摘要:Because of it's difficult to match the object model feature points in Multi-Vision system。In this paper a method which obtains moving-object 3D-information based on single-vision be given。This method based on object-model、camera-model and object-shape model applys a kind of special image-processor which may realize the real time extraction、matching of the object model feature points readize real time obtain moving-object 3D information。
摘要:A new tendency to develop machine vision, invariant vision, is summarized in this paper. The paper has given major contents and applications of vision invariants in broad outline, as well our practical research engaged and engaging in they.
摘要:Based on difference code models and multipoint group models on a special chain code direction, an algorithm is developed for smothing and following of binary image contours.
摘要:A representation of line draw is put forword which supports the bottom up stratrgy for object recognition. The proposed method is built upon the hierachy of line junction area, whose lower level representation is used in pre recognition for its statistical feature, and higher level supports object recognition based on the model constraints.
摘要:An algorithm for detecting man made objects from natural background in space sensing imagery is presented by applying multi resolution analysis based on the wavelet transform and features fusion technique. Experimental results on SAR images have shown that the algorithm is effective and applicable.
摘要:A model for computer image generation is given in this paper. Both geometric and illumination modeling can be simulated by prediction and pseudo noise sequence in this model. The generated image of natural scenery, e.g. cloud, water area and grass land, is rich in details which indicate that the model is suitable for computer simulation.
摘要:A model for computer image generation is given in this paper. Both geometric and illumination modeling can be simulated by prediction and pseudo noise sequence in this model. The generated image of natural scenery, e.g. cloud, water area and grass land, is rich in details which indicate that the model is suitable for computer simulation.
摘要:Wavelett ransform is a new powerful tool developed in recent years,which includes mathmatics、scientific computation and signal processing. Its key is multiresolution analysis. This paper completely reviews the process of the wavelet development as multiresolution tool、difinition of wavelet transform、wavelet decompositions of signals and images、and the multiscale、local time frequency and orientation characteristics of multiresolution image decomposition and so on.
摘要:The method of inverse iterating correction are used to correct the error of VQ coding. It improves the quality of reconstructed images base on wavelet transform and VQ coding. It's SNR improves 3~8dB and the effect is obvious, especially for special images under the same CR.
摘要:The method of inverse iterating correction are used to correct the error of VQ coding. It improves the quality of reconstructed images base on wavelet transform and VQ coding. It's SNR improves 3~8dB and the effect is obvious, especially for special images under the same CR.
摘要:Based on the analysis of neural network competitive learning algorithms and fuzzy C means algorithms, two new fuzzy competitive learning algorithms (FCL1 and FCL2) have been proposed, and the fuzzy membership functions have been investigated. Both the theoretical analysis and the experimental results have proven that the proposed fuzzy competitive learning algorithms are efficient for image vector quantization compression coding.
摘要:This paper studies signals and images obtained by a subsurface radar,presents a adaptive filter and a linear predicate data expand discrete fourier transform which realize the pulse compression.
摘要:In this paper, the algorithm of globally optimal multistage vector quantization based on stochastic disturbing (SD MSVQ) is presented. Experimental results show that not only computation and storage is significantly reduced, but also obtained a good reconstructed image at a high compress ratio.
摘要:The morphological filtering is an effective method to restore the root signal under noisy background. A class of omnidirectional multiple structuring elements are defined in this paper. Consequently, a new morphological filter—The mor phological filter with omnidirectional multiple structuring elements is proposed. A standard test image is filtered in noisy environment. The results of simulation show that the new filter performs better in noise-suppressing and detail-preserving.
摘要:Taking advantage of the threshold decomposition property of stack filter, a stack weighted median filter is presented based on the threshold decomposition architecture. It allows parallel process and VLSI implementation, and is effective in the application of image process.
摘要:This paper introduce a sort of simple Algorithm for Julia set and Mandelbrot set drawing. It don't require high performance computers, but can be realized on common computers.
摘要:This paper combines the methods of image process with the methods of rendering realistic pictures and puts forward a new method for 3D reconstruction of fundus images.It emphatically expounds how to get 3D information from 2D fundus images and how to realize 3D reconstruction of fundus by realistic rendering technology.
摘要:A new method for visualizing fractals in 3 D is put forward in this article.The image is generated in evolutionary manner and, by utilizing existing techniques for visualizing traditional geometric object,the realistic effects in display can be obtained.
摘要:The aim of optimal layout on rectangular or irregular sheets is to cut the shapes out of the sheets in such a way as to minimize the amount of waste produced. In this paper,a method to cluster irre gular shapes in rectangular modules is proposed.
摘要:Based on the framework of the curve extraction algorithm developed in Ref. ,and using the convergence mapping from the tripe point subset to the elliptical parameter set provided by the geometric properties on the pole and polar line of an ellipse, a new method to robustly and directly extract the ellipses of targets is presented. Compared to the old indirect method developed in Ref., the new method has the attractive advantage of making sure the results of every mapping to be the valid elliptical parameters. The extensive experiments for the images exactly simulated by PC computer and for the real images have demonstrated that the method is very robust, fast and exact.
摘要:This paper presents a kind of dynamic searching image reconstruction method that satisfies several criteria. From the image space essence, this paper establishes three criterion functions. The entropy function derivative isn't a linear function, this paper presents a method that transforms it to a proximate linear formula and gets a iteration formula.The dynamic searching method, to the best, avoids the adjustment overdose or insufficiency problem in the course of iteration that exists in many iteration methods.
摘要:In this paper, a simple but effecient algorithm for automatic mesh generation over arbitrary polygonal domains is proposed. The algorithm can generate three different mesh topologies: an all triangular element mesh, a mixed mode element mesh(which consist of both quadrilateral and triangular elements) and an all quadrilateral element mesh. If an all triangular element mesh is the result mesh needed, the method the method can work just as other general triangulation method. If a mixed mode or all quadrilateral element mesh should be generated, the method can generate rectangles or squres as more as possible.
摘要:This paper presents a model applying for the background in computer animation,which represents waving trees in winds.The model combines vector composition technology with the recursive growing process of the 3D trees and achieves easily animation pictures of better effect.
摘要:This paper presents a model applying for the background in computer animation, which represents waving trees in winds. The model combines vector composition technology with the recursive growing process of the 3D trees and achieves easily animation pictures of better effect.
摘要:The system of Graphics Digitizer Manage Software(GDMS) is a Open-system based on graphics database、order database、Attribution database and sound database. The system is common, it may bocame a graphics input sub-system of others system for managing graphics and image.
摘要:In this paper, the radiated waveshape of an electric dipole is shown on the screen by using WINDOWS 95 at different time and this result gives the visibility of the theoretical model.