Research of self-adaptive Chroma Space Model Skin-color Algorithm Based on Illumination and its application to erotic webpage filtering

Network erotic information can do very bad influences on people. Skin-color detection is the key technique to the Face Recognition and Erotic Image Filtering etc. Aiming at skin color Algorithm for now is not adequately considering the illumination information, the paper presents a new approach to detecting skin-color pixels in multicolor images based on the Chroma Space Model-Self-adaptive Chroma Space Model Skin-color Algorithm Based on Illumination. The paper colligates the distributed feature in YUV and YIQ color space, and differentiates skin color according to the I component of YIQ, angle of UV and illumination of images. The experimental results show obvious superiority in the yellow, black (Asian, African) and white skins (occidental). This new algorithm achieves 93.4795%, and error rate and detection time comparatively do not have much changes. The paper applies the algorithm to image filtering, and combines it with URL and keywords filtering to realize erotic webpage filtering. It overcomes the lags and limitations of the filtering software and realizes the possibility of supervising and analyzing webpage online with little affections to the net normal operation.