最新刊期
卷 7 ， 期 7 ， 2002
- 摘要：Human motion tracking has become a hot issue in the fields of image processing and computer vision. It is receiving increasing attention from more and more researchers due to its wide applications in human-machine interfaces, virtual reality and smart surveillance, robot simulation, etc. In order to track human motion precisely, two methods of automatic tracking are proposed. One method is based on 2-D model. Taking region overlay as the matching criteria, the method adopts a matching procedure from coarse matching to fine matching for establishing the precise correspondence between the 2-D model and real human body motion; the other method is based on the feature of body parts and their connection. By locating the body parts according to their different shapes, the joint connecting the body parts can be located also, and then the human movement can be recovered by 2-D skeleton model. Based on the two methods, real human motion is tracked from simple walking movement to complex gymnastics. The result shows that the method can resist the affection of noise and the variation of intensity, and the position of the overlapped area can be estimated coarsely. Precise result can be acquired in the whole course of the movement of the body.2266|259|0更新时间：2024-05-08
- 摘要：Joint transform correlator based on opto-electronic hybrid processing is discussed to recognize high similarity industrial parts images. Experiment system of joint transform correlator is presented. Muitilevel simulating targets based on basic shapes are constructed in order to simulate different industrial images. Joint image similarity degree for joint transform correlator is defined based on morphology method. Joint images are graded based on image similarity degree. Two kinds of method based on principle improment and non-principle improment are refined for high similarity image recognition in order to reduce false decision. Octagon with holes, octagon and pentagon are selected as input images from basic shapes and pentagon is selected as reference image. Structure light pattern is put forward to encode joint images for non-principle improment. Complementary encoding method based on morphological hit-or-miss transform is applied to code joint images for principle improment. Distinct effect is acquired with industrial basic shapes recognition by computer simulation. The results indicate that we can recognize high similarity images by raising JTC recognize ability or reducing image similarity degree.2198|278|0更新时间：2024-05-08
- 摘要：A new small target detection method based on energy accumulation and morphology top-hat operator is presented for solving detection of moving small target with low SNR in infrared image sequence. Accumulating the energy of the image sequence in the sliding window with a setting size can remove the random noise in the images in order to increase the SNR, the principle of accumulating the energy of the image sequence is given. Morphology Top-Hat operator can detect peaks and valleys in images, it can be used to detect candidate small targets. According to motion continuity and motion trajectory consistency in multi-frame successive image, the true small targets can be filtrated from candidate small targets. The algorithm of small target detection based on morphology Top-Hat is given. By means of experiments the performances of noise proof and background suppression and false target suppression are compared between our detection method and high-pass filtering detection method. The experiment results show that the method is better than the high pass filtering method in the performances of target detection and it can effectively and reliably detect the moving point target with low SNR.4762|335|0更新时间：2024-05-08
- 摘要：In this paper an algorithm for detecting blood vessel shape in image of digital subtraction angiograph(DSA) base on mathematical morphology is proposed. In the first step of the proposed algorithm, the central line of the designated blood vessel can be detected by means of matched filter and adaptive tracking, and then the edge of the blood vessel is extracted by using erosion gradient operator. The performances of dilation gradient, Beucer operator and VVL operator are also discussed. The experimental results show that the proposed method has important value for clinic application. The method can be used directly for detecting blood vessel in DSA image without smoothing DSA image, so that some useful information of blood vessel can be preserved and the high precision of edge detection can be achieved without any strict restraint on the shape of blood vessel.2176|257|0更新时间：2024-05-08
- 摘要：A serious disadvantage of Digital subtraction angiography(DSA), inherent to the subtraction operation, is its sensitivity to patient motion. The resulting artifacts frequently reduce the diagnostic value of the images. This paper presents an algorithm based on registration for the image enhancement of DSA, it employs block matching method for the registration of mask image and live image. Because of the high computational cost, it is usually not admissible to determine the displacement explicitly for every pixel. This has led to the concept of control-point based registration. Most control-point based registration algorithms published so far employ regular grids. However, control point selection should be image-feature based. Our experiments have demonstrated the advantage of algorithm with edge based control point selection over algorithms based on regular grids. Other aspects in our algorithm include similarity measure, interpolation and search strategy: energy of the histogram of difference is selected as the similarity measure, fast spline interpolation is used in order to implement the algorithm fast and get better result, and Powell optimization strategy is used as the search strategy by carefully choosing block size. The experiments demonstrate that the subtraction images after registration are of high quality and much more diagnostic value, the algorithm is also fast to be implemented.2654|277|0更新时间：2024-05-08
- 摘要：Informations are playing more and more important role in our life. Most of them are images. However, many kinds of noise will occur during image generation, communication and transformation, and will affect the image quality. By using image-smoothing technique, the noise can be dwindled, but some image details and value information are also lost while the dwindling noise by using the most used traditional smoothing methods. For solving this problem, a novel image noise smoothing method based on high order statistics has been proposed in this paper. It uses the property that high order statistics are not sensitive to Gauss noise to process image-smoothing, and can mostly save the image details while greatly dwindling noise. Experimental results have shown the effectiveness of the proposal scheme, the experimental image quality has been improved about 1.7dB compared with the image processed by other method.2978|284|0更新时间：2024-05-08
- 摘要：In order to speed up object matching and location without degrading its matching accuracy in an image, this paper present a hybrid algorithm for coarse-to-fine object matching and locating method based on grid structure's image. First, the structure information of the image and template is extracted based on grid structure. The extracted information forms a coarse scale representation of the original image or the original template. On the coarse scale image some kind of similarity measure is used for matching. Then, the coarse matching position is guided to a small area of the original image by a guided searching strategy. The partial Hausdorff distance is applied to search the small area and determine the precise location in these small areas. The combination of above two steps of coarse matching and fine locating can reduces the computational cost without degrading the matching and locating accuracy. The proposed hybrid algorithm is compared to the partial Hausdorff distance matching without extracting the information of the image. The improvement of matching speed is remarkable. And the proposed algorithm is tested in integrate circuit images. It shows that the proposed matching and locating algorithm is very efficient.2716|363|0更新时间：2024-05-08
- 摘要：In computer drawing, generally, curves are represented by line segments and surfaces are tessellated up with small plane patches. To draw genuine curve and reduce computational costs as much as possible, curve-generating point-by-point has been studied in many papers, some good results have been achieved. But there is lack of effective approaches for fast surface-generating. If we suppose that surfaces consist of points of space lattice and the coordinates of points on a surface could be calculated one by one, then we would obtain the surface. There are already some algorithms for calculating polynomial curves on a plane, we can generalized them to space curves and surfaces. This paper presents an algorithm based on recursive formulas and algorithms for polynomial to generate a parametric polynomial surface patch point-by-point, in which only integral additive operation is employed, and properly choosing point number minimizes the computational costs, so it is quite efficient. The method in the algorithm can be modified for rational function, therefore it is of significance in fast generating parametric rational curve and surface(such as NURBS curve and surface), as well as in many other areas of Computer Graphics.2296|336|0更新时间：2024-05-08
- 摘要：As a new technique applying to protecting the copyright of digital productions, the digital watermark technique has drawn extensive attention. A method of embedding the watermark in digital images based on the discrete wavelet transform is proposed. The watermark used here is not the conventional patterns such as a pseudo-random sequence or a bit stream but a text watermark. The information which the text watermark contains is abundant and intuitionistic, also the watermark is robust. To ensure the security of the watermark and make the watermark be hard to be extracted, the watermark is scrambled with Arnold scrambling transformation before embedded into the original image. According to the different characteristics of the high and the low frequency components of the wavelet coefficients of the original image, more watermark information is embedded in the high frequency components while less information in the low ones. That is to say, by using the hierarchical structure of the wavelet, the watermark is repeatedly embedded in various places. Moreover, experimental results have proved that the method is robust enough to some image degradation process such as cropping, JPEG compression and sharpening etc. So it is effective.2588|316|0更新时间：2024-05-08
- 摘要：This paper is concerned with the problem of intelligent understanding of engineering drawing. The presented method is based on the engineering semantics classification. The contribution of different kind of semantics and the relationships among them in engineering drawings understanding are firstly discussed, namely, expression mode semantics is the premise, projection relation semantics is the kernel and dimension constraint semantics is the elicitation. Then the technology of engineering semantics obtaining and the corresponding integration are mainly introduced, thus the functional semantics expressed through the engineering semantics information could be obtained. In the end, complete details are provided to illustrate the identification of simple features and the combined relationship among the primitives using this approach. This method has been successfully used in the prototype system and the expected results have been got. It shows that the drawing understanding could be simplified, the recognition and processing could be speed up, and furthermore, the understanding reliability could be enhanced by using the technology.2739|350|0更新时间：2024-05-08
- 摘要：Instead of computely redacting the real world as a Virtual Reality(VR) system does, the user can see computer-generated 3D model superimposed on the real world with an Augmented Reality(AR) system. One of the key technologies of AR is the real-time 3D registration of objects in the real world. This paper describes a vision-based method for AR registration and its application in an AR system with See-Through Head Mounted Display(STHMD). The scheme has a simple structure, which employs one color CCD camera and several mark points with different colors. The real world coordinates of the markers are known, and their screen coordinates are obtained through color filtering of the input video stream. If all the markers are successfully tracked, a translation matrix and an orientation matrix can be uniquely determined using projective transform relations. The algorithm requires only linear equation solving with small calculation errors and little computation time consumption. Therefore it has good real-time performance characteristics and can be employed with either graphic workstations or ordinary PCs. The method can work at long ranges and is applicable to outdoor augmented reality systems.2631|381|0更新时间：2024-05-08
- 摘要：There are many researches of simulative images or growth of biology with Lindenmayer system, but there are few researches on retrieved Lindenmayer gene set from a long Lindenmayer string. If Lindenmayer gene set can be retrieved from Lindenmayer string which can be obtained from image or growth of biology with other algorithms, the image or the growth of biology will be easier compressed or identified. Retrieving Lindenmayer gene set from a long Lindenmayer string which can be reverted with the set of Lindenmayer gene is researched in this paper. Base on iteration statistic laws of single symbol in Lindenmayer string, some valuable relationships of single symbol's quantities in iteration of Lindenmayer system are provided in this paper. The statistic laws of single 'growth' symbol can used for judge whether the Lindenmayer string is a iterated string quickly. Following to the relationships, the Lindenmayer gene set can be found out faster, and the relationships can judge the Lindenmayer gene set whether is true. First it can be got how many research sets from the count of single 'growth' symbol of the Lindenmayer string, then the length of string branch can be obtained with each set, and at last the Lindenmayer string can be judged whether it is a iterate string, the iterate gene can be brought out if it is a iterated string. The relationships can be used with other retrieving algorithm (ex. Iterate gene algorithm) to run better.2346|347|0更新时间：2024-05-08
- 摘要：With the development of internet technology, internet GIS have applied popularly. Distribution, interoperability, component are the main characters of internet GIS. In order to access, manage, process and distributed geodata, the paper proposes a distributed geospatial data model, which is based on hypermap concepts and distributed database principle. According to the distributed geospatial model, various geospatial data including electronic maps, images, pictures, videos, attributes, hyper documents can be constructed in the model, and they are linked by a hyper link. Every medium can have one or more hyper nodes, which link another medium. A hyper node recodes some information including the IP address and catalog of linked medium. After clicking a hyper node, we can move to another IP address and browse linked medium. Furthermore, the paper defines data structure of different type hyper nodes and discusses its implementation. An internet GIS software-GeoSurf based on the distributed hypermap geospatial model have developed by our university. Many application systems based on GeoSurf have been built. These applications has shown many advantages of the distributed hypermap geospatial data model.2712|306|0更新时间：2024-05-08
- 摘要：World Wide Web Geographical Information System(WebGIS) is the combination of Internet/Web technologies and Geographical Information System,it makes the web publishing and sharing of geospatial information in the whole world possible and has bought great influence to the national economical development and our everyday life. The paper first analyses the problem existed in WebGIS, and then introduces the Scalable Vector Graphics,in short SVG, which is a standard of file format for describing 2D graphics on Internet, recommended by W3C. Considering the trait of SVG, the thought for constructing WebGIS based on SVG is presented, and then the method of coding the geospatial data and attribute data with SVG is gived, the system architecture of WebGIS based on SVG is described, the method in developing GIS's function using Adobe's SVGViewer 2 0 plug-in in client side is also discussed. The thought gived in this paper is proved feasible and effective with an experiment system. The author believe that with more research on WebGIS based on SVG,SVG will play greater role in publishing geospatial information, futhermore, the geospatial information web sharing will become easier.3970|336|0更新时间：2024-05-08
- 摘要：H 261 is an important algorithm for video composition in multimedia communication field. Therfore, it is very significant to implement four-image composition of H 261 algorithm. The traditional four-image is in analog domain. So there is a problem that the cost of implementation is high and its quality isn't very satisfying. In order to lower the cost of implementation and enhance its quality, this paper provides a kind of algorithm of four-image composition which is completed in compressed domain. This paper discusses in detail four-image composition methods in the two kinds of videoconference systems of both H 320 recommendation and H 323 recommendation. First, the composition principle is given. Then based on this principle, the composition algorithm and program flow chart are provided. Finally, its experiment's result is given and compared with that of implementation in analog domain. According to the result we can find not only its quality is enhanced but also its cost is lowered largely.2438|356|0更新时间：2024-05-08
- 摘要：Lifting scheme is not only a fast algorithm of existing wavelet transforms, but also a tool to produce new wavelet transforms. Multiwavelets are a new kind of wavelets and have found many new applications to signal processing. In this paper we present a framework of lifting scheme for multiwavelet transforms, in which the existing lifting schemes designed originally for wavelet (not multiwavelet) transforms can be used as building blocks. Thus we needn't concern about the internal structures and parameters of these building blocks. What we need to do is to choose different wavelets for different applications and then make use of the lifting schemes for these wavelets to build a new lifting scheme. This new lifting scheme is equivalent to a multiwavelet transform. We build such a lifting scheme by using the lifting schemes for D9/7 biorthogonal wavelet transform and then apply this new lifting scheme to image coding. We have achieved better results than that obtained by the commonly-used GHM multiwavelet image coding.2224|332|0更新时间：2024-05-08
- 摘要：An improved embedded zerotree wavelet image coding algorithm is proposed in this paper. It is an improved vision of Shaprio's EZW algorithm. First, a single list scheme is used in successive-approximation Quantization(SAQ), dispensing with double dynamic list of EZW algorithm. The single list make the scan process more simple and faster, and lead to better rate-distortion characteristic. Second, a fast algorithm based on spatial orientation tree recursive scan for the zerotree coding is designed. The zerotree symbols of one quantization layer can be calculated with only one access per coefficient, solving the rescaning problem completely. At last, a new Markov model based on quantization symbol field relativity for adaptive arithmetic coding algorithm is used, for higher compression efficient. Experimental results show that it peroforms better than the EZW algorithm in coding efficiency, maintains the fine-granularity scalability.2111|338|0更新时间：2024-05-08
- 摘要：In order to code and decode the stereo video sequences effectually and acquire excellent reconstructed images, a frame estimation and interpolation technique of stereo video sequences is presented in this paper. In stereo sequence coding, the left image stream is coded according to theMPEG-Ⅰ/Ⅱvideo coding standard. In the right image stream, only reference frames (I and P) are coded, the B frames in the right image stream are neither coded nor transmitted, they are acquired from the reference frames with the frame reconstruction and interpolation technique at the decoding end. The variable block segmentation is based on a multi-resolution quadtree-based motion segmentation. This paper also proposes a novel method to process the reconstruction of right B frame as matched-block and unmatched-block separately, the positions and content of these blocks are decided on the correlation between the blocks in right B frames and right I, P frames or left B frames. Anovel frame reconstruction and interpolation technique for overlapped blocks to avoid false matching and to fill the exposed regions correctly is also proposed in this paper. The simulate experiment results show that the average PSNR is about 1.25dB higher than the method mentioned in the refernce[1], Which is based on fixed blocks, and the reconstructed frames have a much better presentation on viewing perception.2522|327|0更新时间：2024-05-08
- 摘要：In video communication bit-errors, block-errors and macroblock-errors often happen owing to channel noises. They seriously affect video quality and QoS. At the same time retransmission can't match the needs of real-time transmission(RTP) when finding errors. Up to now many methodshave been proposed for error concealment in spatial, temporal, frequent domain and their joint domain. They all tr out to conceal the error with simple algorithm to meet the demand of RTP. In this paper we proposed an error concealment method using the frequent information of error blocks' neighborhood to decide how to conceal the error. It employs Markov random field to describe the relationships between pixels and between blocks. Based on the description we get the neighborhood of pixels and blocks by which to restore the error pixels and error blocks. According to the characters of human vison sstem different methods are used to restore different frequent parts. The high frequent parts, i.e. the deges of image are restored first, then the lower parts. In a least-squear sense linear fitting and multi-directional extrapolation are used to obtain optimal concealment. The experiment results have proved its robust error processing ability and good error concealment performance.2354|341|0更新时间：2024-05-08
- 摘要：Automatic target recognition(ART) using Synthetic Aperture Radar(SAR) imagery is investigated in this paper. The feature extraction problem of SAR imagery is first analyzed, then a matching scheme that incorporates relative distance and magnitude between features is investigated. Following the Birkhoff-von Neumann theorem, we relax the match matrix constraints from permutation matrix constraints to doubly stochastic matrix constraints. Via Lagrange multipliers and a barrier function, the constraints are incorporated into the objective function, and the matching problem is posed as a nonlinear optimization problem. Using a combination of deterministic annealing and softassign, the objective function describing the matching problem is minimized. Recognition is performed by comparing the costs of the matches between the test image and deferent pattens. To account for the difference in the number of features, the computed costs are first caled by the ratio of the number of features between the two images. The test costs image belongs to the class of the pattern with the smallest acaled matching cost. Experimental results show the power of this approach in SAR target recognition.2456|347|0更新时间：2024-05-08
- 摘要：SAR interferometry(InSAR) technology uses the phase information of single look complex(SLC) SAR data to get three-dimensional information of the earth surface. This technology has been in the front line of radar field since last decade. In recent years, this technology has been widely used in many remote sensing fields. To research the effect of wavelength to InSAR DEM, in the test site of Jiagedaqi district, Heilongjiang province, DEMs are generated from SIR-C/X-SAR C-band and L-band single look complex(SLC) data. There are some differences between them. Then the analysis are taken with interferometric coherence, phase unwrapping etc. So the relationships between accurate of DEMgenerated from InSAR and coherence, wavelength are illustrated, which also demonstrates the feasibility and affecting factor of DEM generated from InSAR. At last, 1∶50 000 topographic map is used to evaluate the accurate of DEM. The effects of ground control points’(GCP) accurate are analyzed to InSAR DEM.3157|409|0更新时间：2024-05-08
- 摘要：Desertification is one of the most serious environment problems in the world today, The generation, development and reversion of desertification are caused by the results of comprehensive influences from the climatic and environmental change, and human activities. The classification of desertification is one of the key steps in desertification research. Terrain roughness and vegetation growth are important influence factors of desertification. But it is very difficult to describe the terrain roughness in remote sensing image. In this thesis digital elevation model(DEM) was used to express the Terrain roughness of research area and normalized difference vegetation index(NDVI) were used to describe the vegetation growth. Terrain slope(SLOPE) image (produced by DEM) and NDVI image (produced by TM image) were used to fuse with original TM image. The fusion image greatly enriched the desertification information. According to n- dimension separation analysis to fusion image, the separation in templates of different kinds and degrees of desertification was increased greatly. The classification result shows that the accurate of classification was greatly improved through information fusion.2777|365|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