最新刊期
卷 13 ， 期 1 ， 2008
- 摘要：Abstract：Inverse halftoning has been greatly developed recently and it is one of the key techniques in image processing for halftones. First the principles and implementations of inverse halftoning are given. Then the emphasis is put on the analysis of filtering methods, machine-learning based methods and optimum-estimation based methods for error diffusion halftones. Comparisons show that HVS-based fuzzy filtering schemes have a promising application. Finally the open issues and possible future research directions are pointed out.关键词：Keywords：Inverse halftoning;halftone;error diffusion;filtering;machine learning;optimum estimation3262|242|0更新时间：2024-05-07
- 摘要：Abstract：Image segmentation is a key problem in image processing. Curve evolution, which is very difficult to solve previously,was effectively handled by level set method proposed by Osher and Sethian. At present, there are so many image segmentation methods based on level set, and there are different methods to different images. Now, these methods are continually improved to enhance the speed and the veracity. So in this paper, according to the development of the image segmentation methods based on level set, this algorithm is introduced for readers of different backgrounds in this field to use.关键词：Keywords：level set;image segmentation;active contour model;Mumford-Shah model;multiphase6246|443|0更新时间：2024-05-07
- 摘要：Abstract：In order to effective denoising, this paper proposes a new computationally feasible adaptive noise-smoothing algorithm that considers the local dependency characteristics of images under the complex dual-tree discrete wavelet transform (DWT). The variances of local statistics are estimated with a centered square-shaped window for each coefficient. Previous denoising algorithms tend to smear edges severely in the filtered result. The new denoising algorithm overcomes this problem. Some comparisons with the best available results will be given in order to illustrate the effectiveness of the proposed algorithm.关键词：Keywords：complex dual-tree discrete wavelet transform;image denoising;wavelet coefficients3333|288|0更新时间：2024-05-07
- 摘要：Abstract：In this paper, we studied a variational model in image restoration based on p(x) harmonic mapping and acquired the existence and uniqueness of minimized energy function. Then we educe the corresponding evolution equation. From the experimental results, the proposed model show its advantages of combining anisotropic total variational(TV) diffusion model with Gaussian isotropic diffusion model, since it can not only reduce the “blocky” effect while preserving edges, but also improves computing efficiency.关键词：Keywords：harmonic mapping;TV model;variational model;evolution equation2675|253|0更新时间：2024-05-07
- 摘要：Abstract：Based on the analysis of local characteristic of nature images, a fast and non-iterative inpainting algorithm using adaptive mask is proposed in this paper. First, the direction of isophotes is estimated through sorting. Accordingly, the restore mask can be chosen adaptively. Then, the whole damaged area can be restored along the routine which defined by the fast marching method. The experimental results show that this algorithm has better ability in restoring both smooth area and edge-contained area compared to other fast inpainting algorithms.关键词：Keywords：inpainting;adaptive mask;fast marching method2938|355|0更新时间：2024-05-07
- 摘要：Abstract：In order to effectively protect the copyright of multimedia data, a novel digital color image watermarking algorithm based on singular value decomposition was presented in this paper. The watermark used in the algorithm is 8bits gray-level image which is used extensively in these days and includes more information than the binary image watermark. The scheme first encodes the gray-level watermark into the binary-value bit sequence using the singular value decomposition, and then embeds the encoded watermark into the region with high visual masking effect. In order to access the aim of the watermark’s imperceptions, a method which extracts the blue channel in RGB color space of the host image is adopted. The application of watermark encoding scheme makes our method achieve the goal of watermark blind-extraction and large quantity of vivid information’s hiding. Simulation results show that the proposed method is more robust than traditional methods. The watermarks extracted from the host image after different kinds of attacks that usually encountered in the course of image processing and transmission are still clear and their visual information are nearly lossless.关键词：Keywords：singular value decomposition(SVD);blue channel;similarity(SIM)2720|247|0更新时间：2024-05-07
- 摘要：Abstract：Variable coding modes are used in H.264，which makes the mode selection a time-consuming component. To improve this situation， we first the concept of mode reference aggregate a mode reference aggregate decision algorithm is proposed based on region forecast. This method can significantly reduce the number of candidate mode，thus simplifying the procedure of mode decision. Experimental results show that the algorithm can increase the encoding speed significantly with negligible PSNR drop and bit-rate increase.关键词：Keywords：H.264;video coding;motion estimation;mode decision;NMDA2727|246|0更新时间：2024-05-07
- 摘要：Abstract：Digital Cinema encoding system based on DCI specification has special constraints on the bit-rate and image quality of image sequences. In this paper, it is shown that for a given average bit-rate, the average distortion of the image sequence is minimized, when the image quality is constant. A simple and efficient adaptive rate control method is proposed, which is based on the PCRD-opt(post-compression rate-distortion optimization) rate control algorithm of JPEG2000. Without additional computations, the method could meet the rate and quality constraints. The experiment results show that the proposed method is efficient for different types of image sequences, and it suits for the Digital Cinema encoding system.关键词：Keywords：adaptive rate control;digital cinema;JPEG2000;DCI3013|222|0更新时间：2024-05-07
- 摘要：Abstract：A novel multiple description coding framework, called layered based multiple description coding(LBMDC), is proposed in this paper to meet the requirement of unreliable network transmission. The two-description coding scheme is proposed from SNR scalable layer coding, each description has a base layer and an enhancement layer. The base layer is composed of key information of coefficients and duplicated into both descriptions. The enhancement layer is the rest part of coefficients and split into two parts then sent to separate descriptions, respectively. Thus, the key information of the image is well protected. Even in the condition that one of descriptions is lost, the quality of restructured image is still good. Experimental results have verified the performance of the proposed MDC framework, which improves the ability of error resilient as well as the coding efficiency.关键词：Keywords：multiple description coding(MDC);error resilient;X tree coding2945|225|0更新时间：2024-05-07
- 摘要：Abstract：In order to solve the problem of time consuming in the encoding process of the basic fractal algorithm, a faster method based on the comparison with a preset block is proposed, which is an improvement of the fast fractal image encoding algorithm based on shape feature: the nearest domain block to a range block is found in the sense of comparing with a preset block instead of shape feature. When searching for the best domain block in the vicinity of the nearest domain block, the eight isometric transformations are applied on every domain block instead of only on the best one. At the same time, an error threshold is used to control the searching area around the nearest domain block. The experimental results demonstrate that the proposed algorithm is much faster than the basic fractal algorithm and the quality of decoded image is better than the image processed by shape feature based algorithm in the premise of same encoding time.关键词：Keywords：fractal;image encoding;image compression;preset block2702|248|0更新时间：2024-05-07
- 摘要：Abstract：The concept of “effective area” is introduced in this paper based on detailed analysis of the problems of current active contour models: poor performance under interference, large computational cost, inability to converge to complex contours and difficulty of setting initial contour, etc. The above problems can be solved by setting the initial contour inside the so-called “effective area” and controlling the size of the “effective area”. Setting the initial contour inside the “effective area” is consistent with multi-scale edge detection methods based on wavelet transform. The size of “effective area” can be controlled by the number of GVF field iterations. Based on above discussion, a new contour initialization algorithm is implemented by connecting the initial points of different resolutions obtained using multi-scale edge detection methods based on wavelet transform and suitably controlling the number of GVF field iterations. Experiments show that the initial contour can be accurately set inside the “effective area” and the size of “effective area” can be reduced within 5 pixels from the true edges. The new contour initialization algorithm has also reduced computational cost and can perform reasonably well under interference.2791|224|0更新时间：2024-05-07
- 摘要：Abstract：In structured-light vision system, the accuracy of laser stripe center position is one of the main factors that influences the system accuracy. It analyzes current situation of algorithm for detecting the Center of Structured Light Stripe, and compares advantages and disadvantages between current algorithms. We put forward a new algorithm to detect the Center of Structured Light Stripe combining gradient sharpening with barycenter method,which firstly extracts fringe using gradient sharpening,secondly gets quasi-center of Structured Light Stripe, lastly attains the Center of Structured Light Stripe with barycenter method .Experiment results indicate this algorithm can extract Center of Structured Light Stripe accurately, it is of strong anti-noise capability and its accuracy is at sub-pixel level.关键词：Keywords：image process;center of structured light stripe;gradient sharpening;barycenter method3218|279|0更新时间：2024-05-07
- 摘要：Abstract：In this paper we propose a new ultrasound image segmentation scheme for segmentation of low SNR ultrasound images, which consisted of anisotropic diffusion function and snake model. An improved snake model based on shape similarity was presented. This model is able to change parameters of snake model adaptively according to shape similarity between the snake curve and the prior shape information. Furthermore, edge confidence was introduced into anisotropic diffusion method in order to improve its denoising performance. Experiments show that the proposed scheme not only resolves the segmentation difficulty resulted from low signal-to-noise rate which was instinctive nature of ultrasound images, but also provides a method for choosing parameters adaptively in snake models. Various experimental results for synthesized and real images show that this scheme is promising.关键词：Keywords：snake model;edge confidence;shape similarity;image segmentation;anisotropic diffusion3011|242|0更新时间：2024-05-07
- 摘要：Abstract：A new method of some linear objects tracking such as the runway being changed to tracking of the feature point is presented based on Hough transform during the stable-falling of the sub-muniton. Given all the case of each parameter’s decoupling, the parameters of the line in Hough space have been estimated and predicted with the methods of Kalman filtering. In order to establish the relative position between the runway and sub-munition, some modificative processing in equation for predicting the state of objects’ contrail are conducted combining some data information of the current image when considering the influence of the detector’s spinning with sub-munition. In the instance of small perturbation, a new processing data is established based on the theory of parachute-munition’s dynamic analysis in order to obtain the curve of D0 which would ascertain the relative position of sub-munition and runway. All processing data shows that this method has high efficiency and good stability, and reduces the quantity of calculation to some extent. Experiment results show the error of estimation is small and it can meet the general object requirement.关键词：Keywords：object tracking;the runway;Hough transform;Kalman filtering;small perturbation2643|212|0更新时间：2024-05-07
- 摘要：Abstract：Segmentation of endocardium and epicardium from MR cardiac mage sequences is the most important prerequisite for computer aided cardiac function analysis. This paper proposes a new associated scheme for segmentation of endocardium and epicardium from MR cardiac image sequences. Firstly, segment the endocardium improved GACM, then propagate the segmented contour to other time frames. Secondly, using the segmented endocardium as the initialization contour, the curve is evolved with ad hoc designed regional balloon force controlled by the local mean gray level and the distance from endocardium to get epicardium. Experimental results on clinical ECG gated 4-D cardiac MR images are very close to the outlines medical experts manually. Left ventricle surface model can be reconstructed based on the segmentation results and some important clinical cardiac indices can be calculated.关键词：Keywords：geometric active contour model;image segmentation;level set;cardiac MR2764|228|0更新时间：2024-05-07
- 摘要：Abstract：According to the poor tracking ability adopting static feature model, an adaptive feature generating model based tracking program is present. In this program, the object is valid tracking signal, on the contrary, the background is noise constructing the likelihood maps a local SNR(Signal Noise Ratio) is computed to evaluate the tracking ability of current feature space, and the feature space with maximal SNR is selected as the optimal tracking feature space. Object tracking results based on mean shift demonstrated that the proposed method is more robust and feasible than the classical one.2485|236|0更新时间：2024-05-07
- 摘要：Abstract：Target detection and tracking in infrared image is hard and need to urgently resolve. A novel processing method for dim target detection and tracking in infrared image is proposed. Morphological filters by reconstruction on complete lattices have the fundamental property of simplifying the image while preserving contours. In infrared image, background and clutter can be extracted using reconstruction tophat operator. The proposed method includes intraframe and interframe. Intrafrmae processing uses reconstruction tophat operator to detect target combining with target grayscale, shape and area information. Interframe processing uses spatial and temporal criteria to track target. The experimental results show that the proposed method is efficient and robust to detect and track dim target in infrared images.关键词：Keywords：infrared target;connected operators;complete lattices;reconstruction filter;detect and track2380|210|0更新时间：2024-05-07
- 摘要：Abstract：Based on the characteristics of human vision, an edge-detection algorithm based on masking the image with the spatial activity is presented. Edge-detection algorithms usually deal with the image’s original information without the characteristics of human eyes, more powerful perceiving ability on regular images than the other parts. In this paper, methods for measuring the spatial activity and a double-thresholds mechanism are proposed. The regular extent is represented using the spatial activity and the masking area of the spatial activity is located accurately by correcting gradient direction and determining edge widths. The results show that this algorithm effectively detects edges of regular images and depress the edges of the others. So, this algorithm is more consistent with characteristics of human vision and beneficial to information extraction under complex condition.关键词：Keywords：edge detection;HVS;spatial activity2943|222|0更新时间：2024-05-07
- 摘要：Abstract：This paper presents an improved CLUT color classification method based on linear classifiers and combined color space. The CLUT method is a useful method for color classification. However, the segmentation ability of CLUT is always weakened by the inaccurate choices of color space and threshold, especially when dealing with similar colors. Similar colors always have the almost same distributions in one color space, while have separate distributions in another color space. Combined color space can improve the ability to distinguish similar colors. Linear classifier is one of the most popular methods for pattern classification in pattern recognition. The principle of linear classifier is to use lines to separate the color spaces according to the distributions of different colors. The linear classifiers make it very convenient to set up the table and less depend on the experience of operators. The idea of linear classifiers is applied in this paper to build the CLUT. Meanwhile, HSI and YUV color spaces are employed to increase the ability to segment similar colors. The results of the experimentation show that the combined color space classification method, based on linear classifiers, is efficient and easy to establish look up table and to segment similar colors. The method can be applied to the fast segmentation of color image.关键词：Keywords：CLUT(color look up table);color space;linear classifiers;color classify3200|226|0更新时间：2024-05-07
- 摘要：Abstract：The nearest neighbor convex hull(NNCH) classification algorithm is a kind of nearest neighbor classification method, which takes the approximation errors of the convex hulls of all members of every class to the test point as the discriminant measures. However, the higher computation costs of quadratic optimization problems of the algorithm limit its applications on large data sets. So a sample selection method for NNCH named subclass convex hull growth is proposed in this paper. For one class data, the farthest two points are selected first as the initial chosen set. Then, the distances of others to the convex hull of the chosen set are computed respectively. We choose the farthest one and add it into the chosen set. This procedure is repeated until the end conditions. The convex hull of selected samples is taken as the approximation of all. The more samples are selected the less approximation error is achieved, so the valid estimation of sample distribution is realized. Experiments on the ORL database and the MIT-CBCL face recognition training-synthetic database show the abilities of this method to reduce the training data and accelerate the computation while maintaining the generalization performance of NNCH.关键词：Keywords：samples selection;convex hull;nearest neighbor convex hull(NNCH) classification;subclass convex hull growth2694|190|0更新时间：2024-05-07
- 摘要：Abstract：Aiming at scene image’s character, a method to extract feature point in scene image based on chain-code technique is proposed. Firstly an improved chain-code technique is introduced to represent scene image’s contour. Secondly using the invariability of chain code’s reconstruction, the image’s dominating contour is reconstructed simultaneously the information of noise and detail is wiped off, and then the dominating contour’s chain code is gained using the same method. Then all kinds of features are extracted from the chain code, including the end of border, the center of shape, the point of crossing and main corner point. Finally experiments show that the proposed method is effective for extracting feature point, and has capability of anti-jamming and compressing information.关键词：Keywords：scene image;chain code;reconstructing;extracting feature point3374|222|0更新时间：2024-05-07
- 摘要：Abstract：The storage capacity of the conventional Hopfield network is the 014 times of the number of neurons(P=014N). Because of the huge difficulty in recognizing a large number of images or patterns, researchers are looking for new methods at all times. Quantum neural network (QNN) is a young and outlying science built upon the combination of classical neural network and quantum computing. A Quantum Hopfield neural network (QHNN) whose elements of the storage matrix are distributed in a probability way on the base of quantum linear superposition is presented for speeding up the images recognition and increasing the number of the images recognition. Contrasting to the conventional Hopfield neural network, the storage capacity of the QHNN is increased by a factor of 2N, where N is the number of neurons. Besides, the case analysis and simulation tests have been carried out for the recognition of images in this paper. The result indicates that QHNN can recognize the images or patterns effectively and its working process accords with quantum evolvement process.关键词：Keywords：quantum Hopfied neural network(QHNN);probability distribution;image recognition3490|236|0更新时间：2024-05-07
- 摘要：Abstract：A novel approach to building up an objective function of dynamic matching combined with fuzzy modeling is presented in order to carry out off-line signature verification quickly and accurately. Considering that the projection profiles for feature extraction are very simple, the scheme chooses a suitable membership function to obtain the objective function and dynamic programming is carried out. Then the influence of selecting the reference signature sample and threshold decision is discussed. Some satisfactory experimental results demonstrate the effectiveness of the proposed scheme when compared with other schemes.关键词：Keywords：off-line signature verification;projection profiles;dynamic programming;fuzzy modeling2953|204|0更新时间：2024-05-07
- 摘要：Abstract：In the task-relevant visual attention, the saliency map based on task is required to be built to direct the visual attention. The synergetic perception theory which is similar to the cognition of the human is utilized to research the computational model of task-relevant visual attention. Firstly, the perception of ambiguous model is researched through the synergetic recognition theory. The synergetic visual perception theory is available. The patterns in the synergetic visual perception are corresponding to the visual features in the visual attention model. Then, the bias between the visual features influenced by the task is computed by the property of the bias matrix. The task-relevant visual attention saliency map is built from the bias and the visual features. Finally, a computational model of visual selective attention based on synergetic perception is presented. The algorithm is applied to the visual search task. The validity and the rationality in cognition of the algorithm are demonstrated through the experiments.关键词：Keywords：visual attention;task relevant;synergetic perception;computational model3343|226|0更新时间：2024-05-07
- 摘要：Abstract：Motion perception is one of the important parts of the mind mechanism of biological vision. Based on the fundamental principle of Formotion BCS, and combined with monocular boundary perception processing, this thesis develops a monocular neural dynamic model of motion direction detection. It is different from the Formotion BCS that the mechanism of asymmetrical direction detection is expanded at transient network processing stage, inhibition kernel in the reverse direction is taken at the competition network processing stage and the cooperation mechanism of the motion bipole cells is taken at the boundary segmentation processing stage in the model. We have simulated it on the computer successfully. The model helps us to realize a motion direction detection system based on biovision theory that is different from those normal models in the field of computer vision.关键词：Keywords：motion biovision;direction detection;inhibition kernel in the reverse direction;motion bipole cells2690|232|0更新时间：2024-05-07
- 摘要：Abstract：The vision system of human being is sensitive to local contrast. If the wavelet transform and directional contrast are put together, the effect of image fusion may be better. This paper puts forward a new multi-focus image fusion scheme based on wavelet directional contrast after researching the directional contrast. First, the input multi-focus images are decomposed by wavelet transform. Second, the high frequency neighborhood and low frequency neighborhood are ratioed on each disaggregatire layer of each image. Low frequency of this decomposition is obtained by two-dimensional discrete wavelet inverse transform of last decomposition’s low frequency and high frequency. The bigger ratio one corresponding high frequency wavelet coefficient is taken to corresponding wavelet coefficient. Then two-dimensional discrete wavelet inverse transform of high frequency and low frequency from the maximum decomposition to minimum decomposition is calculated. The fusion image is acquived at last. The neighborhood relativity of pixel is taken into account, the error in selecting pixels is reduced. Experiments show that this scheme is improved than the multi-focus image scheme calculating the wavelet directional contrast of each pixel.关键词：Keywords：image fusion;discrete wavelet transform;wavelet contrast2657|246|0更新时间：2024-05-07
- 摘要：Abstract：For redusing the region boundary extraction algorithm complexity, in this study, Z-map model to binary image with value-range is transformed, then region problem is extracted from binary image. Via scanning Row and run, right and left node of run is connected by its relationship between Row，then the loop is constructed by foregoing connection and extract region boundary. According to the relationship of the run’s connection, resume the region boundary information is resumed completely. The time complexity of extraction algorithm is O(n), while n is the number of run. At last, the relation between the time consuming of the algorithm and the number of run, row and column were provided, while comparison of this arithmetic and before was also given.关键词：Keywords：boundary extraction;run;Z-map;NC machining2863|215|0更新时间：2024-05-07
- 摘要：Abstract：Bresenham algorithm is the most fundamental algorithm for drawing line segments in computer graphics. Canonical Bresenham algorithm can only generate one pixel of a line each time. We proposed a new method by improving it in a novel way, which can generate pixel of a line row by row according to the first and second row of a line and inherit the advantages of Bresenham algorithm without division and decimal fraction. Finally, its efficiency has been proved to be much better than those existing methods.关键词：Keywords：computer graphics;Bresenham algorithm;determinant variable3582|241|0更新时间：2024-05-07
- 摘要：Abstract：Collision detection technology for parametric garment design is static collision detection. It asks for the property of precision, timing and validity of constraint system. The parametric garment model is set into four layers, so different demand for collision detection is put forward to aim at every layer. Collision has an effect on constraint solution, so maintaining validity of constraint system is accomplished by modifying constraint solution sequence or constraint chain. Constructing garment initial mesh with unified topology, subdividing the initial surface to get display mesh by piece-wise technology, precision is improved when collision detection is carried out based on display mesh. Garment style cutting line lying on the display mesh also needs to be projected onto the initial mesh. Speed of collision detection is affected by subdivision, so sub-area and piece-wise technology and all kinds of collision detection techniques are adopted to satisfy the need of timing. At last self collision detection and collision response with some constraints are also discussed. This algorithm is robust and effective, which is shown in the examples, and has been integrated into a garment design system.关键词：Keywords：garment 3D design;parameterization;collision detection;surface subdivision;constraints solution2569|238|0更新时间：2024-05-07
- 摘要：Abstract：To improve modeling ability of subdivision surfaces, interactive shape modification algorithms for C-C subdivision surfaces are discussed. The constraints of points, normal vectors and local isoparametric curves on C-C surfaces, which can be converted into those on control vertices, are specified via setting up local coordinate systems in real-time operation. A global linear system is obtained and the shape of C-C subdivision surfaces can be modified with the various geometric constraints. Two methods based on least-square and energy optimization are presented. The former minimizes the total variation of control vertices. It is fast and suits for local, precise modification. The latter gives the optimal solution to the energy equation with penalty function method and it is a global modification with good fairness. Both methods can be solved explicitly by a pseudo-inverse matrix and are of invertible, commutative, and associative properties, which enhance the efficiency and controllability of the manipulation.关键词：Keywords：C-C subdivision surfaces;shape modification;the least-square;the energy minimization2571|230|0更新时间：2024-05-07
- 摘要：Abstract：Virtual-reality city (VR city) is one of the most important technologies in 3D urban GIS. This paper present how to establish city landscape of three-dimension, generalize the development of general situation and basic principle of city 3D simulation, and analyze two kinds of methods of the establishment of city landscape of three-dimension. Then we introduce the methods of 3D scenes of establishing dynamic simulation and real-time driving, and put forward the function, meaning and development foreground of the city landscape of three-dimension dynamic simulation technology. Finally, we summarize the possibility and direction of development of combining city 3D simulation and GIS, and give an example of city landscape of three-dimension dynamic simulation.2346|251|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