3D Foot Reconstruction System Based On Disparity Map

Nowadays, as people's living standard improved, the pursuit of high quality and personality are increasing more and more intensely, the customized shoe is in tune with the "customization" of this trend. The first acquisition of customization is to get 3D foot models. This thesis, based on multi-view stereo reconstruction technique, provides an solution to acquire the foot shapes from the requirement of customization. The main contents focused by this thesis are as follows:1. The principle of stereo matching and overall system solution are introduced in the paper. Firstly, the images are preprocessed to remove noise. And then images are rectified according to camera params. The fallowing is converting the problem of searching the best matchs on the basis of similarity measurement to matching optimization. Finally, we need to remove outliers.2. Stereo matching based on dynamic programming and Graph-Cut are elaborated and implemented, and the modeling time is greatly reduced by using two level image pyramid. The reconstruction of the foot is compared with the model based on feature points reconstruction technology to verify the validity and error.3. Based on the research achievement, the 3D foot reconstruction system is designed and implemented. The output of this system will be the input of our personalized shoes system called ArtShoe for personalized shoes applications.Finally, the current work is summarized and the further work is put forward.