摘要： 传统肤色检测方法无法同时保证较高的检测精度及较快的处理速度。为此，提出一种基于模糊认知图的图像压缩域肤色检测方法。在熵解码的离散余弦变换系数中，提取图像块的颜色特征和纹理特征，利用模糊认知图建立用于表征压缩域图像特征与肤色检测结果之间关系的肤色模型，采用该模型进行肤色检测实验。结果表明，与传统方法相比，该方法检测准确率更高，所需检测时间更短。

关键词: 肤色检测, 离散余弦变换, 特征提取, 压缩域, 模糊认知图, 分类器

Abstract: According to the traditional arithmetic in skin color detection which can not attain high precision detection and rapid rate transact. The paper proposes an arithmetic based on Fuzzy Cognitive Map(FCM) for detection skin color area on compressed domain. Color and texture features of the image blocks are extracted from the entropy decoded Discrete Cosine Transform(DCT) coefficients. The FCM is applied to set up the skin color color model to describe the relationship between the image block features and the skin color detection results, and the initial skin color image blocks are detected based on the model. Experimental results show that, compared with the traditional methods, it has higher accuracy and shorter detection time.

Key words: skin color detection, Discrete Cosine Transform(DCT), feature extraction, compressed domain, Fuzzy Cognitive Map(FCM), classifier

中图分类号: 

• TP391.41