基于高斯混合模型自适应肤色识别算法  

Adaptive algorithm for skin color recognition based on Gaussian mixture model

作　　者：任民宏[1] 鲁秋菊[1] 

机构地区：[1]陕西理工大学数学与计算机科学学院,陕西汉中723000

出　　处：《陕西理工学院学报（自然科学版）》2016年第6期53-56,85,共5页Journal of Shananxi University of Technology:Natural Science Edition

基　　金：陕西省教育厅专项科研计划项目(15JK1134)

摘　　要：肤色识别是色情图像识别、人脸检测等图像处理技术的关键步骤,肤色识别效果的好坏直接影响相关图像处理技术的效果。图像中存在干扰因素时基于单高斯模型的肤色识别效果不佳,为了解决这个问题,在YCbCr色度空间基础上,提出了基于高斯混合模型自适应肤色识别算法。实验表明该算法识别效率较高,能较好地识别存在干扰因素图像中的肤色,具有一定的实用性。Skin color recognition is the key step in the image processing technology such as pornographic image recognition, face detection, and so on. The effect of skin color recognition affects the effect of related image processing technology. In order to solve the poor effect of skin color recognition based on the single Gaussian model when the interference factors exist in the image, adaptive algorithm for skin color recognition based on Gaussian mixture model is presented on the basis of YCbCr color space in this paper. The experiment shows that recognition efficiency of the algorithm is higher, and skin color in image with the interference factors being better recognized with this algorithm and the algorithm has certain practicality.

关 键 词：高斯混合模型 自适应 肤色识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]