摘要:Along with the analysis of color image retrieval using accumulative histogram, the concept of local accumulative histogram is first introduced in this paper. In addition, the distribution of similar colors in HSV space is studied and a new dividing method for similar color regions is obtained. Based on these studies, four new algorithms of image retrieval using local accumulative histogram and new dividing method for similar color regions are proposed. The performances of these new algorithms are proved much better than the algorithms using accumulative histogram as shown by the retrieval experiments with real images.
关键词:Image retrieval;HSV space;Accumulative histogram;Similar color region
摘要:On the basis of the 3-D model of a solid of revolution, the geometric properties and algebraic formula of its limbs and cross-section curves under perspective projection are analyzed in detail. A new technique to recognize a solid of revolution and to extract its 3-D vision information from its a single viewgray-level image is successfully developed. Experiments have
demonstrated that the technique is exact, efficient and reliable.
摘要:An efficient data structure for realizing spilt-and-merge image segmentation method is devised in order to overcome the problems in the current methods. This data structure can very easily realizes each step in the idea of split-and-merge and dynamically assigns memory, it results in a remarkable decrease in computational time and an improvement in memory space. The experimental results show that this method developed is very efficient in realizing split-and-merge.
摘要:In this paper, we propose a new feature extraction method for handwritten Chinese character recognition. A Chinese character image is decomposed into four directional sub-images based on the analysis of its wavelet transform. With elastic meshing technique, features are extracted from the four sub-images respectively. Experimental results show that the proposed approach is very effective.
关键词:wavelet transform;Handwritten Chinese character recognition;feature extraction;Elastic meshes
摘要:In order to save space of newspapers and magazines in an electric way, it is important to classify their contents automatically after they are scanned into computer. An algorithm which can establish optimal trees for the description of document structures in document segmentation is given in this paper. We can get better understanding of the document structures by using this method.
关键词:Document understanding;Document structure;Optimal tree
摘要:In this paper, three methods of fast half pixel search of motion estimation have been proposed. Experiments show the three methods have good MSE performance and decoded picture quality; Analysis show that the slowest method can save compute times about 29.17% and the most fast can save about 58.33%. The proposed methods can be as candidate algorithms for the real time implementation of motion estimation half pixel search.
摘要:This paper compares theKHandKJsign images in the context of the zero-thresholding at single and multiple scales. A function in the real number field is given for implementing the transformation from theKHcoordinate to the KJ coordinate. It is shown that the consistent zero-thresholding inequalityεK≥2H-εH+ε2Hmaintains effectiveness for both KH and KJ sign images. Therefore, from the viewpoint of zero-thresholding, theKHsign image is virtually a better choice for surface segmentation.
摘要:The iterative two-fork tree of perimeter and area on a modified Koch curve island is acquired,and from this, the formulae for calculating perimeter and area is obtained.It shows that the theoretical value of fractal dimension could be acquired from perimeter-maximum diameter relation for the fractal structure.It also shows the relation between that the critical
nested tier of both the theoretical value of the fractal dimension and the ideal value of the fractal dimension,which meets the precision of auto-image analysis obtained from perimeter-area relation and parametersaandncould be ascertained. It is considered that the perimeter-area method is effective to obtain ideal value of fractal dimension in determenation of fractal dimension.
摘要:A filling algorithm of circular window for convex polygons is presented. It serves the needs of filling with clipping as well as clipping-only functions.
摘要:The course of human computer interaction technology and its tendency is analyzed in this paper. An open three dimensional reference model of UIMS is presented so as to solve the problems arising in the process of development of UIMS.
关键词:human computer interaction;UIMS;Three-Dimesional Reference Model
摘要:Harmonic maps can construct the mapping relation between two regions. With this method, the geometric distortion is minimized. Harmonic maps can be used to do the parametrization of mesh over a simple domain. In this paper, we present the construction method for harmonic map. And its application in computer graphics is discussed in detail.
关键词:Harmonic map;mesh simplification;level of detail;texture mapping
摘要:Automatic topological relation generating algorithm influences directly the performance of such spatial operations in GIS as spatial relation establishment and query. Double neighboring point judgement method can be adopted so as to decrease computing complexity greatly in two key stages of this algorithm -determining both direction of polygon and point-in-polygon relation.
摘要:The occluding edge and local shading information are the important cues for surface shape estimation. In this paper, we use the occluding edge sequence at different view points to acquire the 3-D geometry relations of the responding points on the object surface. And then integrating the local shading information analysis, we can compute the depth of the unexposed
areas on the surface. Eventually, by interpolating we can acquire the 3-D model of the surface and implementthe objects modeling.
摘要:A visual computing model based on attention mechanism with the following characters is proposed in this paper.First, it imitates the characteristic of zoom-lens, eye movement and head movement of human vision system. Second, the retina character is modeled by a new nonuniform sampling method with changeable structure. Third, the bottom-up process driven by data and top-down process driven by knowledge are combined. A novel knowledge representation method and a fixation movement controlling mechanism based on depth first search tree are proposed as well.
摘要:This paper describes a new offsetting method for detecting cutter interference in multi-surface NC machining, and generating its tools paths. The method can be applied in 3-ax is NC machining with a cutter of arbitrary shape.