中文摘要：

数字人像识别技术的研究成果有广泛的应用前景,但光照问题是影响人脸识别技术关键的因素之一。对未知光源参数且具有左右光照不均匀的正脸图像,根据小波不同子带图像上的频率特点,提出了一种小波子带图像上的人脸光照归一化方法。在Yale B人脸库上的实验表明,新方法能较有效的解决人脸识别中的左右不均匀光照恢复问题,使人脸识别率有显著提高。

英文摘要：

Research achievements from the technique of digital face recognition have bright future, but problem from illumination is a crucial factor in face recognition. For recognizing image with unknown parameter from the source of illumination and asymmetric illumination frontal image, it was proposed a normalization method of facial illumination based on wavelet Subband image, which concerns different characteristics of low subband and high subband. Experiment results from test on Yale B face database indicate that the new method can solve asymmetric illumination restoration problems in face recognition very efficiently, and greatly improves the recognition rate in face recognition.