最新刊期
卷 16 ， 期 4 ， 2011
- 摘要：Scale problem is one of the central issues in land cover mapping from remote sensing imagery, and downscaling land cover data is a difficulty in this domain. Soft classification can provide more information than hard classification at pixel level. However, the spatial location of land cover compositions within each pixel is unknown. To solve this problem, many super-resolution mapping methods have been developed in recent years. In this paper, a review on recent development of super-resolution mapping methods is presented. The review focuses on the research status, major algorithms and accuracy assessment methods of super-resolution mapping. The main drawbacks, research challenges and future directions of super-resolution mapping are discussed.关键词：super-resolution;sub-pixel;mapping;classification;scaling3309|229|0更新时间：2024-05-07
- 摘要：In this paper, firstly a new method for color image segmentation is proposed. The segmentation algorithm can segment an image into several regions with high visual consistency. The segmented results can simulate the feelings of human observing images, like an image have several regions with same colors or texture. After obtaining consistent regions, some image region properties like color coding, connectivity coefficient, and area ratio are then extracted as descriptors which can be used for content-based image retrieval. The color coding was obtained by quantification the pixels in the HSI color space. The experimental results shows that the proposed method can not only capture the visual impression of an image, but also have a good performance in image retrieval.关键词：visual consistency regions;image segmented;connectivity coefficient;image retrieval3180|324|0更新时间：2024-05-07
- 摘要：Because of the network’s time-varying and heterogeneity, and also the high probability of packet loss in congested network, transmitting media streaming over internet by TCP is a preferred scheme, which can improve the media streaming QoS. But due to the TCP retransmission for error or time out packets, in the congested network, it is difficult to provide real time transmission for high bit rate media streaming. This paper proposes an encoding rate adaptive media streaming transmitting algorithm (TCP_RA) based on TCP. It adjusts the media streaming encoding rate according to the difference between reading and writing points in the application layer buffer, in order to suit the network bandwidth’ variety. It compares the difference for rate adaptive media streaming QoS improvement between TFRC and TCP_RA by simulation in NS2, simulation results show that TCP_RA can effectively improve the media streaming QoS in bad network environment. The algorithm is easy to implement, so it’s worth of popularizing关键词：streaming media transmission;rate adaptive;TCP protocol3875|249|0更新时间：2024-05-07
- 摘要：Aim at the degradation of single image under haze conditions, this paper proposes a new algorithm of automatic haze removal. It first transforms the haze image from RGB to YCbCr color space, and uses multi-scale Retinex algorithm on the luminance component, then conduct a subtraction operation with the control of image quality assessment. After median filter on the image, the transmission map is obtained and the haze removal image can be obtained as well. Experiments show that the algorithm can effectively improve the degradation of image and enhance its quality.关键词：haze;haze removal;luminance component;Retinex algorithm;image quality assessment3684|314|0更新时间：2024-05-07
- 摘要：The document image incorporating with image enhancement method can achieve a good perceptual quality, which is necessary for the document image information system. Contrast-limited Adaptive Histogram Equalization (CLAHE) is widely utilized to expand the global dynamic contrast range, and Lee filter is usually used to amplify the local details. A detail-amplifying CLAHE (DAHE) method is proposed by combining the above two methods. Firstly, a noise-resist policy is employed to differentiate the detail and the noise. Secondly, a flexibility amplifying mechanism is proposed to make the amplified image suitable for the human vision system. The proposed method can not only balance the global gray-level dynamic range, but also enhance the local details. The experimental results demonstrate that the proposed method can both enhance the document image, which is suitable for the document image information system, and satisfy the requirements of the national industry standard.关键词：document image;image enhancement;contrast-limited adaptive histogram equalization;Lee filter3335|226|0更新时间：2024-05-07
- 摘要：Owning to the visual inconsistency of the greedy synthesis algorithm and neglecting of structure information of other optimization based inpainting, a new image inpainting algorithm using structure and texture optimization was proposed. The image inpainting was formulated as minimization of a weighted energy function, which was optimized using an expectation maximization(EM)-like algorithm. To ensure damaged regions with strong structure and high confidence been given priority to restoration, a greedy approach was used to set initial values and to calculate weight values for the EM algorithm. Compared with greedy synthesis and optimization based image inpainting approaches, the proposed algorithm considers the structure information and achieves better repairing results.关键词：image restoration;image completion;ancient painting restoration;quality assessment3276|287|0更新时间：2024-05-07
- 摘要：Based on partial differential equations (PDE), we propose a novel image magnification approach which consists of oriented diffusion, fidelity term and shock filter. In a zoomed image, the blurring can be adjusted and the features be preserved by setting the coefficient of the fidelity term related to edges. The weighting of oriented diffusion and shock filter can be balanced through continuous function based on image content. Thus we adaptively combine the two kinds of PDE with minimum artifacts. Experiments demonstrate that the proposed adaptive coupling approach can achieve better result than traditional interpolation methods and other similar methods.关键词：image magnification;PDE;oriented diffusion;shock filter3338|165|0更新时间：2024-05-07
- 摘要：In order to solve the multi-focus problem, which is a difficult issue for image retargeting based on existing warping methods, a fisheye transformation method is proposed in this paper. The method emphasizes areas of interest while keeping the background information and it can solve the multi-focus problem in image retargeting. For a multi-focus image, the distances among focuses are calculated first, and then the transformation method is selected. According to a predetermined threshold, the selected transformation method can be full-area fisheye transformation after splitting the image or orthogonal fisheye transformation. The visual effect of target images and users’ subjective satisfaction showed that our method is a viable solution for multi-focus image retargeting.关键词：image retargeting;fisheye transformation;multi-focus image;focus conflict3297|255|0更新时间：2024-05-07
- 摘要：Chan-Vese model, which has better ability to handle the blurry boundary and complex topological structures in images, has been widely used in image segmentations. However, the effect on segmentation in the images with intensity inhomogeneity and multiple-objects is less satisfying. Fuzzy c-means clustering(FCM) algorithm works as an unsupervised classification method has been applied in object identification and image segmentation. Nevertheless, it is sensitive to noise because of taking no account on the spatial information. Arming at these problems, a multiphase CV model integrated with improved FCM algorithm is proposed. First, the classes of the intensity are calculated based on the histogram statistics, and the spatial information computed in the neighborhood revise the grade of membership. The improved FCM algorithm applied with the region fitting term of CV model, working as the reliance of evolving the level-set curve. Anisotropic local template is then used to handle the different objects so as to control the split-up of the contour accurately and segment more objects in less time.关键词：Chan-Vese model;FCM algorithm;anisotropic;noise;multiphase3248|204|0更新时间：2024-05-07
- 摘要：Separation individual tooth from three-dimensional dental model is an important step in computer aided orthodontics treatment. Because of different shapes of every tooth and huge difference among different people, it is difficult to partition individual tooth automatically. Therefore, this paper presents an interactive segmentation algorithm, whose process is firstly picking up feature points on triangle mesh of dental model interactively, then approximating the geodesic path between feature points and finally partitioning individual tooth from dental model by region grow algorithm once the geodesic path is closed. The experiment indicates that our algorithm runs fast and needs little interactive operation, while the border of the segmented tooth is smooth and meets the clinical requirement of orthodontics.关键词：geodesic path;teeth segmentation;region grow;triangle mesh3405|192|0更新时间：2024-05-07
- 摘要：An unsupervised two-phase texture segmentation method based on active contour was proposed. First, the texture feature extraction approach based on semi-local image information was analyzed, which revealed that it could not represent texture’s orientation information. In order to segment texture images containing periodic and orientational character, a four-channel texture feature was achieved combing semi-local image information with nonlinear structure tensor. Then Gaussian mixture model was adopted to describe the probability density function of the features. Numerical algorithms were based on split Bregman method. Experimental results for both nature and synthetic texture images show that our method could cope with complex segmentation tasks. Meanwhile it is effective and unsupervised.关键词：image segmentation;semi-local image information;structure tensor;level set;split Bregman method4437|284|0更新时间：2024-05-07
- 摘要：The application of classical Gaussian mixture model to image segmentation has highly computational complexity. A image segmentation method based on Gaussian mixture model with multi-sampling is proposed in order to solve this problem. First, the sampling theorem is given and proved,and the minimum sample size is derived according to the smallest cluster and cluster number. Second, a penalty function, which is to judge the good sample, has been designed to eliminate the error of clustering model,and image pixels are sampled based on the minimum sample size to be clustered according to Gaussian mixture model. Finally, by the means of the definition on the distance between a pixel point and the categories, the remaining points is assigned respective cluster depending on the principles of the nearest distance. The experimental results show the effectiveness of the algorithm.关键词：image segmentation;Gaussian mixture model;muti-sampling3005|269|0更新时间：2024-05-07
- 摘要：The paper proposes a segmentation and tracking method based on adaptively weighted Markov random field (MRF).By using spatial relativity of the adjacent pixel regions, this method adaptively updates the system energy function parameter β, and makes object segmentation more accurate. On this condition, the fragments-based integral histogram feature template is established. And Kalman prediction method is combined with object motion information to improve the algorithm’s accuracy and implement matching and tracking. Experiment results proved that this algorithm can accurately settle with the problem of objects segmentation and tracking problem under the case of partial shelter,and illumination change.关键词：moving object tracking;MRF segmentation;adaptive weight;fragment template;Kalman prediction;integral histogram3045|233|0更新时间：2024-05-07
- 摘要：Obtaining the type of a building automatically is significant for 3D modeling and image retrieval, but it is very difficult. In this paper, a new automatic urban building classification algorithm is proposed. This work analyzes the corner characteristic of common city buildings, applies morphological sieves of large scale to obtain rough contours. Then it uses a new segment and corner optimization process to count all kinds of corners, and realizes the automatic distinguish of flat roof buildings and non-flat buildings. The tests of dozens of images demonstrate the validity and effectiveness of the proposed segments optimization approach. The results indicate that it can achieve over 80% accuracy on type judgment with the presented method.关键词：Tmage classification;contour extraction;corner detection;Hough3313|347|0更新时间：2024-05-07
- 摘要：Cotton impurity inspection is important to control the quality of fabric and reduce production costs. To detect cotton impurity in industrial environment with uneven illumination, an impurity detection algorithm based on Gabor filter is proposed. In the algorithm, the image is divided into multiple zones using Otsu’s threshold method and morphological filter, and the texture features for foreground and background are then extracted using Gabor filter. An adaptive threshold segmentation method is designed and applied to the Gabor filter output, and then the impurities in the image are detected using a morphological filter and connected-zone analysis. Experiment results show that the proposed algorithm can remove the undesired interference caused by light source fluctuations effectively, and the common impurities in cotton also can be accurately detected.关键词：cotton impurity inspection;Gabor filter;adaptive threshold3148|267|0更新时间：2024-05-07
- 摘要：An inspection model for surface defects of copper strips based on vision bionics is proposed in this paper. The model firstly uses Gaussian pyramid and Gabor filter to extract the features of defective images, and combines the features into a saliency map. Then, it simulates the bottom-up attention mechanism, analyzing the content of fovea to acquire “what” information and obtaining a time-ordered sequence of visited regions after simulation of saccades, which constitutes the “where” stream. At last, it uses a discrete and observable Markov model to adjust the probabilities of a single Markov chain according to “what” and “where” stream, and maximizes the likelihood of the training data, and realizes correct classification. The experimental results demonstrate the feasibility and effectiveness of the proposed method in surface defect inspection, furthermore, the average accuracy rate can reach 94.40%.关键词：defect inspection;vision bionics;selective attention;Markov model4120|233|0更新时间：2024-05-07
- 摘要：Scanning infrared imagery has the characteristics of giant data, complex background, heavy clutter and low SNR. A real-time processing method is proposed to detect targets on all-azimuth scanning infrared imagery. The proposed method firstly suppresses the clutter and backgrounds to enhance the target by preprocess. Then it adopts two level feature extraction to dramatically decrease the image data. At last according to the idea of inter-frame image gray-level dependencies, accomplishes the target detection of scanning infrared images. Image processing card is designed based on DSP&FPGA processor with the advantage of real-time data processing and data transmission. The results show that the method is efficient and viable, the image processing card works reliably and stably, which can satisfy with the need of real-time target detection of scanning infrared images.关键词：scanning infrared images;target detection;real-time image processing3032|226|0更新时间：2024-05-07
- 摘要：The cameras in this study are used to extend view field. The paper proposed a three-dimensional information based human tracking method. The method firstly samples the pixels within the human body in one of the video images, and then finds the matching points in the other simultaneous video images by polar line constraints. The three-dimensional points in space corresponding to each pair of the matches are obtained by triangulation and are also clustered. The pixels in the image plane corresponding to each clustering region which includes the three-dimensional points are decided. Finally, Gaussian-smoothing histograms are created based on the pixels in the image plane. Occluded people would be divided into each other by the Gaussian-smoothing histograms. The matching relation of each person in different cameras could be found and thus the human tracking is realized. Experimental results show the efficiency of the method.关键词：human tracking;video surveillance;occlusion;Gaussian-smoothing histograms3302|248|0更新时间：2024-05-07
- 摘要：Transferring color between images is a method that uses one image’s color from another. This paper presents a fast auto color transferring method. Fuzzy clustering algorithm produces a certain number of sample patches with different hue characteristics. Then this method creates correspondence by computing the distance of feature vectors between sample patches in the reference and target images. The local color information is transferred from the matched sample patch in the reference image to the target one in ιαβ space. The weight clustering of effective hue cuts down the number of sample immensely. Therefore it improves the speed of color transfer significantly, and in the meantime does not damage the transfer quality.关键词：color transfer;fuzzy clustering;color space3944|264|0更新时间：2024-05-07
- 摘要：An uncorrelated kernel extension of graph embedding which provides a unified method for computing all kinds of uncorrelated kernel dimensionality reduction algorithms is proposed. Compared with kernel dimensionality reduction methods, the proposed method is better in terms of reducing or eliminating the statistical correlation between features and improving the recognition rate. The experimental results on ORL, YALE and FERET face databases show that the proposed uncorrelated kernel extension of graph embedding method is better than other methods in terms of recognition rate. Besides, the relation between uncorrelated kernel extension of graph embedding and kernel extension of graph embedding is revealed.关键词：optimal discriminant vectors;statistically uncorrelation;kernel extension of graph embedding (KGE)3460|172|0更新时间：2024-05-07
- 摘要：A gait recognition method based on spatiotemporal feature extraction is proposed. Recursive subspace learning is used to extract both time and space feature of gait. In the first subspace learning, the periodic dynamic feature of gait is extracted by principal component analysis and sequence data is represented in the periodicity feature vector form. In the second subspace learning, principal component analysis plus linear discriminant analysis are applied to the oeriodicity feature vector representation of gait and sequence data is compressed into gait feature vector. gait feature vector is an effective representation because it contains both human dynamic and shape feature. Experimental result on the USF gait database shows that the proposed method achieves highly competitive performance with respect to other published gait recognition approaches.关键词：gait recognition;subspace learning;principal component analysis;linear discriminant analysis3308|208|0更新时间：2024-05-07
- 摘要：A new image-based algorithm named IEPVH is proposed to perform exact polyhedral visual hull reconstruction in an incremental fashion. First, edges of visual hull model are chopped by viewing cone of the new view. Second, local orientation information of intersection vertices on edges of visual hull model are retrieved and at the same time intersection vertices on edges of viewing cone are added into the mesh. Third, local orientation information of intersection vertices on the edges of viewing cone are retrieved. Finally, faces of viewing hull are identified by traversing edges of the new model and then triangulated for rendering. IEPVH enables interactions with users during the computation. Moreover, IEPVH has the merit of less memory consumption. Modeling results are presented to demonstrate IEPVH’s robustness and effectiveness. IEPVH’s merits make it preferred by mobile device.关键词：image-based;3D modeling;silhouette;visual hull;polyhedral mesh;incremental processing3261|196|0更新时间：2024-05-07
- 摘要：Point Pattern Matching (PPM) is an important issue of computer vision and pattern recognition, which is widely used in target recognition, medical and remote image registration, pose estimation, etc. This paper proposes a particle swarm optimization (PSO) based approach for pose and correspondences estimation between the feature points of two images under affine transformation. In the method, the point sets matching problem is formulated as an objective function’s optimization problem in the affine transformation parameters solution space. The PSO is used to search for optimal transformation parameters. There are three contributions made in this paper. Firstly, we develop an initial transformation parameters estimation method for PSO, which greatly improve the algorithms efficiency and veracity. Secondly, we introduce a threshold to correspondence finding, which rejects outliers and enhances veracity while using “Nearest Neighbors Search”. Thirdly, we propose two approaches to improve the searching efficiency when using the original PSO. Experiments demonstrate the validity and robustness of the algorithm.关键词：computer vision;pose estimation;point pattern matching;particle swarm optimization3484|229|0更新时间：2024-05-07
- 摘要：For medical image registration of good robustness, high-accuracy and speed requirements, this paper proposes a DoG(difference of Gaussian) keypoints image registration algorithm based on Rényi entropy. This algorithm extracts DoG key points from images, then incorporates grey scale information of the key point into the joint Rényi entropy, and estimates joint Rényi entropy directly using minimum spanning tree. The new algorithm combines the robustness of DoG key points and the high speed of Rényi entropy estimated by the minimum spanning tree. Experimental results show that in the images with noise, non-uniform intensity and large scope of the initial misalignment case, the algorithm achieves better robustness and higher speed while maintaining good registration accuracy.关键词：medical image registration;DoG keypoints;minimum spanning tree(MST);Rényi entropy3311|223|0更新时间：2024-05-07
- 摘要：In this paper, a SIFT feature based fundus image automatic mosaic algorithm is presented, according to the characteristics of low contrast and the geometric distortion between different fields of the view of the fundus image. The SIFT features of fundus images are extracted and described using vector to determine the feature points matching between two images, and outlier points are removed using MLESAC algorithm. A distance-slope similarity measure method is put forward to purify feature points, then the perspective transformation matrix is computed according to matching points between images, and image registration and image mosaic is implemented finally. The mosaic results of multiple images obtained by fundus camera show that the algorithm has good robustness and stability, and high-precision automatic fundus image mosaic can be achieved.关键词：SIFT;image registration;image fusion;images mosaic;fundus images;distance-slope similarity measure4114|298|0更新时间：2024-05-07
- 摘要：DMSP/OLS nighttime light data is effective and economical for mapping urban land on a large scale. Based on conventional dichotomy method, Layered Threshold Method (LTM) was developed to extract the urban land using the DMSP/OLS stable nighttime light data. The administrative regions with close thresholds were divided into the same set by LTM continually. The process was ended until the thresholds in one set were selfsame. The urban land of Chinese mainland in 2002 was extracted based on LTM using the DMSP/OLS stable nighttime light data. Compared to the traditional inefficient dichotomy method, whose efficiency depended on the amount of administrative regions, LTM searched the thresholds of the sets instead of each administrative region, optimizing the previous method. The performance of the LTM was evaluated to be credible and valuable according to statistical data and Landsat ETM+ data.关键词：layered threshold method;dichotomy method;DMSP/OLS;urban land3629|575|0更新时间：2024-05-07
- 摘要：Most CFAR detectors need prior information on interfering targets in a multi-target environment, and hence can not keep stable detection performance when the detection environment changes. A new adaptive constant false alarm rate (CFAR) detector，referred as stepwise cumulation CA (SCCA) CFAR detector, is presented for target detection in a multi-target environment for SAR imagery. By employing a cell-to-cell criterion for accepting or rejecting reference samples according to an adaptive threshold, that is, clutter power estimation and standard variance which are independent of interfering targets, the samples from interfering targets are censored and the clutter samples are accumulated stepwisely. The final estimate of the noise level in the cell under test is formed using cell-averaging method with the accumulated homogeneous clutter samples. The detection performance and runtime of SCCA-statistics-based two-parameter CFAR detector are tested and compared with CA-CFAR and OS-CFAR by simulation, and the result shows the proposed detector increases the detection probability, and its runtime approximates that of OS-CFAR.关键词：SAR;imagery;target detection;CFAR3407|406|0更新时间：2024-05-07
- 摘要：A robust natural feature based parallel registration method is proposed. The method uses two threads for registration on a dual-core computer, one thread using KLT for tracking, while the other using Surf to reset the accumulated-errors. Delayed-homography is proposed to solve the problem of thread Synchronization. At the same time, the idea of key-frame, which is defined in this paper, increased the accuracy and stability of the registration. The experiment result proved that the multi-thread method solved the problem of accumulated-errors, and it is a robust feature based real-time registration method.关键词：augmented reality;parallel registration;thread synchronization3840|226|0更新时间：2024-05-07
- 摘要：Volumetric visual hull reconstruction is easy to be implemented without complicated geometrical computation, but its accuracy and efficiency are still unsatisfactory. A parallel optimized visual hull modeling method is proposed based on CUDA and marching cubes. It parallelizes the voxel intersection judgment procedure and the isosurface extraction procedure, and implements both of them using multiple threads of CUDA to accelerate the reconstruction. During marching cubes process, exact intersection points are calculated and redundant computation of edges is eliminated. The pixel shader is used to carry out the hybrid weight texture mapping so as to enhance the quality of the reconstructed mesh model. Experimental results on the multi-camera platform DreamWorld have proved that this method can reconstruct multiple dynamic objects accurately in real time.关键词：visual hull;real time modeling;computer unified device architecture (CUDA);marching cubes;voxel3264|206|0更新时间：2024-05-07
- Postal code：100190
- Tel：010-58887035/58887030/58887418 Email：jig@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备05080539号-4
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0