Font Size: a A A

3D Model Retrieval System Based On Freehand Sketch

Posted on:2013-01-27Degree:MasterType:Thesis
Country:ChinaCandidate:H L ZhangFull Text:PDF
GTID:2248330362470877Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the development of computer graphics technology and three-dimensional model widelyused in many domains, it is particularly important that how to quickly and accurately retrieve3Dmodels what users intent from the database in the process of design. Thus, in the field of multimediainformation the research of content-based3D model retrieval has been widespread concerned.Currently, the research of3D model retrieval still has some problems in the freely expression of theuser’s search intent and the feature extraction of the model, these problems reduce the ease ofhuman-computer interaction and affect the retrieval performance and results of3D model retrievalsystem.Accordingly, in this paper, there is a study on how to get user’retrieval intention and the featureextraction of3D model. Users can use freehand sketch to freely express their retrieval intention.bycalculating the degree of the similarity between the feature extracted from sketch and the feature ofprojection, users can achieve their retrieval purpose.On this basis, the framework and implementationof the retrieval system based on freehand sketch are proposed in this paper.The main researches are:(1) Design a polygon fitting algorithm based on feature point. After pretreatment in the sketch,use the algorithm designed in this paper to fit the profile of sketch. The algorithm makes up for thetraditional inadequacy of fitting algorithm and improves the accuracy of the feature extraction.(2) Researched and improved feature extraction methods of three-dimensional model. Obtain themain plane used to project by calculating according to the properties of3D model.uses the improvedSUSAN algorithm to detect the edge of projected image, Obtain the feature used to describe3Dmodel from the profile which get by the profile tracking algorithm.(3) Proposed a retrieval model based on Bayesian network and relevant feedback. Use featuredata extracted from the profile to train the model, according to the characteristics of user-drawnsketches, users not only can achieve3D model retrieval by the Bayesian network inference, but alsocan improved efficiency and3D model retrieval accuracy by relevant feedback.(4) Completed the design and the development of main function module of3D model retrievalsystem based on freehand sketch, highlights the technology used in such as the pattern recognitionmodule、feature extraction of three-dimensional model and so on.
Keywords/Search Tags:Freehand Sketch, Sketch Recognition, 3D Model Retrieval, Bayesian Network, ProfileExtraction, Feature Vector
PDF Full Text Request
Related items