最新刊期
卷 4 ， 期 10 ， 1999
- 摘要：A real time algorithm is proposed for the detection of abrupt scene change in uncompressed as well as compressed video. The proposed algorithm is based on the dual (one big and one small) sliding windows. The big window is used for selecting the probable positions of abrupt scene change, and the small window whose center is at the probable position selected is used for determining the real positions of abrupt scene change. The decision made in the small window is helped by a combination of double side checking and a single side checking. This algorithm can effectively avoid the false detection and miss detection caused by the violent motion of camera and/or large objects. The effectiveness of our algorithm has been demonstrated using five video clips (about 9000 frames, 98 abrupt scene changes), the recall rate is 100% and the precision rate is 96%.2362|135|0更新时间：2024-05-08
- 摘要：This paper presents a new face recognition method in Fourier domain. The proposed method combines the Fourier transform and the wavelet transform for face recognition. Yale and Olivetti face image databases are selected to evaluate the proposed method. The results show that the proposed method gives higher accuracy than PCA method and template matching in spatial domain. Spectroface is extremely effective for eliminating the errors that bring from different expressions and small occlusion.关键词：Face recognition;Fourier transform;Wavelet transform;Minimum distance classifier2677|156|0更新时间：2024-05-08
- 摘要：Automatic face recognition/identification is very useful in some areas. Usually, an ID number is applied for that purpose. This can't prevent the forge ID to be used. The visual feature (such as face, gesture etc.) is relatively stable, and different between each other. So these information can be used to identify the human being. In this paper, a multi candidate based approach for pose varied face recognition is given. To reduce the affection of pose variety, some pre process methods are also applied. Actually, the quality of the image is a very important factor for automatic recognition, so a method for image quality estimation is proposed. The experiments show that the image quality is associated with the recognition result. When the input image is relatively clarity, it can reach a good recognition result.关键词：face recognition;Pose varied;Image quality estimation;eigenface4130|182|0更新时间：2024-05-08
- 摘要：This paper develops a multi template matching based method for single face detection in general background images. Firstly two eyes in whole template is used for determining the potential location of the face roughly. Secondly 5 whole face templates with different width height ratios are used for searching and localization. Finally some edge features are used for verifying the detection results. Input image is resampled via constant ratio in order to deal with the problem of multi scale detection. Experiment results show that it is relatively fast and rather successful.4085|237|0更新时间：2024-05-08
- 摘要：Radial basis function neural network (RBFNN) and multilayer perceptron neural network (MPNN) are very similar with topologic structure.They play very similar roles in providing techniques for approximating arbitrary nonlinear function mappings between multi dimensional space.Two models are compared in this paper.This paper describes a method to construct an RBFNN classifier by a fast clustering algorithm and computes the optimal weights between the middle layer and output layer statistically.We applied the proposed method to construct a RBFNN classifier for remote sensing image classification.The experiment showed that the method could construct classifier quickly and the performance of the classifier was better than others.2358|170|0更新时间：2024-05-08
- 摘要：The disadvantages of the OPTA thinning algorithm, i.e. incomplete thinning and slow speed, are analyzed and studied. A new algorithm is presented which has the advantage of quick speed and complete thinning. The thinned fingerprint skeleton is smooth and located in the center of the ridge.3549|150|0更新时间：2024-05-08
- 摘要：The commonly anti aliasing algorithms in texture mapping are analyzed in this paper, and the modified summed area tables algorithms based on the simplification of the geometry transformation in the texture mapping are proposed. Through the local linearity of the transformation the mapped pixel become parallelogram, then the adaptive rectangle subdivision is proceeded on the bonding box to eliminate the aliasing. The result shows that the modified algorithm can improve the quality and reduce the calculation at one time.2518|126|0更新时间：2024-05-08
- 摘要：Direct Volume Rendering (DVR) algorithm is an important method to display the three dimensional data and it is widly used in medical image rendering and sicentific visualization. Furthermore, ray casting method is a basic method of DVR algorithm. However, it is limited by plenty of computation. From the view of reducing the number of casted ray, this article provided an adaptive ray casting method based on the optional sampling pattern. The advantage of the algorithm is not only using the simple human vision model and powerful fourier analysis tools, but also making the advantage of the sampling pattern being more suitable to the human vision model and reducing the aliase pattern. As a result, more noise is changed to be more insensitive. So, this algorithm not only improves the image rendering quality, but also reduces the image rendering time.2373|178|0更新时间：2024-05-08
- 摘要：Transportation networks have apparent characteristics of spatial distribution, which is the basic difference between the sparse graphs describing the transportation networks and other planar graphs describing topological structure or hierarchical structure. A time shortest path algorithm based on the restricted searching area is presented in this paper. On the basis of quadary priority queue and inverse adjacent list data structure, a restricted searching area algorithm is established, based on the statistical analysis on the spatial distribution of transportation networks, with the appropriate restricted ellipse and rectangle areas for appointed start and destination nodes. The restricted rectangle searching area is proved more appropriate as the restricted area. Considering the different road travelling impedance and turn impedance, a time shortest path algorithm is presented. The presented algorithms can efficiently decrease the searching scale and improve the running efficiency. It has been proved by some tests on the real-world transportation network.关键词：Shortest path algorithm;Transportation network;Restricted searching area;GIS2950|178|0更新时间：2024-05-08
- 摘要：Corner feature is widely used in pattern recognition for it is invariant to scale, shift and rotation. This paper introduces three algorithms to extract corner feature, named splitting merging algorithm, the extended Hough transform algorithm and the extended corner algorithm. Among them the splitting merging algorithm is the easiest one to implement but is the most sensitive to noise. The extended Hough algorithm is much strong to noise and the last algorithm is much more suitable to describe the characteristics of practical contours. Result of computer simulation for recognizing four occluded objects by using these algorithms is given.关键词：pattern recognition;Local feature;Corner feature;Occluded image recognition2357|131|0更新时间：2024-05-08
- 摘要：The moment invariants play an important rule in feature extraction and pattern recognition. This paper presents an efficient method for moment computation. The experiment shows that the new method has higher speed than the conventional way of moment computation.2423|104|0更新时间：2024-05-08
- 摘要：This paper is the literature bibliography on computer graphics in China in 1998. Based on the current status of the research on graphics engineering as well as the distribution of related publications in China, we selected about 268 references from 12 important Chinese journals published in 1998 and classified these references into different categories according to their contents. We hope that the paper could provide readers with a general and up-to-date overview of graphics engineering in China.关键词：Computer Graphics;Graphics engineering;Graphics processing;Technique application2338|142|0更新时间：2024-05-08
- 摘要：In order to explain the construction of the multimedia information system, this paper demonstrates and analyses the fundamental data model of time based media, hypermedia document, video information and composite multimedia object. Furthermore, the logic representation method of the vision data which is used to implement content based retrieval and feature identification is exposed. And then some opinions about current focuses of graphic's database study and future development are put forward in the paper.1955|107|0更新时间：2024-05-08
- 摘要：Image mosaicing is to“stitch”together overlapping multiple images into a large image. One of the difficulties in image mosaicing is, when the overlapping region has not evident features or has many similar features, pseudo-match will happen and cause mosaicing failure. Based on MSE curve and constraints between the images, an approach was proposed for distinguishing and solving of pseudo-match. For there are constrains between the images, each image has several calculating paths. When the results of several paths are not identical, pseudo match happened. The difference of two minimal extremum on MSE curve can be used as the degree of belief of matching. We select the path with the highest degree of belief as right matching path. Experimental results demonstrate the efficiency of this method.2368|121|0更新时间：2024-05-08
- 摘要：Due to the inherent characters of motion estimation algorithms, the matching speed is greatly influenced by the implementation technique. Based on the detailed analysis of the block based exhaustive searching algorithm and making full use of the programming language C/C , a best program technique suitable for motion estimation is given. The results of using the proposed technique for different motion estimation algorithms are also given.2330|138|0更新时间：2024-05-08
- 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