摘要:Astudy of Image Coding based on Wavelet Transform is made in this paper. On the basis of a systematic analysis of the quantization noises and the reconstruction noises of the sub-images.an optimal bit allocation scheme is presented .which minimizes the system reconstruction distortion at a given bit一rate. When uniform quantizers are used .it is shown that the quantization step of a succeeding level in the wavelet pyramid is a hall of that of its predecessor level. The simulation experiments demonstrate the con- elusion oI this paper and give satisfactory results.
关键词:wavelet transform;Noiscs analvsis;Optimal bit allocation
摘要:Alter analyzing existing methods of constructing hierarchical model.this paper present a new simplification method for polyhedral model based on feature angle criterion.It is one of the important techniques for constructing hierarchical model.It consists of three parts:1. preprocessing data including creating topological structure and some basic computation;2. deciding whether a vertex should be deleted according to feature angle criterion;3. filling hole using a modified local greedy triangulation algorithm and updating topological structure of model.At the end of the paper.is given a sample of hierarchical model through defining different feature angle threshold.
摘要:This paper introduced briefly the principle oF 3-D Color Scanning System.and analyzed the facial features as well as its reflectivity in detail.From above analysis a divisional algorithm.embeded the framcwork of expert system,was presented.
摘要:In this paper,a kind of description and recognition method for 3D objects is presented. The multiple description approaches are firstly presented,including general description.unit attributes description and relational description. In the procedure of recognition.after creating the primary assumption about object idcntilication,authors present three prunning algorithms to prune the search tree. Then,the distance measure is used to realize the final recognition. Finally,some experiments are given out.
关键词:computer vision;Relational data structure;recognition;Isomorphism
摘要:In the process of 3-D ce11 reconstruction of serial sections through medical tissues,the registration of cell profiles in consecutive sections is a critical step. This paper presents an automatic registration then method that builds up the equivalent ellipses of cell profiles with the help of the moment of inertia and then matches these ellipses to obtain the required geometric transformations.This method has been applied to the 3-D registration of megakaryocyte cells in human bone marrow specimens with satisfactory results.
关键词:3-D Reconstruction;Image registration;Geometric transformations;Serial sections;Ellipse of inertia
摘要:Quadtree representation of DEM is of importance in DEM compression and terrain visualization. An improved quadtree is proposed.which has the advantages of both an ordinary quadtree and a pyramidal structure. Each node of the improved quadtree is added with a maximum height value and a minimum height value,and the whole quadtree is threaded,thereby making raytracing efficient in terrain visualization.
摘要:A classified VQ (CVQ) tehchnique.based on the edge-oriented classifier.can reduce the edge degradation as well as the encoding complexity. In this paper according to the WT coefficients feature of a still image transformed,we propose a classification algorithm in the wavelet transform(WT)domain for the CVQ. This approach can be easily extended to the wavelet transform coding technique. Coded images based on this classification are shown to enjoy considerable bpp reduction over three existing methods in the same or a little higher PSNR conditions.
摘要:The paper presents a new fractal image compression scheme,which merges the range blocks as much as possible in a 4-connected extending way to decrease the amount oI the transformations under a given error tolerance,with a quick searching algorithm. Through rearranging the storing order of the transformations according to their edges,we only need to record their surrounding rectangles for those complicated ranges blocks .whose shape can be recovered using a paint-like rendering method in decoding.Our experiment on Lenna is pleasing.the compression ratio reach 41 at a PNSR of 29.63dB.
摘要:Egdes are of very visual importance.In this paper a novel visual primary component(VPC)model is proposed.By using the B-spline wavelet transform which has the advantage of fast algorithm.and according to the edge information such as strength,length and integrity.an effective algorithm for the extraction and coding oI VPC is developed in the paper.Computer simulation results are given as well.
摘要:wavelet transform is adopted to decompose an image into a group of multiresolution sub-images.An improved coding scheme is developed which have considered the gray distribution properties of sub-image and human being's vision.image data have been reduced farther by using dual-pyramid data structure for sub-image coding.which have great similarity with each other.The mean compression rate and signal-to-noise ratio for 10 cow's image is 26.2 and 30.6.respectively.The reconstructural quality of image is high.
关键词:Wavclct transform;Image coding;Pyramid data structure
摘要:Graphics clipping and covering is one of the key points in computer graphics field. Alter more study and practice.a new algorithm is give.which is for clipping and covering in circular window.Adopting shilling principle to do multiply and division operation.we have got a more speed in clipping and covering.raised the capability of computer in graphics processing.
关键词:Clipping and covcring;Circular window;“hrue/false”field
摘要:This paper presents an automatic module wrapping tool,Module Extender(ME).ME is a relatively independent part of the General Interactive Visualization Environment(GIVE).GIVE provides a convenient module-level programming environment.And it supports users to set up a visualization map by selecting appropriate modules from the Module Librarian.Thus,the contents of the Module Librarian decide the system's application fields.By the visual operative interface ,ME supports users to directly build modules from existing algorithm codes into the Module Librarian on the basis of the API functions and the module's syntactical definition.This makes the system an opened environment and adaptable to different application fields.
摘要:Using sparse range image and gray level image in combination.features extracted from the gray level image can be assigned conditionally with corresponding range values by means of a projective mapping.It is called rank information propagation transformation.The range information propagated in this way can be used to calculate the Euclidean invariant of corresponding surface planar patterns and hence to generate initial hypothesis.Since initial hypotheses are extracted based on reliable information.and hybrid control strategy with feedback is used.the proposed method can be used in practice situations.In this paper the principle of the method and its implemental issues are addressed.and some preliminary experimental results are presented to demonstrate its effectiveness and efficiency.
关键词:Model-based vision;Object recognition;data fusion;Projective transformation;Range information propagation
摘要:Electrical capacitance tomography(ECT)provides a new way for imaging the crosssection of process pipelines and vessels containing components with dillerent permittivities.This paper describes an emulation method used for two component flow imaging.When the 1 dielectric distribution of the flow pattern is given.the capacitance between any two electrodes of the transducercan be calculated through a finite clement method (FEM).Then the reconstructed imaging can be gained with the contribution of Applicanon Visualization System and back project algorithm. Finally.three typical two-phase flow regimes.which encountered in practice .are reconstructed on a workstation SGI indigo 2.
摘要:This paper studies an algorithm for estimating camera zooms.rotations and pans from video sequences.Three new options are presented:a simple 3-D model of camera motion.image pyramid.and selective iteration.This algorithm can be used in video compression.paranomic imaging.content_based retrieval for video.and virtual reality.
摘要:Center weighted median filters has better characteristic on edge preservation and fast filtering.but fail in noise depression. This paper deals with a method of improvement if center weighted midian filter.and the mutilevel center weighted filtering method is presented which makes the filters ellicient in edge preservation and noise depression.and faster filtering as well.
关键词:Median filters;Center weighted median filting;Image processing edge preservation;Noisc depression;fast algorithm
摘要:This paper presents our viewpoints of content-based image database.The construction of content-based object-oriented image database on the current RDBMS is mainly discussed.Moreover.a layer model is presented and a data structure of image database is given.
摘要:Since its extreme irregularity and complexity of real topography.3D realistic terrain generation is one of the most challenging problems under some given condition.In this paper.real terrain is regarded as a hierarchical and stochastic process.through the integration of fractal geometry.random number generation and our established terrain model for description of terrain structure.a realistic and feasible hierarchical method for 3D terrain generation is proposed and established. The method we propose is formed by three stages:determination of initial simulation condition.formation of controlling terrain sketch;and addition of localized terrain details by fractal interpolation. Any kind of wonderlul terrains can be simulated according to some given condition such as required topographic parameters(average height,standard deviation.average slope.terrain frequency.terrain relative length etc.)or some known characteristic points by the use of our method.Experimental results show its ellectivity and realiability of this method.
摘要:Spatial Engineering Design is a complex system engineering which is strongly characterized by large volume of data,complicated procedure and invovling operation both on spatial-referenced inlormation and thematic information.Geographic information system is a powerful tool for the management and operation of both spatial and attribute information.The combination of these two field are explored in details in this paper.The hardware and software structure.function requirement and the characteristics as well as the the principles of construction of spatial engineering design GIS system are discussed.The conclusion is that the efficiency and the quality of the spatial engineering design is substantially improved by adopting GIS technologies in the computer assistant design of spatial engineering.