最新刊期

    19 11 2014
    • Visual perception and processing for intelligentvideo surveillance:a review

      Huang Tiejun, Zheng Jin, Li Bo, Fu Huiyuan, Ma Huadong, Xue Xiangyang, Jiang Yugang, Yu Junqing
      Vol. 19, Issue 11, Pages: 1539-1562(2014) DOI: 10.11834/jig.20141101
      摘要:With the increasing maturity of video surveillance technologies and popularity of surveillance equipment, video surveillance applications are increasingly widespread. The amounts of surveillance video are showing the explosive growth. In the era of big data, the data of surveillance video has become one of the important data objects. However, due to the unstructured nature of video data, the processing and analysis of multimedia data is relatively difficult. In face of huge video data captured by a large number of surveillance cameras, how to effectively transmit, store, analyze and identify in accordance with the multimedia content and features, has become an urgent need. For the problems of large scale visual perception and intelligent processing in the area of intelligent video surveillance, this report is organized around surveillance video encoding, target detection and tracking, augmented surveillance video together with video motion and identifying abnormal behavior four research directions, and elaborate their development status in 2013 and future development trend outlook. China's latest national standards AVS2 has twice coding efficiency of surveillance video than the latest international standard H.265/HEVC, which marks the video coding technology and standard of our country has already realized the leap in the field of video surveillance. Study on the detection and tracking of video moving object is focused on effective feature extraction and classifier training, and the introduction of machine learning method makes the moving target detection and tracking based on multiple instance learning and sparse representation become a hot spot of research. Surveillance video quality enhancement includes removing of fog, rain, snow, and night, deblurring and super-resolution enhancement, whose existing algorithms are good for a certain class of image clear effect, and on the other class is relatively poor, universality is not high. Intelligent action analysis and abnormal behavior recognition technology existing got continuous development, the performance of the algorithm is also rising, but from a practical point of view, in addition to a specific scene or controllable simple, there is not yet too many mature application system. With the arrival of the era of big data, intelligent video surveillance needs to be increasingly urgent, in the face of many challenges at the same time, this research field will welcome great opportunity hitherto unknown, will have more and more research to practical.  
      关键词:video surveillance;target detection;target tracking;video enhancement;behavior recognition   
      5668
      |
      678
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121450 false
      更新时间:2024-05-08
    • Hu Wenrui, Xie Yuan, Zhang Wensheng
      Vol. 19, Issue 11, Pages: 1563-1569(2014) DOI: 10.11834/jig.20141102
      摘要:The challenge of image denoising is how to filter the noise as more as possible while preserve the image information at the same time. To reduce this contradiction, we propose an iterative image denoising method by combining the non-local techniques and the high order singular value decomposition (HOSVD). Our method first uses the non-local patch clustering and the HOSVD to construct the data-adaptive 3D transform basis and coefficients. Then an inverse transformation is taken to get the denoised patches after filtering the coefficients by a threshold value. Usually, denoising once is not enough to get the desired denoised result. Therefore, we design an iterative denoising strategy for the proposed method, which is verified to be a process to take the trade-off between bias and variance and optimize the mean square error (MSE). Experiments illustrate that our method not only filter the noise effectively but also preserve the texture well. Our method combines advantages of both non-local collaborative filtering and data-adaptive denoising. Compared with the other three advanced denoising methods, such as, BM3D, NCSR and PLOW, our method outperforms them in terms of texture preserving.  
      关键词:high order singular value decomposition (HOSVD);mean square error;non-local collaborative filtering;data-adaptive   
      4253
      |
      333
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121224 false
      更新时间:2024-05-08
    • Single image super-resolution in wavelet domain with double sparse

      Yang Bo, wu jitao, Xie Xiaozhen
      Vol. 19, Issue 11, Pages: 1570-1576(2014) DOI: 10.11834/jig.20141103
      摘要:Super-resolution is a challenging technique for recovering lost information according to natural image priori. As an important priori, sparse has been widely studied in the field of image processing, such as in image recovery, inpainting, demosaicing, and denoising. Given the development of compressed sensing theory and the L optimization method, a large number of super resolution methods have been proposed based on sparse representation. An example is the single image super-resolution based on sparse representation, which has been widely studied in recent years. Based on the super-resolution model through sparse representation, where the feature image patch can be represented sparsely, a novel method conducted in the wavelet domain is proposed in this work. Our method is based on the sparse of high-frequency image patches and high-frequency image patches in redundant dictionaries. First, a decomposed coefficient image is obtained after using the discrete wavelet transform in a low-resolution image. Second, in connection with the high-frequency coefficients of the low-resolution image, a double sparse model with super-resolution is established to recover the detail coefficients of a high-resolution image correspondingly. Third, the decomposed coefficients of the low-resolution image and the recovered high-frequency coefficients of the high-resolution image are merged into wavelet coefficients for second floor decomposition. Finally, with the multi-scale property of wavelet and the assumption that a low-resolution image can be used as a high-resolution image of low frequency coefficient approximation, a super-resolution image is reconstructed with two layers of inverse wavelet transformation with low-resolution image wavelet decomposition and estimated high-frequency coefficients of high-resolution images. In the model solving process, we adopt a fast solving method called the constrained splitting Bregman method, which is widely used to solve the L problem. Unlike the method for joint feature spaces, the constrained splitting Bregman method uses two dictionaries with high and low feature spaces. The low-resolution dictionary is learned from a low-resolution feature space using the k-svd method while the resolution dictionary is learned from least square approximation. The sparse model of redundant dictionary is known to recover good texture and denoise at the same time. The double sparse method has the advantage of the famous SR method through sparse representation and obtains good denoising performance. Through several experiments for standard pictures, the double sparse method restores image local texture and edges well and achieves a good effect on a noised image because of the use of multi-scale property and sparse with high-frequency coefficients. Our method involves less computational complexity compared with the popular sparse super-resolution method because of the use of only three quarters of the image patches of the original image. The sparse model is widely used in image recovery. Based on the sparse of detail coefficients in wavelet domain and on the sparse of feature image patches under redundant dictionaries, a novel single image super-resolution method in wavelet domain is proposed in this work. The double sparse method preserves good local texture and edges and obtains better results for noised image with less computational complexity compared with the conventional method via sparse representation.  
      关键词:wavelet domain;double sparse;sparse representation;super-resolution   
      3309
      |
      302
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56118864 false
      更新时间:2024-05-08
    • Magnifying imperceptible variations in infrared thermal videos

      Fu Chuanqing, Gu Xiaojing, Gu Xingsheng
      Vol. 19, Issue 11, Pages: 1577-1583(2014) DOI: 10.11834/jig.20141104
      摘要:Infrared thermal devices are widely used in the industrial, medical, military, and other fields and significantly expand the visual perception range of humans. However, infrared thermal videos have low contrast and blurred details, and the subtle variations in these videos are difficult to observe. In this work, we design an amplification method for magnifying the barely seen changes in infrared thermal videos based on Eulerian perspective. Our goal is to reveal temporal variations that are difficult or impossible to be seen with the naked eye and to display these variations in an indicative manner. The proposed method uses an infrared thermal video as input and applies contrast pyramid decomposition to each frame, followed by temporal filtering of the decomposed images. The signals selected by temporal filtering are then amplified to reveal hidden information. The contrast pyramid is then constructed. Finally, noise reduction is performed on the reconstructed images, and the final output is obtained. Corresponding infrared thermal videos were obtained, and experiments were conducted on these videos to magnify the subtle colors and motion variations. For example, the signal from 0.75 Hz to 1 Hz of a profile video was filtered and magnified 100 times as output to augment the color changes in the video. The signal from 100 Hz to 120 Hz of a guitar video was also filtered and then magnified 30 times as output to augment the motion variations in the video. Experiments showed that the proposed method can effectively magnify imperceptible variations in infrared thermal videos. This study demonstrates that imperceptible variations in infrared thermal videos can be magnified and shown to observers in an indicative manner, which is valuable for both military and civil fields.  
      关键词:infrared thermal videos;magnification of imperceptible variations;Eulerian perspective;contrast pyramid;BM3D noise reduction method   
      3762
      |
      277
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121511 false
      更新时间:2024-05-08
    • Calibration of Kinect sensor with depth and color camera

      Guo Lianpeng, Chen Xiangning, Liu Bin
      Vol. 19, Issue 11, Pages: 1584-1590(2014) DOI: 10.11834/jig.20141105
      摘要:Given the low efficiency and poor accuracy of the recent calibration method for the Kinect sensor with depth and color camera, we propose an improved algorithm that can quickly and accurately calibrate the color and depth camera of the Kinect sensor based on color and disparity images. The algorithm considers the features of color and depth simultaneously to improve the calibration of the Kinect sensor. The color camera was calibrated using Zhang's method, and the depth camera was calibrated using a Kinect disparity distortion correction model. This model represents the geometric relationship between disparity and depth and uses a spatially varying offset simplified by Taylor's formula that decays as Kinect disparity increases. The Kinect sensor was calibrated using multiple views of the calibration plane to obtain the color and disparity images that would be used to obtain the distortion parameters and the rotation and translation matrix between the two cameras. The spatial distortion parameter by Taylor's formula was streamlined to simplify the distortion correction model of the depth camera and to optimize the solution process. The reprojection error of the color camera was 0.33, whereas that of the depth camera was 0.798; the running tine of our calibration process was 116 s. Experiments show that the proposed algorithm can ensure the accuracy and improve the efficiency of the calibration process.  
      关键词:disparity image;Kinect sensor;distortion correction;camera calibration   
      7055
      |
      383
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56118950 false
      更新时间:2024-05-08
    • Contour detection based on spatially unified modulation model

      Xiao Jie, Cai Chao, Guo Zhaoli
      Vol. 19, Issue 11, Pages: 1591-1595(2014) DOI: 10.11834/jig.20141106
      摘要:The response of the central neuron in the primary visual cortex (V1) is not only affected by the stimulus inside the classical receptive field but also modulated by the surrounding stimulus (NCRF). Contextual modulation is mediated by horizontal connections across the visual cortex. This study proposes a contour detection method based on the visual mechanism in V1. First, the response of every visual neuron in V1 is computed by local energy. Second, facilitation and suppression (contextual influence) on a neuron through horizontal interactions are obtained by constructing a spatially unified modulating function. Finally, the total output response of one neuron to complex visual stimuli is acquired by combining the influence of local visual context on the neuron and energy response. Unlike previous studies, the present work unified the facilitation and suppression modulated by surrounding elements in the image without separating the regions of excitatory and inhibitory lateral connections. The proposed method was tested on synthetic and natural images; encouraging results were obtained. The proposed model is expected to improve contour detection performance when images contain objects of interest on a cluttered background.  
      关键词:contour detection;visual perception mechanism;Gabor filter;spatially unified modulation   
      3104
      |
      229
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121534 false
      更新时间:2024-05-08
    • Image segmentation by integrating similarity fitting and spatial constraint

      Zhang Zhengrong, Zhan Tianming, Wei Zhihui
      Vol. 19, Issue 11, Pages: 1596-1603(2014) DOI: 10.11834/jig.20141107
      摘要:An image cannot be segmented well using one feature in one subclass because the objects in an image region contain different subclasses. The sparse representation classifier (SRC), a combined result of machine learning and compressed sensing, can solve such problems by constructing a dictionary that contains all the features of all subclasses. However, the performance of the SRC depends on an over-complete dictionary. Thus, the SRC needs a large number of training samples to obtain a good segmentation result. Therefore, we propose an interactive segmentation method based on similarity search and spatial constrain. First, each object dictionary that contains different features from corresponding subclasses is designed separately by manual labeling. The whole dictionary is then constructed by arranging the object dictionaries obtained by manual labeling in sequence. The most similar feature of the test sample is searched in each dictionary. The fitting term is built using the distance between the testing sample and the most similar feature in each subclass. Afterward, the object function is built by integrating the fitting term and the spatial constraint using total variation. With the addition of the spatial constraint, our model can reduce the effect of noise on the image segmentation results. The continuous max-flow algorithm is applied to minimize the objective function and to effectively obtain segmentation results. The continuous max-flow algorithm, which is linked to the continuous min-cut problem, is a novel method for solving the standard TV-based formulations of the Potts model. This algorithm can avoid extra computational load in enforcing simplex constraints and naturally allows parallel computations over different labels. Our method can segment different nature images with different object shapes or contents and is robust to the location and number of training samples. Compared with that of the SRC method, the efficiency of our method is relatively better when using the same dictionary. In addition, the segmentation results of our method are much better than those of traditional segmentation methods, such as N-cut and the logistic regression classifier. These results demonstrate that our method can segment a whole region of different objects in an image. The SRC can segment an image that contains different subclasses using an over-complete dictionary and a reconstruction strategy. However, the segmentation performance of the SRC is weak when only a few training samples are used. In this work, the similar feature fitting method and a spatial constraint are used to build a Potts model. The continuous max-flow is applied to solve the objective function and to obtain segmentation results. Our method offers the following advantages. 1) The feature fitting strategy is suitable for image segmentation with a small number of training samples, and it can solve the problem where image objects contain different subclasses. 2) The spatial constraint based on total variation can avoid the noise effect during image segmentation and improve the accuracy of the segmentation results. 3) The continuous max-flow applied to solve our objective function can avoid extra computational load in enforcing simplex constraints and naturally allows parallel computations over different labels. Comparative experiments with SRC, N-cut, and LRC methods demonstrate that our method can segment a whole region with different objects in an image.  
      关键词:image segmentation;similarity search;spatial constraint;continuous max-flow   
      3171
      |
      361
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56120963 false
      更新时间:2024-05-08
    • Zheng Shan, Fan Huijie, Tang Yandong, Wang Yan
      Vol. 19, Issue 11, Pages: 1604-1612(2014) DOI: 10.11834/jig.20141108
      摘要:The detection of optic disc and cup is important in fundus image analysis and in the computer-aided diagnosis of optic nerve disease. In this study, an elliptical multiphase active contour model is proposed based on the oval shape of the optic disc and cup. The proposed algorithm can segment the optic disc and cup simultaneously and exactly. First, a multiphase active contour model is developed based on gray images of the optic disc and cup with different brightness levels. Afterward, the oval constraints are embedded in the model. The parametric equation of an ellipse can be obtained by finding the solution for the energy function of the model. Two oval initial curves are initially set. The evolution equation drive curves move toward the direction of the optic disc and cup. The curve evolution stops upon reaching the object contour border. The proposed algorithm is validated through experiments on its anti-noise ability and on the selection of different initial curves using different medical fundus images. The obtained results are compared with those of other algorithms. The experimental results show that unlike other models, the proposed model can simultaneously segment the optic disc and cup. The approach proposed in this paper can achieve superior image segmentation results. A novel image segmentation method is presented in this work. The experimental results show that the method can accurately segment the optic disc and cup. The algorithm is proven to be not pretreatment, robust, and efficient.  
      关键词:fundus image;optic disc and cup segmentation;C-V model;elliptical multiphase active contour model;multiphase level set function   
      3467
      |
      285
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56119151 false
      更新时间:2024-05-08
    • Precise local feature description for facial expression recognition

      Hu Min, Jiang He, Wang Xiaohua, Chen Hongbo, Li Kun, Ren Fuji
      Vol. 19, Issue 11, Pages: 1613-1622(2014) DOI: 10.11834/jig.20141109
      摘要:To identify facial expressions accurately, we propose a precise local feature description method for facial expression recognition. First, the eyebrows, eyes, and mouth in a facial expression image are identified and extracted. The local features from the organ images are then obtained and processed by the expanded vector triangle pattern. The outline and detail features of the images can be statistic. Finally, different scales of sufficient vector triangle patterns are used to describe the features of the different organs. Various scales of sufficient vector triangle patterns are then combined to describe the features of the same organ. In this way, key organ information can be expressed fully. Experiments on the proposed method were performed using the JAFEE, Cohn-Kanade (CK), and Pain Expressions database. The average recognition rates were 95.67%, 97.83%, and 84.0%, and the average durations of feature extraction were 11.70 ms, 30.23 ms, and 11.73 ms. The cross validation results showed that the precise local feature description method for facial expression recognition is fast and accurate. Through organ segmentation and the construction of flexible full vector triangle patterns, the precise local feature description method performs well in image feature description while consuming little time. The recognition results of the proposed method are better than those of the typical facial expression recognition method.  
      关键词:facial expression recognition;precise local features;sufficient vector triangle;a variety of scales   
      3810
      |
      298
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56119020 false
      更新时间:2024-05-08
    • Hu Haifeng, Chen Suting
      Vol. 19, Issue 11, Pages: 1623-1629(2014) DOI: 10.11834/jig.20141110
      摘要:A method for textural feature extraction based on Gabor wavelet and improved local binary pattern (LBP) is proposed to classify and recognize the surface roughness of a work piece through its image and improve accuracy. Given that the LBP operator ignores the magnitude differences between neighbors, the magnitude-considered LBP (M_LBP) operator is proposed. The magnitude of the gray level differences between neighbors is defined as . The gray mean of the image is used as threshold for the binarization of . The binarization result is appended to the top digit of the LBP in this neighbor, which is obtained by dividing the LBP according to the value. Before the recognition of the surface roughness of the work piece, the surface image is obtained with a stereomicroscope and then preprocessed. A self-similar Gabor wavelet filter bank is acquired by changing the scale and orientation parameters. The filter bank is used for surface image filtering. The multi-scale and multi-resolution Gabor texture features of the image are acquired. Afterward, the magnitude of the Gabor texture features is calculated, and the Gabor magnitude maps (GMMs) are extracted. The proposed M_LBP operator is then applied in the GMMs to extract M_LBP feature maps. Based on these M_LBP feature maps, the texture feature vector for each surface image can be constructed. After extracting these texture feature vectors, the k-nearest neighbor (KNN) algorithm is used in roughness recognition. The feature vectors of both the training samples and the samples to be recognized are extracted. Then, the, nearest neighbors of the samples to be recognized are selected from the training samples. The roughness class of the sample to be recognized can be acquired according to the roughness classes of these nearest neighbors. Different values are selected for (, ) in the M_LBP operator and for in the KNN algorithm. The comparative experiment shows that the recognition accuracy is highest when (, ) is set to (8, 8) and is set to 4. We use the LBP, Gabor combined with LBP, and Gabor combined with M_LBP to extract texture feature vectors. Through these vectors, we compare the time consumption (0.2886, 0.9546, and 1.1562 s) and the recognition accuracy (74%, 82%, and 98%). The experimental results demonstrate that the proposed method can recognize the surface roughness of the work piece with 98% accuracy and a difference of 0.2 μm, which is better than that of the other two algorithms. The proposed M_LBP operator can refine LBP information, and the Gabor wavelet combined with M_LBP overcomes the limitations of LBP, which includes single scale, single orientation, and disregard for magnitude. Hence, the method can be applied in roughness recognition with high precision.  
      关键词:roughness recognition;Gabor wavelets;M_LBP;KNN classifier   
      3188
      |
      276
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56119058 false
      更新时间:2024-05-08
    • The scalable flowers category recognition

      Miao Jinquan, Cao Weiqun
      Vol. 19, Issue 11, Pages: 1630-1638(2014) DOI: 10.11834/jig.20141111
      摘要:The methods based on pattern recognition achieve quite different accuracy when using different features or classifiers. The purpose of research is to construct the flower category recognition method rapidly, measure its performance, reduce the development workload of researchers and improve efficiency. According to the normal procedure of flower category recognition using pattern recognition technology, the steps of the algorithm are programmed as different kinds of processors to build up an expandable system using the plug-in technology, and the researchers can change the image processing and recognition algorithm by choosing the corresponding processors. Based on this, the algorithm is constructed by linking the processor in the way of data flow and represented as a network. We use the XML description file to describe the dependencies of a plugin and object pool to provide the communication foundation between processors which is the base of the scalable construction methods. The processors are encapsulated into processor plugins and managed by the main framework. The processors are divided into three kinds of processors to input data, process data, and buffer data. The core processor plugin controls the execution order of processors and make sure the data inputted to the processor valid. The constructed processor network of flowers category recognition algorithm normally contains input-processor, feature extractors, buffer processor and classifier processor. The input-processor can generate one input data and data index each time. Feature extractors will be run for several times to process the input data sequence. In the procedure of classifier training or accuracy statistic, a feature vector buffer is needed. While the network that represents the algorithm is processing, the ports in processor can transport messages and data produced. Finally we compared and analyzed different features in color, shape and texture using K-Nearest Neighbor and support vector machine classifier respectively. We use HSV color space histogram to reduce the effect of light, Hu moment and edge curvature histogram to represent shape and polar gray-level co-occurrence matrix for texture. Then different accuracies are achieved by changed the parameter. We can use the constructed method to optimize the parameter of each flower feature extractor rapidly. After that, we combine the features which are extracted by flower extractor processors using the optimized parameter as a flower feature vector. At the same time, multithread technology is used to speed up the process of the algorithm constructed. We construct and optimize the algorithm using the method proposed in the paper. The recognition rate is 91.26% for first hypothesis and 98.41% for fifth hypothesis while using the optimized algorithm on a dataset of 68 flower species. We use the proposed method to construct a flower feature extraction algorithm on 853 training images and finish the whole procedure in 2125 s. The connection method of workflow can construct an algorithm fast and benefit the rapid assessment of the performance of different features and classifiers in flowers category recognition. It is applicable to the research and optimization of algorithms. The proposed method based on workflow and plugin technology is easy to use and flexible and the algorithm constructed has good scalability. Furthermore, the system can be applied to the other researches using the methods of pattern recognition based on digital image.  
      关键词:plug-in;flow structure;scalable;flowers category recognition   
      3321
      |
      279
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121387 false
      更新时间:2024-05-08
    • Medical image fusion based on lifting wavelet transform

      Li Junfeng, Jiang Xiaoli, Dai Wenzhan
      Vol. 19, Issue 11, Pages: 1639-1648(2014) DOI: 10.11834/jig.20141112
      摘要:Medical image fusion is important in the field of disease diagnosis because it can improve the availability of information contained in images. To address the problem of multi-modal medical image fusion, this study proposes a new algorithm for medical image fusion based on the characteristics of lifting wavelet transform. First, the source multi-modal medical images after registration are decomposed into low and high frequency sub-bands by applying lifting wavelet transform. Second, image fusion rules are put forward according to the different features of the low and high frequency sub-bands. A fusion rule based on weighted region average energy is adopted for the low-frequency sub-band coefficients. For the high-frequency sub-band coefficients, the weighed box-counting method is applied in the fusion rules of low-rise sub-bands with low noise content, and the fusion rule of the weighed local area energy of the image gradient is used for high-rise sub-bands with high noise content. Several experiments that compare the previous with new medical image fusion algorithms are conducted for gray and color images. The experiment results are then analyzed in terms of visual quality and objective evaluation. The proposed algorithm can effectively preserve edge information. This study demonstrates that the proposed algorithm based on lifting wavelet transform can effectively preserve a large amount of information and significantly improve the performance of fusion images in terms of visual quality and objective evaluation index.  
      关键词:medical image fusion;lifting wavelet transform;region energy;box-counting methods;local area energy of image gradient   
      4005
      |
      328
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56120710 false
      更新时间:2024-05-08
    • Robust coding transmission for SAR image based on compressive sensing

      Hou Xingsong, Tian Wenwen, Gong Chen
      Vol. 19, Issue 11, Pages: 1649-1656(2014) DOI: 10.11834/jig.20141113
      摘要:Consider a wireless communication system with a radio station in an airborne synthetic aperture radar (SAR) system operating in a time-varying channel. Unpredictable packet loss occurs during transmission. Therefore, building a robust and efficient SAR image coding transmission scheme is necessary. Although traditional joint source-channel coding (JSCC) can achieve excellent and efficient transmission performance under fixed channel conditions, the predetermined redundancy of channel coding was adopted to achieve robustness. However, when the deterioration of channel condition exceeds the correction capacity of the channel codec in a time-varying channel, reconstruction performance declines at the decoder. In this work, we propose a robust SAR image coding transmission scheme over a time-varying channel using the democracy of compressive sensing (CS). A range of methods to improve the rate-distortion performance of the proposed scheme are also adopted. The reconstruction performance depends only on the number of measurements received and not on the actual measurements received; that is, every measurement is independent and nearly equal. Given the rich edge information of an SAR image, directional lifting wavelet transform (DLWT) is adopted as sparse representation to improve the representation of the edges of the SAR image. Although DLWT can attain good sparse representation for SAR images, this method cannot ensure strict sparse representation in CS; that is, representation coefficients still contain small coefficients that would interfere in the recovery of large coefficients. Thus, sparse filtering (setting small coefficients to zero) is also adopted in this study to eliminate the interference of small coefficients. Compared with the deterministic model-based CS reconstruction algorithms, the Bayesian model-based CS reconstruction algorithm is more reliable in a random signal scenario. Thus, we adopt an efficient Bayesian reconstruction algorithm called tree structured wavelet that exploits the structure dependencies of wavelet coefficients to attain high-performance image reconstruction. The proposed scheme achieves more robust SAR image coding transmission compared with the traditional JSCC scheme, CCSDS-RS, at the same rate. When the packet loss rate (PLR) reaches 0.05, the reconstruction performance of DSFB-CS is higher than that of CCSDS-RS. Therefore, CCSDS-RS is highly sensitive against packet loss and can easily lead to the cliff effect with channel deterioration. By contrast, the R-D performance in DSFB-CS decreases gracefully with channel deterioration. As the proposed scheme is based on reconstruction algorithm TSW, DSFB-CS is compared with the traditional scheme based on TSW without sparse filtering that uses DWT as sparse representation. DSFB-CS can improve the peak signal-to-noise ratio (PSNR) up to 3.9 dB. The proposed DSFB-CS achieves more robust transmission for SAR images compared with the traditional JSCC scheme, CCSDS-RS. The reconstruction performance of the proposed scheme declines slowly with the channel deterioration. Even under a time-varying channel, the robustness of DSFB-CS can still ensure that the decoder side can attain a relatively stable reconstructed image. Compared with the traditional CS scheme based on TSW, the proposed scheme exhibits significantly improved PSNR. The percentage of the reserved large coefficients in all coefficients and the bits of every measurement are both key parameters.  
        
      3576
      |
      256
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121023 false
      更新时间:2024-05-08
    • Wang Mingfei, Jia Jinyuan, Zhang Chenxi
      Vol. 19, Issue 11, Pages: 1657-1668(2014) DOI: 10.11834/jig.20141114
      摘要:Recent studies have developed virtual environment technologies based on peet-to-peer(P2P) networking. Some articles have reviewed these technologies from different aspects. However, analysis from the aspect of scene data transmission remains lacking. To date, an increasing number of scholars are beginning to investigate scene data transmission. Therefore, a survey of virtual scene transmission works based on P2P networking is indispensable. A summary research on this challenging issue was performed, and existing achievements from diverse aspects were analyzed. The transmission technologies of large-scale 3D virtual scenes were outlined. Lightweight modeling and streaming technologies for large-scale virtual scene, virtual environment network architecture, interest management, resource/neighbor discovery mechanism, scene data delivery strategies of different network architecture, cache update mechanism, and data perfecting mechanism were proposed for virtual environment transmission. These technologies were reviewed and compared, and their relationships were determined. A significant number of similar objects and model elements exist in a large-scale virtual scene. Thus, reusing repeated components can significantly reduce the amount of scene data. Concerning the lightweight works of 3D models, this study discusses the proposition and improvement of a model lightweight method and a new voxel unit-matching method. The advantages and disadvantages of symmetry methods for singly connected models were analyzed, and the proposition and definition of model streaming technologies were introduced. Several streaming methods and data structure of the base/incremental mesh were also expounded. The maturity of cloud computation depends on its flexible contraction, powerful performance, stable bandwidth, and cost-effective characteristics. Thus, an increasing number of distributed virtual environments were deployed for Cloud service, particularly in massively multiplayer online games applications. Combining P2P networking with cloud computing, we deployed the super-node on the cloud to maximize the stability of Cloud and the scalability of P2P. Such properties can enhance the stability and load balance of the entire driver vision enhancement (DVE) system. Currently, studies on P2P-Cloud only focused on using Cloud to improve DVE data delivery, load balancing, and QoS. Studies that adopt P2P-Cloud architecture and analyze its infrastructure, deployment principle, and cost should be expounded.In the structured DVE, the nodes undergo a resource discovery mechanism to search the desired scene data. Dihydrotestosterone (DHT) is usually used to accurately locate the requested data of the nodes. Therefore, in the structural network, sending a considerable number of requests to locate the resource is not needed with a low network latency and overhead. However, when the nodes frequently join and leave the system, churn can easily occur and cause maintenance difficulty. Classic mechanisms such as SimMud, Colyseus, and PBT were described and compared in terms of advantages and disadvantages. The resources in the unstructured DVE do not necessarily have a relationship to the locations or have a global mechanism similar to DHT to manage the connections, movements, and exits of the nodes. The nodes were used to detect nearby neighbors depending on a mutual notification system. They also send updated information to their neighbors. The neighbor discovery mechanisms pSense and VON were used to illustrate this principle.A data-distributed strategy can expound the mechanism for dividing the labor between nodes to achieve efficient and real-time transmission of the scene data. A comparative review was conducted using pull/push strategy, publish/subscribe mechanism, unicast, and multicast. Several classic DVE systems with P2P architecture were analyzed from the aspects of topological architecture, resource/neighbor discovery mechanism, and data-distributed strategy. The demand of users for online virtual scene roaming has increased with the improved intelligence, power, and popularity of mobile terminals, process ability, and 3G/WiFi, respectively. Package loss would seriously affect mobile devices when receiving virtual scene data because of the limited bandwidth and unstable link of mobile networks. To overcome this problem, this study introduces strategies for lost packet recovery and rate control. This study proposes the transmission of large-scale virtual scenes on mobile networks using scene lightweight pretreatment and P2P-Cloud hybrid architecture.  
      关键词:peer to peer(P2P) overlay network;cloud computing;driver vision enhancement (DVE);scene transmission;mobile internet   
      4510
      |
      272
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56121417 false
      更新时间:2024-05-08
    • Song Jianwen, Qiu Jinming
      Vol. 19, Issue 11, Pages: 1669-1676(2014) DOI: 10.11834/jig.20141115
      摘要:This study proposes a plant leaf model based on the color rough law of the natural environment. This study provides algorithms and theories that can be used for large-scale plant dynamic simulation. The functional S-rough set is used through (,) color rough law and the color coarse rule of the natural environment, which provide a color generation algorithm based on a dynamic color system and plant leaf color reasoning simulation. This study proposes a system modeling theory for dynamic plant leaf colors based on the natural environment. The developed dynamic color system theory is shown as a function of two direct S-rough sets for broad leaves. Leaf color changes with the adjustment of external factors (e.g., light, temperature, and humidity). This adjustment verifies the reliability and validity of the plant leaf rough law theory. In the intelligent simulation system of plant leaves, the dynamic change in leaf color is consistent with the natural law of leaves. This system can be used for large-scale plant dynamic simulation based on the billboard technology. The current algorithm simulates leaves; however, its applications are still limited. For example, the time and space complexities of the algorithm have not yet been analyzed. In addition, the algorithm has yet to be used to simulate uneven spots on leaves. Future studies on grove simulation should include random functions to obtain the desired effect. The light, temperature, and humidity factors considered in this study are only for simple calculations. Normal calculation methods, such as large-scale simulation, should be performed. Simple methods should be included in future studies. This study has shown through simulation that the color of plant leaf surface varies under the influence of scene appearance factors. The results of this study may be used as a basis for the large-scale simulation of forest fire and vegetation emergency management plan design under drought disasters.  
      关键词:natural environment;function S-rough sets;plant leaf;color rough low;intelligent color rough set;reasoning simulation   
      3736
      |
      269
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56120390 false
      更新时间:2024-05-08
    • The quasi-cubic Bézier spirals with three control points

      Gao Hui, Shou Huahao, Miao Yongwei, Wang Liping
      Vol. 19, Issue 11, Pages: 1677-1683(2014) DOI: 10.11834/jig.20141116
      摘要:This study aims to design a simple and effective transition curve. First, three quasi-cubic Bézier curves with monotone decreasing curvature are constructed after studying the base functions. Three other quasi-cubic Bézier curves with monotone increasing curvature are then obtained by parameter symmetry. These new curves have similar properties with the cubic Bézier curves, including endpoint, convex hull, and geometry invariability properties. However, unlike the cubic Bézier curves, the quasi-cubic Bézier curves only have three control points. We then provide strict mathematical proofs in relation to the sufficient conditions of the monotone curvature of these curves. Two of the quasi-cubic Bézier curves covered a broader scope than the cubic Bézier curves with monotone curvature. The specific positional relationship between the quasi-cubic Bézier spiral curves and the cubic Bézier spiral curves were determined. Four of the six quasi-cubic Bézier curves had zero curvature at one endpoint and can therefore be combined into four pairs of S-shaped or C-shaped transition curves for separated circles; the ratio of two radii had no restriction. The remaining two quasi-cubic Bézier curves can be used to form a single transition curve with no curvature extreme for separated circles when the difference of two radii is large. Finally, examples were given to show the effectiveness of these curves. In transition curve design, the quasi-cubic Bézier spiral curve with three control points is more simple and effective than the cubic Bézier spiral curve.  
      关键词:cubic Bézier curve;quasi-cubic Bézier curve;curvature monotony;G;continuity;transition curve   
      4175
      |
      320
      |
      0
      <HTML>
      <L-PDF><Meta-XML>
      <引用本文> <批量引用> 56118849 false
      更新时间:2024-05-08
    0