A Modeling and Comparison Method for Foot Based on Passive Stereo Vision

An accurate 3D modeling and comparison method for foot is presented to solve the difficulty in acquirement and searches of 3D foot in the exploitation of shoe last．The stereo image pair is acquired by a pair of calibrated cameras．Global disparity of foot is estimated at first．After the image pair is preprocessed automatically, ausing non-contact information-collecting technique and phase-correlation method, awhich is based on complex wavelet, asurface matching with sub-pixel accuracy in the frequency domain is carried out．By taking order matching constraint, acontinuity constraint and correlation constraint together, a 3D point cloud of foot is reconstructed and 3D model of foot is built by mosaicking point cloud at various angles automatically．After normalization, ahigh-order moments are compared, awhich is useful for the subsequent shoe last design．Experimental results show that with our proposed method, a foot information collection is fast and reliable, a and entire reconstruction can be done without manual intervention．