Then, the coordinate value of short axis was acquired at the pixel level with Canny edge detection algorithm. According to the equations of the straight line, rough errors were removed through 3σ rule, and the coordinate values of short axis at the sub-pixel level with curve fitting algorithm is obtained. Finally, based at the sub-pixel level of coordinate values, straight line equations of two short axes was presented with LSM, and thus the coordinate of cross center would be accurately detected with the intersection of two short axes. Experimental results showed that the accuracy of the central coordinates value of cross approaches to sub-pixel level and can satisfy the measurement system requirements of high precision.

To effectively improve the low contrast of human body region in the infrared images, a combing method of several enhancement methods is utilized to enhance the human body region. Firstly, for the infrared images acquired by Kinect, in order to improve the overall contrast of the infrared images, an Optimal Contrast-Tone Mapping (OCTM) method with multi-iterations is applied to balance the contrast of low-luminosity infrared images. Secondly, to enhance the human body region better, a Level Set algorithm is employed to improve the contour edges of human body region. Finally, to further improve the human body region in infrared images, Laplacian Pyramid decomposition is adopted to enhance the contour-improved human body region. Meanwhile, the background area without human body region is processed by bilateral filtering to improve the overall effect. With theoretical analysis and experimental verification, the results show that the proposed method could effectively enhance the human body region of such infrared images.

In a binocular visual system, to recover the three-dimensional information of the object, the most important step is to acquire matching points. Structure tensor is the vector representation of each point in its local neighborhood. Therefore, structure tensor performs well in region detection of local structure, and it is very suitable for detecting specific graphics such as pedestrians, cars and road signs in the image. In this paper, the structure tensor is combined with the luminance information to form the extended structure tensor. The directional derivatives of luminance in x and y directions are calculated, so that the local structure of the image is more prominent. Meanwhile, the Euclidean distance between the eigenvectors of key points is used as the similarity determination metric of key points in the two images. By matching, the coordinates of the matching points in the detected target are precisely acquired. In this paper, experiments were performed on the captured left and right images. After the binocular calibration, image matching was done to acquire the matching points, and then the target depth was calculated according to these matching points. By comparison, it is proved that the structure tensor can accurately acquire the matching points in binocular stereo matching.

The existence of vertical parallax is the main factor of affecting the viewing comfort of stereo video. Visual fatigue is gaining widespread attention with the booming development of 3D stereoscopic video technology. In order to reduce the vertical parallax without affecting the horizontal parallax, a self-adaptive image scaling algorithm is proposed, which can use the edge characteristics efficiently. In the meantime, the nonlinear Levenberg-Marquardt (L-M) algorithm is introduced in this paper to improve the accuracy of the transformation matrix. Firstly, the self-adaptive scaling algorithm is used for the original image interpolation. When the pixel point of original image is in the edge areas, the interpretation is implemented adaptively along the edge direction obtained by Sobel operator. Secondly the SIFT algorithm, which is invariant to scaling, rotation and affine transformation, is used to detect the feature matching points from the binocular images. Then according to the coordinate position of matching points, the transformation matrix, which can reduce the vertical parallax, is calculated using Levenberg-Marquardt algorithm. Finally, the transformation matrix is applied to target image to calculate the new coordinate position of each pixel from the view image. The experimental results show that: comparing with the method which reduces the vertical parallax using linear algorithm to calculate two-dimensional projective transformation, the proposed method improves the vertical parallax reduction obviously. At the same time, in terms of the impact on horizontal parallax, the proposed method has more similar horizontal parallax to that of the original image after vertical parallax reduction. Therefore, the proposed method can optimize the vertical parallax reduction.

In recent years, we have witnessed the prosperity of the face image super-resolution (SR) reconstruction, especially the learning-based technology. In this paper, a novel super-resolution face reconstruction framework based on support vector regression (SVR) about a single image is presented. Given some input data, SVR can precisely predict output class labels. We regard the SR problem as the estimation of pixel labels in its high resolution version. It’s effective to put local binary pattern (LBP) codes and partial pixels into input vectors during training models in our work, and models are learnt from a set of high and low resolution face image. By optimizing vector pairs which are used for learning model, the final reconstructed results were advanced. Especially to deserve to be mentioned, we can get more high frequency information by exploiting the cyclical scan actions in the process of both training and prediction. A large number of experimental data and visual observation have shown that our method outperforms bicubic interpolation and some stateof- the-art super-resolution algorithms.

This paper aims at designing and demonstrating a video and audio signals synchronous wireless transmission of free-space optical communication (FSO) system. Video and audio signals were modulated by the acousto-optical modulator (AOM). With the help of a designed circuit and programmable microcontroller, the system based on time division multiplexing (TDM) achieves the functions of signal de-multiplexing and wireless transmitting. Proved by experiments, the system meets the actual requirements with advantages of flexibility, practicality and low cost. And it provides an efficient scheme of synchronous wireless transmission of video and audio signals for monitoring system and TV access network.

With the availability of more powerful computing processors, iterative reconstruction algorithms have recently been successfully implemented as an approach to achieving significant dose reduction in X-ray CT. In this paper, we propose an adaptive iterative reconstruction algorithm for X-ray CT, that is shown to provide results comparable to those obtained by proprietary algorithms, both in terms of reconstruction accuracy and execution time. The proposed algorithm is thus provided for free to the scientific community, for regular use, and for possible further optimization.

In order to study the influence of nonlinear sweep voltage on the range accuracy of streak tube imaging lidar, a nonlinear distance model of streak tube is proposed. The model of the parallel-plate deflection system is studied, and the mathematical relation between the sweep voltage and the position of the image point on the screen is obtained based on the movement rule of phoelectron. And the mathematical model of the sweep voltage is established on the basis of its principle. The simulation of streak image is carried out for the selected staircase target, the range image of the target can be reconstructed by extremum method. Comparing reconstruction result and actual target, the range accuracy caused by the nonlinear sweep voltage is obtained. The curve of the errors varying with target ranges is also obtained. And the range accuracy of the system is analyzed by the means of changing the parameter relate to sweep time.

The variation of the different parameters of the exciting radiation and the registration of the fluorescence of the investigated object allows to obtain multidimensional spectral images: from three-dimensional (length of the exciting radiation, the wavelength of the emitted radiation, the fluorescence intensity) to eight and more dimensions (in addition to three of these dimensions: spatial coordinates x, y, z; time of measurements; the duration and the intensity of the exciting radiation and etc.). In the case of measurements in natural conditions is highly desirable that the result of the processing performed during a single measurement for the operation in real time.

In this paper we consider the approach described for the treatment of fluorescence measurements of dissolved organic matter and chlorophyll-a in seawater. Joint analysis of the various pairs of wavelengths of excitation / emission fluorescence, fluorescence analysis at different durations of the exciting radiation and the time-spatial analysis of the received signal allow identifying different types of fluorescent dissolved organic matter and estimate their stage of biodegradation, to study the functional state of phytoplankton cells. So it is possible to provide real-time investigation of environmental indicators of seawater.

Excitation-emission matrix (EEM) technique is widely used to study the properties of organic matter in seawater. Additional variations of the exciting radiation parameters (in addition to the wavelengths), such as duration and intensity, allow to get more spectral information about the object of research, and allow to identify some individual components or to study the dynamics of the molecular changes in the sample of analyzed liquid.

One of the main limitations of the EEM is that usually necessary to carry out research in the laboratory on the specialized equipment. The paper describes the hardware system allowing carry out measurements in the field studies.

The hardware system includes a set of light-emitted diodes (LEDs) working in the range from 245 to 600 nm, which produce sequential excitation fluorescence of the sample liquid, which is detected using a 32-channel photomultiplier tube (PMT). Number of LEDs can vary from 7 to 16. Through the use of multi-channel photomultiplier reached the required measurement sensitivity and efficiency measurement is less than 1 minute for full cycle.

Minimum 4 channel in excitation range of 350-600 nm support the work of variable duration excitation pulses from microseconds to seconds. It makes possible to measure the fluorescence of the chlorophyll-a with closed or open reaction centers of phytoplankton cells. The use of multiple excitation radiation allows to activate various mechanisms of energy transfer in the photosynthetic apparatus of phytoplankton cells, improves measurement accuracy, reduces dependence on the variation in species composition and functional state of phytoplankton.

Laser range-gated imaging has great potentials in remote night surveillance with far detection distance and high resolution, even if under bad weather conditions such as fog, snow and rain. However, the field of view (FOV) is smaller than large objects like buildings, towers and mountains, thus only parts of targets are observed in one single frame, so that it is difficult for targets identification. Apparently, large FOV is beneficial to solve the problem, but the detection range is not available due to low illumination density in a large field of illumination matching with the FOV. Therefore, a large field-of-view range-gated laser imaging is proposed based on image fusion in this paper. Especially an image fusion algorithm has been developed for low contrast images. First of all, an infrared laser range-gated system is established to acquire gate images with small FOV for three different scenarios at night. Then the proposed image fusion algorithm is used for generating panoramas for the three groups of images respectively. Compared with raw images directly obtained by the imaging system, the fused images have a larger FOV with more detail target information. The experimental results demonstrate that the proposed image fusion algorithm is effective to expand the FOV of range-gated imaging.