Research On Sub-graph Learning Based 3D Model Retrieval

The graph matching problem is a basic problem in the multimedia field,and plays an important role in the fields of target detection,target tracking and 3D model retrieval.A graph matching algorithm based on subgraph learning is proposed in this paper.The influence of outliers on matching results in traditional graph matching problems is reduced,and the matching precision is improved by introducing penalty terms to optimize the selection of subgraphs.This paper also proposes a 3D model retrieval algorithm based on graph matching based on the graph matching algorithm of subgraph learning.This paper extracts two-dimensional views of multiple angles from 3D models,converts the problems of 3D model retrieval into graph matching problems,and then uses the algorithm of sub-graph learning to reduce information redundancy and obtain the similarity between models.In addition,this paper builds a complete massive 3D model retrieval system using large-scale crawler and distributed technology.This system can realize automatic data crawling,cleaning and index construction.The system utilizes distributed technology to enable it to carry a large number of concurrent retrieval requests,which is extremely practical.In order to verify the graph matching algorithm based on subgraph learning,a lot of experiments were carried out on synthetic graphs,real graphs and 3D model datasets.The correctness and effectiveness of the graph matching algorithm based on subgraph learning are verified by comparison experiments with other graph matching algorithms.Finally,the main interface and usage of the 3D model retrieval system implemented in this paper are introduced.