最新刊期
卷 9 ， 期 8 ， 2004
- 摘要：D modeling is currently one of the most important issues in computer vision and computer graphics. Human faces are so common and reachable that many new algorithms are first tested on the face data. Also, due to the complexity and flexibility of human faces, many researchers have taken on the challenge of building realistic 3D face models. This paper provides a comprehensive overview of recent developments of 3D face modeling and keeps up with the latest research mainly from 1990 to now. Different from previous reviews, this paper puts emphasis on four major issues involved in a general frame work of 3D face modeling, i.e. 3D capture, generic 3D face modeling, individual face model generation and its application. This paper focuses more on overall methods and general characteristics involved in above four issues. To provide more detailed discussions, each issue is accordingly divided into categories of approach and various main methods proposed in each issue are clearly discussed to examine the state-of-the-art in human face modeling. Based on the discussion and analysis, this paper tries to summarize the intrinsic principles and common methodologies. Finally, some detailed discussions on research topics and future direction on 3D face modeling are provided.关键词：D face modeling;face animation;3D data acquisition;face recognition4566|286|0更新时间：2024-05-07
- 摘要：The article compares several universal integration techniques and methods via to access and integrate Spatial Heterogeneous Data on GIS. And in the paper, the authors discuss the key theory and technique of Spatial Heterogeneous Data Integration. Last, we give emphasis introducing the Integration method based on XML and Middleware technique. And we consider that the most major of XML is the independence of it's data storage format with display form. Along with the development of RDF and VDB, with data describing tool and transition tool based on XML by way of integration layer, middleware possessing capability of data sources integration is suit to the need of spatial heterogeneous data source integration.2684|153|0更新时间：2024-05-07
- 摘要：This paper presents a SPIHT(set partition in hierarchical trees) and Motion Compensation Based Rate Scalable Video Compression algorithm. We will refer to this new technique as the Scalable Adaptive Motion Compensated SPIHT (SAMCS) algorithm, which can be used in video surveillance, video meeting, etc. SAMCS has three characteristics: Firstly, it uses adaptive Intra/Inter block coding and the lowest rate based reference frame reconstruction to adapt to the various bandwidth; Secondly, it makes wavelet transform on prediction error frame (PEF) to create a lot of zero pixels and achieve much higher compression rate; finally, SPIHT is a fine scalable coding, so SAMCS can get scalable rate bitestream based on the lowest rate reference frame reconstruction, which is very appealing for network-oriented application. Experimental results show that, SAMCS has an excellent performance on compressing low-motion video. For example, the PSNRs of Akiyo's three components(Y, U, V) are 36.33dB, 40.22dB and 42.52dB by SAMCS at 27kbps, and the compression rate is 355∶1. As to high-motion video, such as Singer, the performance of SAMCS is no better than MPEG-4 VM. What's more, compared to MPEG-4 VM, the PSNRs of Y components is improved 4.0dB on the average at the highest bit-rate.关键词：video compression;scalable encoding;SPIHT;discrete wavelet transform;motion compensation2974|181|0更新时间：2024-05-07
- 摘要：Integer transform has been adopted by the newest international video coding standard H.264 for its high efficiency of computation. Our first national audio and video coding standard AVS will also adopt 8×8 integer transform for video compression. As transform module has strong influence on the efficiency of codec, it is important to select an excellent transform basis (transform matrix). In this paper, after introduction of basic theory of DCT-like 8×8 integer transform, the authors bring out three principles of transform basis evaluation: (1) orthogonality, (2) transform efficiency, and (3) computation complexity. First, the transform bases satisfied principle (1) can be found in a predefined data range. Principle (2) is embodied in a novel method combining evaluation of transform efficiency from two aspects: energy compaction and decorrelation efficiency in a proper range of correlation coefficients of input image prediction difference data. By using such evaluation method and taking principle (3), computation complexity of fast transform algorithm into consideration, several theoretical highly efficient transform bases such as (10,9,6,2), (5,6,4,1) and (4,5,3,1) have been picked out and tested in the actual video sequences. Based on careful analysis of testing results, the authors find out the proper situation that each basis fits for and prove the validity of transform basis performance measurement that uses this theoretical evaluation method.关键词：integer transform;DCT;energy compaction efficiency;decorrelation efficiency;AVS4528|212|0更新时间：2024-05-07
- 摘要：This paper presents a new color correction algorithm called Illumination &Hue Color Constancy. The algorithm mainly resolves the color constancy issue of which color features of the various spatial points are often influenced by the effects of the scattered lightness of the near light source with the complex background in video color recognition research. Based on the illumination principle and the mathematic model of color sense, this paper combines the illumination corrective algorithm with the hue corrective algorithm to achieve the color constancy. By constructing a simulation of the machine vision system, the paper evaluates the algorithm' s color recognition ability, and also shows the errors values of RGB and hue to analyze its corrective ability. With the comparison of the experimental results for illumination corrective algorithm, hue corrective algorithm, illumination & hue corrective algorithm, and linear regression algorithm, the paper demonstrates that the algorithm has not only good corrective ability for the color influenced by the effect of the invariant illumination under scattering light, but also has accurate recognition ability for the color features. As a result, the paper presents that the theory of the hue & illumination algorithm is rational and corrective. Taking the advantage of the compute complexity which isO(n), the hue & illumination technique is easy to achieve and might have extensive usefulness in real-time video recognition.关键词：video;color constancy;illumination;brightness;hue3595|210|0更新时间：2024-05-07
- 摘要：This paper presents a new wavelet based image coder with the property of progressive multi-resolution. After the image is decomposed with lift wavelet, the wavelet coefficients are quantized visually lossless by utilizing the properties of HVS. Then, in the constraint of the bits budget, the bits are allocated optimally to each subband according to the rate-distortion curve and their contribution to the visual effect. At last each subband is independently encoded with the procedures of layered quantization and arithmetic coding. Without the iteration of all the quantization steps, the bit allocation algorithm deduces the theoretic formula of allocation by modeling the wavelet coefficients as generalized Gaussian distribution. The allocation result is optimal in the sense of rate-distortion. The experiments show that, compared with the classical coders, the coder has the near PSNR values and the better visual effect in low bit rate such that it displays the superior performance.3219|178|0更新时间：2024-05-07
- 摘要：Conventional interpolation algorithms of image, such as zero-order or nearest neighbor, bilinear, cubic spline interpolation, can be analyzed in two steps: (i)upsampling by zero filling, and(ii) low-pass filtering. But ideal low-pass filtering of interpolation can not be practically achieved, which results in high-frequency artifacts in the interpolated image. On the other hand, due to the low-pass filters fixed, these algorithms fail to utilize the information of the image itself. In order to improve the quality of the interpolated image and enhance the resolution of it, a novel regularized image interpolation algorithm based on genetic algorithm is proposed in this paper. This genetic algorithm has real-valued coding, the induced mutation operator and the fitness function for evaluation containing the term of some subjective quality measures, so the convergence of the genetic searching in the solution space is very fast. Finally, we analyze how to choose the regularization parameter in the fitness function, and compare the results with that of iterative regularized interpolation algorithm. The experiments demonstrate that the proposed algorithm is practical and applicable.关键词：image interpolation;real-valued genetic algorithm;regularization;resolution enhancement2832|230|0更新时间：2024-05-07
- 摘要：An image super-resolution algorithm based on wavelet-domain local gaussian model is proposed. Wavelet-domain local gaussian model approximates the local probability distribution of the wavelet coefficients with a single gaussian function. Because the model adaptively characterizes the local statistics of real-world images, the algorithm presented in this paper specifies the prior distribution of the real-world image through it and converts the image super-resolution problem to a constrained optimization one which can be solved with the conjugate gradient method. Experimental results show that the algorithm properly retrieves various kinds of edges and the PNSR and subjective visual effect of the reconstructed images are improved significantly.关键词：image superresolution;wavelets;local gaussian model;conjugate gradient method4304|158|0更新时间：2024-05-07
- 摘要：Noise filtering is a necessary step on image pre-processing stage in digital image processing. In most situations we will deal with image noise with high density, especially it is very difficult for processing image corrupted by salt-and-pepper noise with high probabilities. In this paper, the disadvantage of normal noise filter working in spatial domain was analyzed, and a new filtering technique, that is, extended adaptive median filtering technique was proposed. The extended adaptive median filtering not only inherites the advantages from standard adaptive median filter, but also remedies the corresponding disadvantages. It resolves the filtering problem for image corrupted with high density, and derives reasonable perfective effect. Finally several resulting figures from filtering technique effect figures are presented to verify the new noise filtering technique developed in this paper.关键词：filtering;median adaptive;extended;implement3663|191|0更新时间：2024-05-07
- 摘要：A new self-adaptive median-opening-closing(MOC) filter is proposed to remove the salt noise of images in this paper. The self-adaptive MOC filter is composed of a MOC filter and a learning optimizing module. In order to achieve the best effect of removing the salt noise, a group of sample images are given to the optimizing module for learning regularity of salt noise and a genetic algorithm is applied to the module to automatically adjust the structure of the MOC filter according to the result of the learning in the optimizing procedure. An optimized structure of MOC filter is generated at the end of the learning and optimizing process and then may be used as an optimized filter in practice for reducing the salt noise in image. Computer simulation experiments are done separately when image contain salt noise of different intensity. The results obtained in the simulation process show that using the self-adaptive MOC filter can achieve good effect of filtering the salt noise of the images compared to use individual median or morphological filter, specially when the image contain more salt noise. How to select parameters about the optimizing process is also discussed in this paper.关键词：simage filtering;median filter;mathematics morphology;genetic algorithm;image denoising2714|180|0更新时间：2024-05-07
- 摘要：This paper is based on Canny algorithm.An adaptive Canny algorithm of edge-detection method is proposed.This algorithm not only keeps the Canny's excellent performance in good localization,only one response to a single edge and good detection,but also improves the performance in the detail edge-detection and good detection. Canny adaptive algorithm divides image into sub-images and detects them with adaptive threshold value according to the whole image edge information, that improves the automaticity of edge-detection.With the mathematic analysis and test result,it is demonstrated that the adaptive edge-detection method is an efficient improving approach on edge-detection.关键词：image processing;edge-detection;Canny algorithm;adaptive threshold5463|316|0更新时间：2024-05-07
- 摘要：This paper presents a correlation tracking algorithm based on correlation coefficient, which overcomes the disadvantages of traditional correlation tracking based on point-to-point multiple with accumulation and has the advantages of good accuracy and high stability. At the same time, many measures are put forward to improve the speed of the algorithm, which resolve the requirement of real-time of object tracking. And during the object-tracking, there may be many changes in a sequence of the object images, therefore, a reasonable strategy of template updating will be the key of the object-tracking problem. On the basis of the similarity measurement and template buffers, a suitable template updating strategy is given, which effectively decreases the accumulation of the object-tracking error, and greatly improves the stability of the object-tracking. The failure judgement of object-tracking presented effectively resolves the transitory failure of tracking, caused by sudden changes, such as the changes between the dark and the bright, or the tracked object is covered temporarily. The experiments show the solution decreases the complex of correlation tracking, and has the advantages of good accuracy and high speed as well. Now this algorithm has been applied to real-time object-tracking system.2814|217|0更新时间：2024-05-07
- 摘要：To construct lighting model of the real world is one of the critical technologies in augment reality. The theory of lighting modeling used in Computer Graphics(CG) is applied in our research. At first, the position of the real light-source is conjectured from the lighting and shade of the registration image elementarily by using ray tracking algorithm. Then the virtual light-source and virtual fiducial are constructed in the CG environment, in which, Phong model is used to draw the light effect. By comparing the CG scene with the real image, we can modify the parameters of lighting model over and over again, until the lighting effect of the CG scene is similar to that of the real image. It is proven that this method works well in the indoor AR system. Our method could be used feasibly in most applications after some improvement.2620|204|0更新时间：2024-05-07
- 摘要：In this paper, a new approach is proposed for reconstruction a three-dimensional planar object from its single line drawing depicting it under orthographically axonometric projection. The line drawing is supposed to be a nidden-drawn one. A new method with uniqueness for representing lines in two-dimensional projective plane and three-dimensional space is introduced. The coordinates relations between a line in three-dimension and its orthographically axonometric projection line in two-dimensional plane are put forward. It is also provided the coordinates relations when two lines are intersecting. On the basis of these coordinates relations, some new constraints about lines and planes have been proposed, and a linear system based on these constraints is eatablished. The three-dimensional information of a planar body can be obtained by solving the linear system from its orthographically axonometric projection. It is proved also that, in the interpretation of line drawing of a planar object under orthographically axonometric projection, there exist at least four degrees of freedom, the three of which are used to determine the location of the planar object, whereas the last one of which is used to determine its pose.2937|184|0更新时间：2024-05-07
- 摘要：Based on the assumption of a generalized Gaussian model for the additive noise, this paper develops the regularized image restoration algorithm and proposes anlp-norm data item to the regularization objective functionnal instead of the usual quadratic data item. Meanwhile, the paper applies an adaptive method for choosing the regularization parameter. An iterative algorithm is utilized for obtaining the restored image and the regularization parameter, which can be determined in terms of the partly restored image at each iteration step therefore allowing for the simultaneous determination of the restoration of the degraded image and the value of the regularization parameter. Numerical experiments demonstrate that our method results in high restoration performance when the image was blurred by a Gaussian PSF and an additive generalized Gaussian noise, especially when the shape parameter p≤1.3205|179|0更新时间：2024-05-07
- 摘要：It is important to organize the unstructured video data properly for content video analysis and application. However, existing shot-based video analysis can not get good performance for effective browsing and retrieval because shots are too numerous to handle and can not convey meaningful semantics, so it is necessary to organize video content with the scene structure, which is a more meaningful and high-level semantic video unit. In this paper, a simple and efficient method is presented for video scene detection. As the buildup of video scene should obey the film grammar, it concludes the types of the scenes that are widely used in filmmaking firstly. The rules to generating the video scene are put out then. Based on the rules, a scheme for video scene detection is proposed: The video stream is segmented into shots through shot boundary detection; Key frame extraction is performed based on the content variation of shots; A new clustering method based on the sliding shot window is used to group the shots into shot clusters; A correlation function between shot clusters is defined to analysis the correlations of them and to construct the final video scene structure. Experimental results verify it is an efficient method.关键词：shot;shot cluster;scene;sliding shot window;correlation function of shot clusters3924|210|0更新时间：2024-05-07
- 摘要：As the ASAR sensor has successfully passed the primary validation period, more and more users in China will be able to get ASAR data in the near future. If users apply the data to practical mapping, the first key problem which should be solved is to geolocate the SAR image. Based on Range-Doppler(RD) model, a geolocation method has been developed for the major imaging products of ENVISAT Advanced SAR (ASAR). As one special contribution of this paper, the way to construct range equations for different ASAR products has been solved successfully. The range function of slant range image products is just a linear equation in one order, which can be defined by the first range distance and the range pixel spacing. For precision ground range products, the one and only SRGR record provided by the metadata can be used directly to construct the range equation. But for median resolution ground range products, several SRGR records have been provided by the metadata, the range function should be established through interpolation method according to the relation between the azimuth time for each image line and the azimuth updating time for each SRGR records. Five scenes of ASAR image have been geolocated with this method and the relative geolocation accuracy has been evaluated taking the geographic locations of four image corners derived from SAR metadata as reference. The geographic locations of the four image corners have been calculated and stored in the metadata of SAR image product by the original SAR processor. The product type ID for the five image products is APP -1P, APM -1P, APS -1P, IMP -1P and IMM -1P respectively. It has been shown that the location error in both latitude and longitude direction is below 1.0 meter for precision products such as APP -1P, APS -1P and IMP -1P; For median resolution products such as APM -1P and IMM -1P, the maximum longitude error is 59.73 m, about 0.8 times of the image pixel space; and the maximum latitude error is 83.38m, about 1.1 times of the image pixel space. It shows that the geolocation result achieved by the developed method is quite consistent with that achieved by the SAR processor to produce these ASAR image products. Also, the result will be useful for the development of orth-rectification and the other relevant processing algorithms for ASAR image products.3140|187|0更新时间：2024-05-07
- 摘要：It was nearly 20 years ago that the landuse map was made, which is inconsistent with the facts gradually and needs to be updated to provide accurate landuse information for the rapid development of the region. Having the advantage of the high resolution and spatial view, the SPOT5 image is taken as the suitable remote sensing data for updating the landuse map. Based the experimental study in Jiu Yuan district of Baotou city, we put forward a practical technology and method to realize the 1∶10000 landuse map updating. From the view of the users, we use the landuse map (DLG) and topography (DRG) to produce orthophotoquad image, and extract the landuse changing information from the overlaid maps composed by landuse map and the SPOT5 image. The uncertain information is further investigated by field work. The experiment show that the updating technology based on the SPOT5 image is more efficiency, economical and practical than the TM image based technology which were used before.2599|220|0更新时间：2024-05-07
- 摘要：Multiwavelet is an extension from wavelet theory, and has several particular advantages in comparison with scalar wavelets on image processing. Multiwavelet analysis can offer a more precise way for image analysis than wavelet multi-resolution analysis. In this paper, we research the characteristics of multiwavelet and give a fusion method by means of discrete multiwavelet transform. In the experiment, we have obtained the fused picture through merging a pair of remote sensing images with the different resolution of the same district by applying this image fusion method. Comparing with other fused images and evaluating them in way of objective and subjective performance, we can draw the conclusion that this method can fuse details of input images successfully, and it can perfectly display information of the each input image, therefore using this image fusion method can get more satisfactory result than using others.关键词：image fusion;discrete multiwavelet transform;fusion algorithm;remote sensing images2702|236|0更新时间：2024-05-07
- 摘要：Handwritten character recognition using the hidden Markov model (HMM) has been an active research topic for the past decade. One of the major problems, however, is that the handwritten characters may not exhibit consistent patterns due to different people's different writing styles. To enhance HMM's encoding stability and to reduce its modeling complexity, we propose a new approach in this paper. Specifically, we first obtain a set of uncorrelated optimal discriminant vectors by conducting feature extraction and dimension reduction using the uncorrelated Foley-Sammon transformation. Next, using a new feature space spanned by the optimal discriminant vectors, we obtain the projection coefficients of the raw data onto this new feature space. We then use these coefficients to form the observation sequence of the HMM. Because the uncorrelated Foley-Sammon transformation ensures minimum intra-class distance and maximum inter-class distance, it significantly improves HMM's encoding stability and difference classes' separability. In fact, the transformation allows different characters to be separable in many projection directions. To validate the accuracy and robustness of the proposed approach, we conduct experiments on the widely used US Postal Service (USPS) data set. Experiments show that the integration of the uncorrelated Foley-Sammon transformation and the HMM performs very well, achieving a recognition rate of 92%. It not only is better than regular HMM, but also is superior to the widely used nerual network based approaches.3034|187|0更新时间：2024-05-07
- 摘要：Finite element analysis (FEA) plays more and more important role in sheet-metal forming. However, if the analysis is performed on a low quality mesh, the FEA can be inaccurate. The algorithm of combined mesh generator for complicated surface is proposed to meet the requirement of sheet-metal forming CAE software. The boundary node of surface is generated according to mesh density first. The combined mesh is paved inside until the region is filled. The offset point acquirement of mesh and combined mesh generating steps are proposed. According to the collision type of mesh, three different solving methods (combination of nodes in loop, separation of loop and join between loops) are discussed. The algorithm has been used to metal forming CAE software. The practical application proves that the algorithm is robust, accurate and fast. The meshes which are generated by this algorithm have no crack and overlay in neighboring boundary of surfaces. It is satisfied for metal forming CAE software.2804|178|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