Conventional contrast enhancement methods have two shortcomings. First, most of them do not produce satisfactory enhancement results for face images with back-lit or front-lit. Second, most of them need transformation functions and parameters which are specified manually. Thus, this paper proposes an automatic and parameter-free contrast compensation algorithm for color face images. This method includes: RGB color space is transformed to YIQ color space. Fuzzy logic is used to classify the color images into back-lit, normal-lit, and front-lit categories. Image illumination analysis is used to analyze the image distribution. The input image is compensated by piecewise linear based compensation method. Finally, the compensation image is transformed back to RGB color space. This novel compensation method is automatic and parameter-free. Our experiments included back-lit and front-lit images. Experiment results show that the performance of the proposed method is better than other available methods in visual perception measurements.
如有問題歡迎與系統管理員聯繫 
