3D Shape Feature Extraction And Matching Based On Voxelization

Researches of three-dimensional shapes have experienced decades of development and have achieved substantial achievement. Which has provided great convenience for applications in many fields such as games, health care, virtual reality, three-dimensional printing, molecular biology, industrial design, security, etc. With the burgeoning development of Internet, cloud computing and big data, it’s forseeable that three-dimenstional shape application will become more and more frequent. Thus, it is of great significance to research on three-dimensional shapes. This thesis proposes a method of three-dimensional inner-distance extraction based on adaptive voxelization. Further, studies on shape matching and retrieval based on three-dimensional inner-distance are made. This thesis mainly centers on below three aspects to launch.Firstly, this thesis will talk about three shape feature extraction methods, and set out some classic three-dimensional features and extraction algorithms proposed recently, such as Reeb Graph, Shape Context, Shape Diameter Function, etc. And evaluate their advantages and disadvantages in different application fields.Secondly, the main contents will be talked about in detail, including representation of three-dimensional adaptive voxelization which can vastly reduce the computational resources when calculating the inner-distance, further based on inner distance a method of extracting average inner distance will be introduced, shape key points extraction algorithm which based on the inner-distance and two extraction methods will be talked about, shape matching algorithms which play a vital role in the whole research and shape retrieval which based on shape matching result. In this section, we will do many experiment results to varify above theories and some graphic libraries are used including the Princeton Segmentation Benchmark, TOSCA, SCAPE and SHREC07.Thirdly, to conclude this thesis and make some prospect of this fantastic and lifeful field which in my opinion will achieve great success in the near future.