基于改进的二值形态学人脸识别算法

摘要

针对人脸检测识别中存在的识别率低、图像易受光照、亮度等因素影响问题,提出了一种肤色检测算法与二值形态学处理算法相融合的人脸识别算法.首先对捕捉到的图像进行YCrCb模型检测减少图像中亮度的影响,运用HSV模型检测算法降低光照对图像的影响;然后将得到的检测图像转换为二值图像,进行二值形态学去噪处理;最后得到了人脸识别图像.实验结果表明,所提出的算法对于不复杂、较为复杂和复杂人脸图像的检测识别率分别达到了98.6%、93.3%和90%.%In order to solve the problems of low identification rate and vulnerable to illumination and brightness in face detection and recognition, this paper proposed a face recognition algorithm which integrates the color detection algorithm and the two-value morphological processing algorithm.Firstly, the YCrCb model checking was applied to the captured image to reduce the impact of image brightness, and the HSV model checking was used to reduce the impact of light on the image;And then, the detected image was converted to binary image which was processed with binary morphology denoising method;Finally, the face recognition image was obtained.The experimental results show that the face detecting and recognition rates can reach 98.6%, 93.3%, and 90% respectively when the proposed algorithm was used to process complex, more complex and sophisticated face images.