Reserch And Realization Of3D Face Retrieval Based On Significant Analysis

With the continued improvement of information security, the3D face whichcontains rich information and high recognition accuracy has become a hot researchdirection in human face recognition. Study of this typical3D model helps to solve thegeneral3D models recognition academic problems.Currently,3D face retrieval methods focus on local geometry features or overallcharacters. They are not the same as human logic. In this paper, we simulate the waypeople observe things and process information. Visual saliency reflects that wehuman always focus on the significant region and ignore the unimportant parts whenwe looking at things in a scene. We detect the visual saliency analysis for facerecognition.We firstly proposed algorithms based on the analysis of depth, color, normal andfeatures fusion. It combined the significant analysis method of2D image with3Dhuman face information and applied the significant analysis into3D space.Secondly, the proposed algorithm was applied to the pixels on the entire surfaceof3D human face. It considered the impact of all the points for the target region andcalculated the value of all points. Then abstracted the target parts and calculated thedistance of the samples.Finally, we devised a3D face recognition algorithm based on partialcharacteristics. It reduced the computational and space complexity at the same timeand got a good result91.3%for accuracy rate. The method also reached a good recallrate with showed that our system has a better ability to find all relational samples.