复杂环境下的快速人脸检测算法

A New Algorithm of Fast Face Detection under Complex Condition

使用一种新的假面变换和模板匹配的人脸检测算法,能够对复杂环境中大小不等的人脸进行快速、准确的检测.算法首先通过假面变换来预测人脸位置上边沿的中心,然后对图像中的预测位置进行模板匹配,设计了一类对照明变化、噪声干扰具有较强适应性的模板匹配方法,最后对匹配结果进行验证,确定人脸准确位置.采用多种环境下的大量图片进行实验,结果显示该算法具有较快的检测速度和较高的准确性及鲁棒性.

A novel algorithm of face detection based on mask transform and template match makes face detection more accurate and faster in different face sizes with complex condition. The center of the face edges is predicted by mask transform firstly, and the candidate position is then matched with template, in which an adaptive template match method which is immune to illumination and noise is introduced. Finally, the accurate position of face is verified on the basis of the match results. The experimental results on many images in various conditions show that the algorithm is fast, accurate and robust.