人脸检测中的图像光照处理研究

在人脸检测过程中,由于光照条件的不同对人脸的检测结果影响很大.提出了一种基于灰度统计模型的对数变换和局部归一化相结合的复合光照处理算法,并且有效的消除了局部归一化所产生的虚假边界.该算法不仅可以调节光照条件较差的图像,对于光照条件较好的图像,也产生较少负面作用.通过与其他常用的一些光照调节方法也做了相应的比较,实验表明,此算法能有效地改善图像质量,减少不同光照条件对人脸检测准确率的影响.

Studies on the image illumination compound adjustment of face detection

Different environment illumination has a great impact on face detection.This paper presents a solution including logarithm transforms and local normalization based on gray statistical model which can also reduce the illusive boundary.The algorithm can not only improve the poor lighting condition image,but also decrease the negative effect for the image with good lighting condition.The comparison between this algorithm and other normal algorithms indicates that it can efficiently improve the image lighting q...