基于小波的光照补偿及其在人脸识别中的应用

Wavelet-based Illumination Compensation and Application to Face Recognition Using Eigenface Method

提出了一种基于小波分析的人脸图像光照补偿技术,通过舍弃图像中的低频分量,仅用高频分量进行图像重构,便可较好地消除原始人脸图像中光照的影响.并将该技术应用于特征脸法人脸识别,在Yale库和自建库图像中,实验测得在较大尺度的情况下,图像识别率都有明显提高。

Engineered for human face recognition and based on Wavelet transform, a novel algorithm for the illumination compensation is proposed. By removing the low frequency components in the face image and reconstructing the image using only the high frequency components, the undesired effects of lightings on face image recognition can be greatly reduced. Experiments are carried out using the Yale face database and our own face database, and the testing results show that, for the lighting-compensated images, the recognition rate using the Eigenface method is increased significantly.