最新刊期
卷 9 ， 期 2 ， 2004
- 摘要：A study has been made on the algorithm of solving optimal set of discriminant vectors in this paper. A concise representation method of between-class scatter matrix and population scatter matrix is proposed theoretically based on theories of blocking matrix and optimization under certain conditions. A new algebraic method of feature extraction is presented. The most obvious advantage of the proposed algorithm is that the computation time decreases drastically. The statement is supported by the numerical simulation experiments on facial database of ORL. The experimental results indicate that high recognition rate can be obtained through the appropriate selection of the dimension of block matrix although there exists nonlinear relationship between recognition rate and dimension of block matrix. The proposed concise representation method of scatter matrix suits for all the applications of pattern recognition using Fisher criteria.关键词：pattern recognition;feature extraction;disciminant analysis;generalized optimal set of discriminant vectors;face recognition2595|218|0更新时间：2024-05-07
- 摘要：In image that results from the projection of depth or orientation discontinuity in physical scene or from the effects of illumination and shading, edges are not in general ideal step edges but more typically a combination more than one type of edge primitive (e.g., steps, pulse and ramp etc). These edge points are termed as composite edges, and composite edges detection and localization accurately is a challenge for researcher in this field. In this paper, a composite edge detector based on multi-wavelet is proposed. Take the zero systematic localization error condition as constraint, the new multi-wavelet function is constructed in spatial domain, at the same time, the scale and wavelet functions are even or odd symmetric. Theoretically, based on this approach ones have the possibility of getting arbitrary precise localization for the edges that are composed of steps and pulse with wavelet transformation. Experimental results with simulated and real gray level images demonstrated the feasibility of our edge detector, it shows that the new multi-wavelet based edge detector has a better edge detection ability than Canny edge detector and Mallat-Zhong wavelet based edge detector.2439|201|0更新时间：2024-05-07
- 摘要：Gradient based edge detection methods are mainly focus on step edges. However local energy based edge detection methods and phase congruency based edge detection methods are effective both to step and peak edges. According to the characters of mammals visual cortex, this paper designs a group of quadrature filters by Hilbert transforms, by which the local energy and phase congruency are calculated. After noise analyzing and frequency distribution analyzing, edges are detected according to phase congruency value. As phase congruency based edge detection method is insensitive to variations in image illumination, and it makes looking for the maxima in local energy become unnecessary, we get good experimental result on the location and accuracy of edge detection.3238|235|0更新时间：2024-05-07
- 摘要：As one of the important applications in Synthetic Aperture Radar(SAR) images, ship wake detection has been received considerable attentions in the area of marine remote sensing. Recently, most researches on ship wake detection depend on the mathematical tool of Radon transform. However, the result of Radon transform is the image with which wake character being enhanced but not the end points of the wakes. In order to calculate the ship's velocity, the aim of the research is to detect the end points of ship wakes. Found on the improvement of the conventional Hough transform, a normalized Hough transform algorithm based on grey level is proposed. It can make the detection result independent of the noise and the different lengths of different lines using the normalization technology. The steps of algorithm are presented. The expression of the wake's end points is derived. On the other hand, the calculation algorithm of ship's moving velocity is presented. Both the time complexity and space complexity of the normalized Hough transform and the conventional Hough transform are analyzed respectively. This algorithm is applied to the ship wake detection in SAR image. The experiments obtained good results and can extract ship's velocity automatically.3920|208|0更新时间：2024-05-07
- 摘要：An effective inpainting algorithm for filling-in of spots and cracks in jacquard images is pro- posed·It first improves the classical Mumford-Shah model by imposing some explicit smooth constraints on the formation of discontinuities·Then, a numerical solution of the improved Mumford-Shah model is implemented by its gradient descent flow which leads to two coupled second order partial differential equations, one for the gray levels and the other for the signature function of discontinuities·Experiments on noisy jacquard images are presented to illustrate the feasibility of the algorithm5011|236|0更新时间：2024-05-07
- 摘要：A new fast fuzzy C-means(FCM) clustering without a priori information about the number of clusters for color image segmentation is proposed to solve the problem of heavy calculating burden and the disadvantage that clustering performance is affected by initial centers for FCM, which is simple and easy to implement in color image segmentation. It uses the hierarchical subtractive clustering(HSC), which could reduce the heavy computation load when clustering a large number of data points, to partition the image data into a certain number of subsets with similar color. For one thing, the centers of the subsets are used to initialize cluster centers; for another, centers of the subsets and the number of points in the neighborhood of centers are used in FCM. The computation speed of the fuzzy clustering algorithm is improved greatly because the number of color image data points used in fuzzy clustering is reduced notably and the computing load of HSC is much less than that of subtractive clustering. Furthermore, it can use the cluster validity index to find the number of clusters quickly. Experiments show that without changing the clustering function, the proposed approach has much faster computation speed than plain FCM algorithm and can segment the color image quickly and effectively.4389|242|0更新时间：2024-05-07
- 摘要：When the images are assayed by the statistics, all most of images are assayed by normal distribution. But in large numbers of experimentation most of the images can' t submit to normal distribution perfectly. Some simulated result is different from the actual images. So it is necessary to assay the images by better distribution. When the color and the intensity of images are assayed, it can be found that the color of the object what people have seen is impacted by plenty of environments, such as the intensity light of the environment、diffused reflection and mirror reflection and so on. Then an object's color isn't symmetrical distribution. So it will turn up larger warp and can't simulate the distribution curve of the image greatly, if the image is assayed by normal distribution. So here the method that simulate curve byχ2distribution is proposed.χ2distribution is one of the methods that are common used in statistics. Sometimesχ2distribution can replace normal distribution. Soχ2 distribution is proposed. And better result of the segmentation has been gotten in the experimentation byχ2distribution.2716|228|0更新时间：2024-05-07
- 摘要：Using leather's texture character to performance intelligent segmentation is an efficient way to improve leather product efficiency and production quality. A new method for leather image segmentation is proposed based on proportional template and region boundary adjustment. First, according to the previous experience of leather segmentation, a proportional template used to performance leather image's coarse segmentation is built, then, the leather image is divided into several coarse regions. Second, pixel on the boundary is selected as the center to design a window, and the mean texture period inside the window is computerized using auto-correlation function distribution. By the same way, the mean texture period of the window belong to the region center is also calculated. Two mean texture period are compared to distinguish whether the location of pixel on the boundary need to adjust. The above processes are repeated until the region's texture consistency is limited to scheduled error scope. Simulations show that the proposed method is simple and efficient to segment leather image.2262|178|0更新时间：2024-05-07
- 摘要：This paper presents a wavelet-domain Hidden Markov Tree(HMT)-based color image superresolution algorithm. Because there exists correlations among the three channels of a RGB color image, a channel by channel superresolution method almost certainly leads to color distortions. In order to solve this problem, first the low-resolution color image is converted into a gray-scale image using the spatially-adaptive approach presented in this paper and the resulting gray-scale image must reflect the human perception of edges in the color image; then by superresolving this gray-scale image, a high-resolution image is obtained; finally, wavelet-domain HMT-based image superresolutions are performed for the three channels of the low-resolution color image using the same posterior state probabilities, which reflect the hidden states of the wavelet coefficients of the high-resolution grayscale image obtained before, and thus the resulting high-resolution color image is what we desired. Because the correlations among the three channels of a RGB color image are considered, there are no color distortions in the reconstructed high-resolution image. Experimental results show that the reconstructed color images have high PSNR and are of high visual quality.2894|345|0更新时间：2024-05-07
- 摘要：In 3D model reconstruction from isometric drawing, it's a key step to reason the hidden elements. This paper proposes an algorithm to reason the hidden elements of the isometric drawing based on connected edges. Reason the hidden information is to reason through analysing these vertexes that have hidden edges, reason the direction and the length of connected edges and the edges that go through connected vertexes. After reasoning, we get some possible vertexes. Validating them with superposition and the edge circuits, we get the correct hidden element's coordinates and those circuits information.Then a method is introduced to seek the lost depth information based on the vertex's coordinates and those circuits information of the edges, its matrix is also brought forward. A prototype system To3D is developed using C and OpenGL tools to validate this method, the test result proves it's effective to reconstruct plan solid.关键词：computer vision;3d reconstruction;distinguish from sketch;isometric drawing2716|208|0更新时间：2024-05-07
- 摘要：One of the most challenging work in computer art research field is to studies of the simulation of Chinese ink wash drawing. On the basis of the graphical analysis to the sensible art effects of ink wash drawing, this paper extracts the typical artistic effects which can represent the characters of Chinese ink wash drawing. Then we set up a brush model, and digitize and standardize it. In order to produce the interactive controllable and structured ink wash drawing effects with simple input instrument shcu as the mouse、keyboard and electronic pen, we initialize different data structure of brush to simulate different brushworks. In order to realize the daedal effects of the real brush, we change the state of the brush continually during the course of drawing. In this paper, more attention is focused on the simulation of special effects such as side edge of brush, and finally we realize a fine interactive system. The experiment results indicate that this algorithm simulates the Chinese ink wash brush successfully.2654|184|0更新时间：2024-05-07
- 摘要：Research on image quality assessment is meaningful for image processing projects. Since human being is the final receiver of the image, the key point of the image assessment is that it should match the characteristics of HVS(human visual system). In this paper, a novel image quality assessment according with perceptual property of human eye is proposed. In this algorithm, wavelet transform is used because it matches well with the multi-channel model of HVS, bandpass property of CSF(contrast sensitivity function) is integrated with, and the complexity of the computation is considered. The simulation results show that the correlation coefficient between the algorithm and subjective MOS(mean opinion score) is 0 95, but the correlation coefficient obtained by the PSNR(Peak signal noise ratio) measure is 0.81.4655|315|0更新时间：2024-05-07
- 摘要：With the technology of image compression and codec booming, and in order to control the transmission of their bit-stream, the term of image quality is given a kind of new degree and challenge. Today the research of Image Quality Assessment has also become one of the basis technologies in the realm of images and information. This paper analyzed the characters of Human Visual System (HVS) at first, that included the visual nonlinear, multi-channel and masking effect as well, then constituted a kind of mathematic model to simulate the course that the human visual system proceeds the image information. In the model, original images and distorted images are transformed respectively to the perceptual field, that is, both of them are processed by the basic visual nonlinear firstly, and with that enter a band pass filter created by the modulate transform function. During the processing of the multi-channel in the HVS model, according to the characteristic of masking, the function of band pass filter is divided into five bands by the varying spatial frequency of initial image. Finally the value comes into being that we gained from the vision pathways, and then the HVS-E (Error) is obtained. The result from experiments indicates the fine agreement between the HVS-E and the MOS (mean opinion score).4331|193|0更新时间：2024-05-07
- 摘要：In many image applications, the coding technology based on Region of Interest(ROI) is very important. With ROI coding technology, the interesting region in image obtains better quality than the region of the background(ROB) at a low compression rate, and the ROB obtains a low level quality at a high compression rate. So the technology solves the conflict between the image quality and the compression rate. General scaling based method and maxshift method are used as the ROI coding technologies in JPEG2000, but there are still limitations in some applications. In this paper, a Multi-Bitplane Interleaving(MBI) method is introduced. The method arranges the bitpalne effectively by using ROI parameter and bitplane parameter. The method allows for arbitrarily shaped ROI coding without transmitting any shape information explicitly to the decoder. And it allows for more flexibility to control the image quality of ROI and ROB. ROIs coding is also allowed for. Because of the characters such as ROIs coding, significant coding of ROI and ROB, the MBI method can greatly meet the coding requirement in different applications.2840|200|0更新时间：2024-05-07
- 摘要：A novel optimization method using partially evolved Hopfield neural network is proposed in this paper. The method uses Hopfield neural networks and a genetic algorithm on a local area of Hopfield neural networks to compensate each other for defects. The defect of the Hopfield neural network is captured by locally optimal solutions. The defect of genetic algorithms is the lower convergence speed when it optimizes large scale problems. In the proposed method, the Hopfield neural network and a genetic algorithm are used alternately. Solutions obtained with the converged Hopfield neural network are applied to the genetic algorithm to escape from locally optimal solutions. The genetic algorithm is only carried out on some local areas of Hopfield neural network so as to effectively save the computational consumption. The method is evaluated by investigating two large scale optimization problems: image segmentation and 200 cities TSP problem. Experiments show that the local minima of large scale networks can be greatly improved by the partially evolved Hopfield network and the convergence speed is obviously enhanced.2938|284|0更新时间：2024-05-07
- 摘要：Modern communication standards, such as Digital Imaging and Communication in Medicine (DICOM), include non-image data for a standardized description of study, patient or technical parameters. However, these tags are rather roughly structured, ambiguous , and often optional. In this paper, we present a novel monohierarchical muti-axiel classification code for medical image retrieval. Our so-called MOAB coding scheme consists of four axes with three to four positions, each in {0,…,9,A,…,Z}. In particular, the modality code (M) describes imaging modality and relevant technical detail, the orientation code (O) models examined body orientation, the anatomy code (A) refers to the body region examined, and the biology code (B) describes the biological system examined. So far, the coding scheme is easily used in a medical image database of about 7000 single digital images, the loading time for 256×256 images was short, and the error rate of query result was low. The MOAB classification coding scheme enables a unique classification of medical images so as to develop further content-based medical image retrieval. The code is flexible and easily to be extended.3249|217|0更新时间：2024-05-07
- 摘要：In this paper, a new medical image retrieval approach based on low level features and semantic features is proposed. The low level features include gray, moment and texture features, which are selected by genetic algorithm. These features can't express the human's understanding of the images. Directly using these features can't get satisfying results, so the semantic features are needed. The image describing in the image report by doctors are chosen for semantic content. Experiment results show that the retrieval result by low features and semantic features are better than only by low features.2941|295|0更新时间：2024-05-07
- 摘要：The model of state balance of image pixels from the view of pixels' correlative degree is proposed to extract vascular network from DSA images. The image is regarded as a balanced system that consists of object and background areas. Under some certain force, there is an inner balanced state between these two kinds of areas. But the introduction of noise breaks the balance, this disturbance will make the boundary of vascular and background indistinct. In the circumstances, extracting vascular network directly is very difficulty. If the balanced state can be gained, the segmentation becomes easy and accurate relatively. Therefore it's possible to remove the noise and separate the object and background areas by resuming such balanced state. Based on this theory, a new binarization algorithm for gray images can be developed and used in the vessel extraction from the DSA images of cerebral vessels. An overlap algorithm is presented to resumes the balance state in this paper. Then using this algorithm, the vascular network can be segmented from background perfectly. It can extract the whole vascular network from the DSA subtraction images with high level noises and the experimental results are very satisfying.2840|213|0更新时间：2024-05-07
- 摘要：Cytoskeletal distortion was hard to describe straightly with some quantities due to the complexity of cytoskeletal structure and the morphological uncertainty. For describing the morphological modification of cytoskeleton under simulated microgravity and estimating the confront effect of drugs on cells under simulated microgravity, this paper focuses on get the characteristic parameter that quantify the morphological change of cytoskeletal images. Fractal analysis was applied here to quantify the morphological variation of the cytoskeleton under simulated microgravity. This paper used the methods of covering blankets and box-dimension estimation to count the image complexity for microtubules and microfilaments. To study the medical effect of quercetin, a kind of flavonoids, which could counterwork to effects of microgravity on cells, cytoskeleton image of cells treated by quercetin under simulated microgravity were also analysed. The image complexity of cytoskeleton was decreased under simulated microgravity, but it was partly inhibited by quercetin. The present results demonstrate that fractal dimension could show the texture change of cytoskeletal image under simulated microgravity, and it would be a promising approach to quantitative analysis of the morphological modification of cytoskeleton in different environments.2448|195|0更新时间：2024-05-07
- 摘要：Medical image 3D reconstruct is an important application filed for volume rendering, for it special using, it required fast interactive speed and high image quality. Ray casting is a widely used basic volume rendering algorithm. It can get high quality image but the rendering speed is very slowly. In order to improve its rendering speed and apply it to medical image 3D reconstruct, a synthesized accelerate algorithm is proposed in this paper. Firstly, accelerate algorithms for ray casting are fully studied and compared. Secondly, proximity clouds algorithm has been selected and extended to continuous ray casting. Then the synthesized accelerate algorithm base on ray coherence has been proposed. Finally, the algorithm is realized on PC platform. Four data sets are used for test at two different image resolutions. The result shown that compare with standard ray casting algorithm, the accelerate algorithm not only improve the rendering speed for nearly ten times but also bring no depress on image quality. At 512×512 image resolution, a nearly real time interactive speed has been get. It gives a more capability method for the application of medical image 3D reconstruct.3965|219|0更新时间：2024-05-07
- 摘要：Polygonal vertices convexo-concave determining is one of the fundamental issues in Computer Graphics. In this paper, plane space is separated into 2 region defined as positive region and negative region based on the direction and the stretched of the line segment. Accordingly transform identification of convexity or concavity of the next vertices into identification of the region position. Consequently the operation of multiplication can be averted. In order to do the best in practical determining, 3 areas is separated of the plane space, whereas not the same as earlier. These area are based on two horizontal lines, two vertical lines and two 45°diagonal lines through the vertices of the line segment. Because the partition is based on the polygon vertices and it has dynamic, therefore it's a dynamic distinct partition. These areas are defined as positive area, negative area or mixed area. In the case of the positive area or the negative area, we can confirm the convexity or concavity of the vertices through simple distinguishing. The result of test indicates it is a simple, highly efficient and reliable algorithm with high speed requires only (n/4) multiplication and(12n) comparisons.3145|209|0更新时间：2024-05-07
- 摘要：Measured data are obtained through a laser scanner in reverse engineering. The real data inevitably contain unreasonably noise error during measuring. The noise error causes the reconstructed curve and surface rough. Therefore it is essential to remove the noise error. This paper investigates the method on reducing noise error of the measured data obtained through laser line scanning. The method on reducing noise error is closely related to the organization of the point cloud data. This paper analyzes the mathematical model about the point cloud data error. The noise error is mainly caused by random error. The characteristic of noise error is that the swing value is bigger and the peak arises on the scanning line. According to this feature, a method named the random filter algorithm is put forward for reducing noise error, and it is simple, quick and practical. The procedure of this algorithm is first to compare the relative position among the successive points. Then the points that their positions oscillate bigger are judged noise error according to a threshold and will be removed. The principle and the step are described in detail, and it is proved by some examples that the processing result of the method is effective and can meet the requirements of curve and surface reconstruction.2811|225|0更新时间：2024-05-07
- 摘要：Steganalysis is a technique of finding, distilling and destroying the hiding information in stego-cover. The methods of finding the hiding information can be divided into two kinds. One finds the secret information by contrasting the original cover and stego-cover, and the other,named blind detection does not have the originalcover for contrasting. The original cover for secret communication usually can' t be obtained, so the blind detection is widely adopted. In order to give an outline of the research work of steganalysis and compare the advantages and disadvantages of various methods available now, some analysis methods are introduced. The discrimination function test,χ2-test andRS-test are analyzed from both principle and experiment results, and comparisons of the three methods above are also given in certain aspects, such as the applicability and effectiveness of tests. The thresholds ofRS-test are selected at the condition of minimizing the ratio of missed detection while keeping the number of false accusations reasonable.3547|232|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