Use the image illumination correction, reduce the dimension of face images and different lighting conditions, the use of human face images improved the BP neural network for recognition.

Secondly, the anterior segment image, which was acquired by the method of direct focal illumination, was analyzed automatically to give some physiological and pathological parameters.

A new regression algorithm of an adaptive reduced relevance vector machine is proposed to estimate the illumination chromaticity of an image for the purpose of color constancy.