Three dimensional face reconstruction based on average face model and photometric stereo

摘要

Three dimensional reconstruction of the classical photometric stereo is based on the reflected light of the object. The reconstructed results have good high-frequency characteristics. However, owing to the uneven distribution of the reflected light, the low-frequency characteristics of the reconstructed results are poor. In this paper, the average face model has been used to improve the low-frequency characteristics of the classical photometric stereo. By fusing the low frequency of the average face model and the high-frequency of photometric stereo, the accuracy of the three dimensional face shape has been largely improved. The face images in YaleB database have been used to verify our algorithm. Experimental results of the proposed approach are more effective and accurate than those of classical photometric stereo.