Eyeglasses removal from facial image based on phase congruency

Abstract

Studies have shown that eyeglasses have negative impact on the face recognition. Based on the investigation of face features changes introduced by eyeglasses, and the fact that phase congruency is not sensitive to the intensity and contrast, we proposed an eyeglasses detection and removal algorithm based on phase congruency and progressive inpainting. Experimental results showed that the proposed method significantly improved recognition rates, and exhibited strong robustness against variations in facial images.