Face Recognition Based on Projected Color Space With Lighting Compensation

In this letter, we propose a novel color space conversion method called adaptive projection color space (APCS). This method includes two portions: adaptive singular value decomposition and an inner product conversion algorithm for color images. We employed images from the Color FERET and CMU-PIE databases for training and experiment. The results revealed that the recognition rates from our proposed APCS approach were higher than other color spaces and those of methods proposed in relevant studies.