Analysis of Discriminant Features in Fourier Domain Compensating Shadow Areas on Facial Images

摘要

This paper proposes a novel compensation method for shadow areas on the human face by analyzing magnitude components of the facial images on the Fourier domain. A feature extraction algorithm based on PCA+LDA is utilized to extract features of the magnitude components, and create a compensation handling mask which identifies and catalog the necessary compensation levels of darkness pixels. The proposed algorithm is applied to the facial data for the face recognition. The experimental results demonstrate that the proposed algorithm can effectively compensate for the dark areas in the human image as well as improve the accuracy of face recognition.