最新刊期
卷 7 ， 期 10 ， 2002
- 摘要：Data fusion is a powerful tool to process tremendous remote sensing data for its capability to combine useful information of multisource or multitemporal data, and eliminate their redundancy, so it has been applied on many areas extensively, especially, on remote sensing image processing. At first, this paper simply reviews the necessity of data fusion's appearance and development, together with several definitions brought forward by several investigators. Afterwards, this paper not only attempts to analyze the Data Fusion Engine framework and configuration comprising of pixels-level?feature-level and decision-level fusions detailedly, but also principally analyze the fusion methods and their applications of each data fusion level. In addition, because of close relation between data fusion and image classification, object detection, change detection, and target recognition, these data fusion based applications are also analyzed briefly. Finally, summary and prospect are described.3272|365|0更新时间：2024-05-07
- 摘要：To avoid noticeable seams between texture patches and minimize the stretch and distortion of the pattern when tiling a texture on surfaces, several texture synthesis methods have been proposed in the past few years, most of them however are time consuming and suitable only for part of textures. This paper presents a new approach for texture synthesis. The algorithm first selects multiple seeds from the sample image and puts them in the synthesized images as initial matching points, those points can be a line, a few lines or rectangle, the algorithm then searches for new matching points among the neighbors of the initial matching points along a spiral path and continues this process in a recursive manner until the synthesizing image is fully filled. Compared with the traditional approaches, the synthesizing process of our technique is dramatically speed up and reach to 360 quantitative grade, and the synthesizing speed can be accelerated further in the luminance space while retaining almost the same result as the image obtained in RGB space. In addition, our method can cope with the variation of the texture along different directions with ease. Experiments show that synthesizing results are quite satisfactory.2912|349|0更新时间：2024-05-07
- 摘要：In this paper, a blind meaningful watermarking algorithm is proposed. The host image is first transformed into wavelet domain where the feature set is extracted. Through subspace decomposition of the feature set, the watermark signal sets are generated which holds orthogonality between each two watermark signals and between the watermark signals and the feature set. For the orthogonality between the watermark signals and the feature set, the meaningful watermark can be extracted without resorting to the host image. And for the orthogonality between each two watermark signals, the characters embedded can be retrieved robustly using matched filtering. Because the watermark signal sets are image content dependent, the algorithm security can be improved. In order to measure the reliability of the extracted characters, the concept of credibility threshold is proposed by which the wrong characters can be correctly pointed out. Therefore the credibility of the whole message can also be evaluated, which is another contribution of this paper. In the experiment, we test the algorithm performance under JPEG compression and Additive White Gaussian Noise(AWGN). The results show that the robustness and imperceptibility of the watermark are satisfactory. The credibility threshold also proves its effectiveness.3039|408|0更新时间：2024-05-07
- 摘要：The anisotropy features(orientation) of texture is very important for texture analysis. They provides an effectively approach to texture representation and graphical modeling. In this paper we try to introduced an new multi-channel model for texture analysis and synthesis based on analysis of these features. We give the spatial/frequency parameter description of anisotropy/isotropy features with the special distribution of Fourier transform coefficients. Based on this description a new algorithm for determine the dominant orientation of texture images is provided by mathematics method of binary scale. It is interesting that this algorithm is convenience to implement by combination of some ordinary image processing algorithm. The Image power characteristic by Fourier coefficients distribution along the dominant orientation is not completely. For the remain Image power, we use the stochastic field model to approximate them, because these power corresponding to the isotropy part of texture image. The experiment results for texture synthesis with this model and the performance comparison with other approach show that the efficiency and flexibility of this method. Our study shows the multi-channel decomposition based on orientation power will be an available method for orientation texture representation and reconstruction.2946|356|0更新时间：2024-05-07
- 摘要：After sudying the method of generating texture image based on BP neural networks in, an improved method based on Logistic mapping and multi-layer forward neural networks is proposed. With the aid of using Logistic mapping, the parameters of multi-layer forward neural networks are adjusted by the new method. That is, using Logistic mapping generat for generating a set of chaotic variables, and each chaotic variable corresponding to a parameter of neural networks which need to be adjusted. Since Logistic mapping has the characteristic of chaos, which make the multi-layer forward neural networks defrent parameters, the defects of BP algorithm which lead the neural networks to converging is overcome. The method based on chaotic mapping not only ratain the virtue of the method in but also improved the original method. The improved method need not calculate the error of neural networks, so not only the operation proceeding is simple but also more texture images is able to be generated. Simulations show that the improved method is simpler and more efficiency the original method in .2679|349|0更新时间：2024-05-07
- 摘要：IHS transform was one of typical method for remote sensing data fusion. In resent years, newly developed method that combines advantages of IHS and wavelet algorithms makes image fusion. The result is better than single method implemented. In this paper a high frequency wavelet substitution method is introduced to improve spatial resolutions of the imagery. ETM remote sensing data 543 bands is transformed in IHS munsell color space and then I intensity was produced. Using high energy in PAN image substitutes I high energy by using wavelet. A new image is prodeced which combines both information PAN and I. Finally the new image as I' to finish IHS inverse transformation. The test result is proved that the resolution was greatly improved comparing original 543 composite in spatial resolution while keeps multi|spectral resolution. In cooperating with the demand of on going Minjiang river retum farmland into grassland project, Si Chuan, China, A 15m resolution PAN band and 30m resolution 3 bands of ETM data were selected for the method testing. Improved imagery raised the quality for monitoring land cover and land use change monitoring.2727|355|0更新时间：2024-05-07
- 摘要：Macroscopical distribution pattern of land cover is a foundation for the land cover's interannual change that is being occurred by the outside driving force. In this study, unsupervised classification method have been used based on the different NDVI images which have different spatial resolution and same temporal, and the macroscopical distribution pattern of East Asian land cover was summed up. Furthermore, annual 5~9 months anomaly percentage grading images have been made using Anomaly Percentage Analysis methods based on the NOAA/AVHRR Pathfinder database, which are 8km spatial resolutions and 10 days temporal resolutions, and generalized the characteristic of interannual changes of east Asian land cover. Which are can be seen as follows: (1)Land cover dynamic of summer(5~9 month) in east Asia apparently differentiated three regions such as the northwest inland area, eastern monsoon area and northern part of east Asia. (2)Inside the monsoon area, land cover dynamics have shown to have trends of differentiate of southeast to northwest. As regards the whole, the interannual changes of land cover of Northeast China Plain, North China Plain, southwest China and Huaihe basin are relatively heavy. (3)In the edge of monsoon area, there exist an obvious dynamic zone which is expanded from northeast to southwest. This is consistent with ecological environment fragile area of northern part of China. (4)In the northern part of east Asia, the land cover's interannual dynamic have feature of zonal distribution. In the same zone, there exist difference of east to west, but the interannual changes are very simple. (5)From the interannual changes of land cover, we have seen that the land cover have trends of increase except northwest inland arid area, especially in the northern part of east Asia and midwest Qinghai-Tibet.2596|354|0更新时间：2024-05-07
- 摘要：Massive data of remote sensing images has been generated as remote sensing technology develops rapidly. The disfluency when displaying massive data was once a problem. We found that these phenomena mainly occurred at the operation of panning or zooming out. In this paper, we analyze the reason of these phenomena at current level of computer software and hardware technology. We also present one method to solve this problem based on the intermediate-image technique, raster-division technique, cache technique and intermediate-image-sequence technique. We had programmed on Windows 9X/2000 platforms by Visual C 6 0 and applied this system at the investigation project of natural environment by remote sensing method in Zhejiang Province. In this project we accomplished the selection of reference points, image inlay and the matching of the TM images of Zhejiang province at 1997 and 2000. Our practices showed this method can avoid the stagnation when displaying in existing device conditions.2844|387|0更新时间：2024-05-07
- 摘要：Shape feature is a very important feature used to depict image, it is difficult to describe shape. This paper proposes "distance distribution histogram", a region-based shape feature which uses the dimensional distribution information of pixels in binary image to describe the image shape. The basic idea is to get shape feature by calculating the distribution of pixels in different distance regions. In the method, the selection of datum mark and the partition of distance region are two important parts to assure that the feature can have good invariant performance under some rigid transformation. After the distance distribution histograms have been calculated for different images, Quadratic-form distance is then used to compare the similarity of the images. The experiments have been done in a database consisting of 3000 binary trademark images, the results show that this shape feature can depict the image well, and it has very excellent invariant performance under translation, scale and rotation. When the feature is used for binary trademark image retrieval, it is obvious that the retrieval results fit with human's visual perception well. From the experiments, it can be concluded that the distance distribution histogram feature is really an effective shape feature for binary trademark image retrieval.2772|373|0更新时间：2024-05-07
- 摘要：The methods of multi-perspective mosaic ask mostly estimate the parameter of camera, the depth of every pixels, and more images. The ubiquitous disadvantage of these methods is over heavy calculation. On account of the difficulty of shooting on the sea, these methods will not serve the view of ship-handing simulator. So the mosaic method based on the simple 2D overlapped image is required, but the multi-perspective and shelter between the scenarios bring more difficulty. This paper construct a imaging geometric model of the overlapped images according the demand in view and background of ship-handing simulator, then proposes a automatic mosaic and smoothness method based on the most depth line on the vertical middle plane and the parallel line on the same depth plane. The experimental results show that the jointing efficiency of this algorithm is better, and it has wide application perspective at fields of multi-perspective mosaic.2850|351|0更新时间：2024-05-07
- 摘要：The paper describes the scheme and design of a high-speed image fusion system based on TMS320C6201 DSP, a miniaturized hardware operative platform for digital image fusion processing with high performance and good universality. Some specific problems on hardware designing multi-channel image fusion processing system are also discussed. As a result of using the latest high-performance DSP and many considerations of structural optimization, the system can meet the requirement of real-time(or quasi-real-time) two-channel image fusion processing from visible and infrared band and has the flexibility of applying various algorithms. Additional manual pixel registration function is provided with moving capturing windows. In a sense the system fairly balances heavy computation burden of multiscale-based fusion algorithms and real-time request for hardware. Actual experiments and tests indicate that the image fusion processing system works steadily and reliably. It takes on satisfactory performance and can realize quasi-real-time image fusion processing with complex algorithms such as Laplacian pyramid algorithm. This job stably founded the way to develop multi-channel real-time image fusion system for the practical purpose.2449|370|0更新时间：2024-05-07
- 摘要：Feature plays an important role in reverse CAD modeling. Based on the multiresolution modeling, a series of approximation models and algorithms for feature curves extraction from the triangular Bézier patch is studied. First, a near coplanar merging algorithm based on Gauss sphere is used to generate superfaces. Then the outer and inner boundaries of every superfaces are extracted, which are original or coarse feature curves. Taking account of the complexity and arbitrary of triangular Bézier patches model, editing operation applying on feature curve is inevitable during practical application. The feature curves are the important reference fo rectangular boundaries definition. This research has laid a good fundamental for fulfilling the local B-spline surface reconstruction from triangular Bézier patches. Some experimental results show that the above method is feasible and satisfying, and reasonable rectangular topological B-spline surfaces can be reconstructed according to the feature curves extracted from the triangular Bézier patches.2691|325|0更新时间：2024-05-07
- 摘要：The vecorization of dot matrix image is a classical question of computer graphics. The least-square algorithm, which distinguishes the result of vectorization right or not by defining whether the fitting errors is in the range of threshold or not, is a main method of vectorization of dot matrix image. So how to define the threshold is a key. These errors are analyzed while dot matrix image is recognized as vector diagram by the least-square algorithm, the threshold is defined scientifically, and the judgement formulas to evaluate recognition resul are put forward to in this thesis. In order to complete vectorization automatically, a new method of the vectorization of dot matrix image--the double direction roll least-square algorithm is presented, and some recurence formulas are deduced. The judgement formulas are scientific and accurate, the mothod of the vectorization is rapid and efficient. The formulas and the method are all quite well and universal.2647|333|0更新时间：2024-05-07
- 摘要：The generation of lines is elementary problem in computer graphics. A four-step incremental generation of lines is proposed. The algorithm firstly gives the choice standard of the pixel that the most closed to the straight line, and then converts it to integral variable form. So it develops and describes a new generation of lines, the four-step incremental generation of lines. The algorithm is easy to be implemented by hardware. It's only integer addition and left shift operation being used in the algorithm of incremental generation of lines. So the algorithm has lower complication in hardware and high speed. The four pixels being drew on the same time by using this algorithm. So the speed of the four-step incremental generation of lines is as three or four times as the speed of the Bresenham algorithm. Scan-symmetrical based on the symmetry of lines cause speed of the four-step incremental generation of lines increased as well. Binary search tree method based on geometrical characteristic of lines is used in the four-step algorithm. So almost the same times of decision per pixel between the four-step algorithm and the Bresenham algorithm, and the times of iterative per pixel of the four-step algorithm is much more less than the Bresenham algorithm.2704|335|0更新时间：2024-05-07
- 摘要：It is well known that Bézier curve and rational Bézier curve play a very important role in the field of Computer Aided Geometric Design &Computer Graphics. They can be used to generate different curves in given condition. So far, some problems can' t be well solved in the regulation of weight factor selecting and the algorithm complexity of curve design. Based on Bernstein polynomials, a new kind of function-Bernstein Function Class is constructed in this paper. Its correlative properties are deduced in details. At the same time, the generating method of Bézier curve class is given. It can be concluded that Bernstein polynomials is a special member of Bernstein function class and the same relation exists between Bézier curve and Bézier curve class. In this article, RB function is chiefly introduced and the method of curve representation based on it is reasoned either. The relation between weighted RB curve and rational Bézier curve is also discussed. It is not difficult to find that weighted RB curve can partly supply the gap of traditional rational Bézier curve in weight factor selecting. So the flexibility of curve design is improved. In the end, some available conclusions and visual consequences are obtained.关键词：Bernstein function class;Bézier curve class;RB function;Weight factor;RB curve2799|359|0更新时间：2024-05-07
- 摘要：A class of piecewise trigonometric polynomial curves of degree m(m=1,2,3) is presented in this paper. Weighted trigonometric polynomial curves are given by using a shape parameter. Analogous to the cubic B-spline curves, each trigonometric polynomial curve segment is generated by four consecutive control points. For equidistant knots, the given trigonometric polynomial curves of degree m are C 2m-1 continuous. The construction methods of an open and a closed trigonometric polynomial curves are described. The given curves can be used to generate ellipses conveniently. The comparisons between the trigonometric polynomial curves and the cubic B-spline curves are given. By choosing m or the shape parameter, the trigonometric polynomial curve can approach to the given control polygon in a different way. Therefore, the construction method of the trigonometric polynomial curves is simple and useful for curve design.2578|377|0更新时间：2024-05-07
- 摘要：For video coding, the algorithm of motion estimation based on blocks is widely used. Under the prerequisite of assuring estimation accuracy, a fast classified search method applying different search range and different search steps for different blocks is given in this paper. According to the criterion of minimization of mean square error(MSE) between two frames, two threshold valuesT1andT2are determined. Each block's MSE is compared withT1andT2. Then blocks' movement is classified into fast, slow and still moving. Different search range and different search steps are applied for different blocks. For fast moving blocks three steps search is used. For slow moving blocks two steps search is used. For still blocks one step search is used. Based on the center-biased property of the motion vectors, the first and second search step use the 5×5 window and the third search step uses a 3×3 window. The results show that this method is better in quality than 3SS and reduce 3SS search steps by 23% and FS search steps by 91%. Furthermore this algorithm is more suitable for VLSI implement as well.2364|335|0更新时间：2024-05-07
- 摘要：In this paper, a new shape coding algorithm called curvature scale space based shape coding(CSS) is proposed. For dealing with video image, we regard a shape as a set of points that are parameterized by arc length. Then, the evolution of curve at different resolution levels in Gaussian scale space is achieved by convoluting the curve with Gaussian kernel. By calculating the curvature, the feature-points including significant information of shape contour can be found. But for shape coding, this shape representation will not very efficient when the shape consists of arcs. So, the CSS is modified, then extract the feature points of shape using the modified CSS algorithm. The octants algorithm is modified and encode the feature points by using adaptive arithmetic code algorithm. Simulation results show that our proposed algorithm has higher compress ratio, better subjective quality compared with other results from the CAE in MPEG-4.2532|392|0更新时间：2024-05-07
- 摘要：To compress the stereo video data, we propose an algorithm for stereo image coding based on hierarchical disparity estimation/compensation. A stereo image pair consists of a reference image and a target image. The reference image is coded by method using JPEG standard, while the target image is coded using the disparity estimation/compensation technique. We apply the wavelet transformation to code the created residual image, and use DPCM(difference pulse coding modulation) to code the disparity fields. Hierarchical Markov Random Field(MRF) model, Rate-Distortion(RD) model and the overlapped block matching/compensation are applied in the hybrid disparity estimation method to obtain a smooth and consistent disparity field. In terms of peak signal to noise ratio(PSNR) of the disparity compensated image, the proposed disparity estimation algorithm achieves about 2 5 dB higher PSNR, as compared to conventional variable size block matching. The experimental results also show that the disparity estimation/compensation in half pixel accuracy and dual-direction can obtain a 3dB higher performance than that of integer pixel accuracy only in horizontal direction. In the same quality of reconstructed image, our coding system can obtain 0 387 bits reduction in terms of coding bit rate of the target image compared to VSBM based scheme, and 1 003 bits reduction compared to the one using JPEG standard to code the target image. The proposed hierarchical model for disparity estimation effectively improves the efficiency of stereo image coding.2896|359|0更新时间：2024-05-07
- 摘要：Real-time simplification of terrain is very important in visualization of 3D terrain and Virtual Reality(VR). To best exploit the rendering performance, the scene complexity must be reduced as much as possible without leading to an inferior visual representation. The most common way to increase efficiency is the use of different levels of detail(LODs) for different areas of the scene. The algorithm in this paper is a real-time terrain simplification algorithm, based on local entropy. The algorithm adopts quadtree structure, which can provide real-time computation and high efficiency. The algorithm uses "Local Entropy" as error metric, improving the quality of simplified results. In an aspect of fixing "gaps" that is unavoidable in quadtree structure, the algorithm adopts more rational data structure to acquire higher efficiency. The experiment results indicate that the algorithm is practice and effective, which can meet the demand of real-time rendering in 3D terrain simplification and VR.2924|331|0更新时间：2024-05-07
- 摘要：Realistic 3D Models of Humans are widely demanded in many fields such as production design ,computer animation and virtual reality. A low cost method based on shape from silhouette is presented to construct a virtual human. Only one digital camera is used to take photos of a human from different position. Each view is calibrated individually with a calibration pattern placed on the ground. From these obtained calibrated camera views, the volume model of a person is estimated using shape-from-silhouette approach. An algorithm of surface reconstruction is used to extract the surface of the person from volume model. After texture mapping with original camera views, the realistic 3D Model of the person is obtained. In order to make the result model more realistic, some advanced techniques like texture filtering and texture synthesis are applied to the textured 3D model. Compared with some active scanning methods such as laser scanning and structured lights, the running experiment results show that this approach is very efficient and the accuracy is good enough.2847|431|0更新时间：2024-05-07
- 摘要：To obtain high accuracy and real time processing in wheel wear detection, firstly, the article put forward a method of camera calibration based on the previous research, which simplified the computationin camera chlibration by use of the special features in this imaging system; secondly, it suggested a new edge growing method based on easy differential algorithm and the classical 8 neighbor fields method. in which the neighbor area of a pixel on the wheel edge is analyzed in datail and then the method was put forward according to the special character of wheel image. It has been verified by the test on spot that the calibration algorithm is easy to be realized and can meet the accuracy requirement and that the edge growing algorithm is fast enough to be processed in real time and can obtain the better tracking result. In a word, the algorithms are effective and of a certain value.2631|318|0更新时间：2024-05-07
- 摘要：The quantitative analysis of medical image becomes more and more important in disease studies, diagnosis, and treatment. Our goal is to construct the image sampling module of self-developed image capture card and design a specialized pathological analysis system to meet the requirement of domestic doctors and researchers. Successfully making use of the powerful functions of Image-Pro Plus(IPP), we select Visual C and software development kit(SDK) of IPP as programming language to develop self-designed plug-ins ans algorithms. Our system, which is much easier to use, includes seven functional modules. And the system, with perfoumances of image sampling, displaying, storage, preprocessing, data analysis, and resut report, is practical and convenient to learn for domestic clinical researchers.2470|572|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