A breast detecting algorithm for adult image recognition

Information spreading on the Internet is increasing explosively. Juveniles should be protected from the harmful information. Adult image recognition is one of the important works. In order to avoid the low recognizing accuracy of present algorithms, an adult image recognizing algorithm taking model's breast as recognizing object is proposed and discussed in this paper. The algorithm uses fuzzy cluster method to segment candidate region according to common colors of breast feature. After getting the preliminary candidate region, the algorithm applies "what" and "where" way for information processing of visual system to compose integrated breast feature vector. The algorithm uses Forward Propagation neural network to judge whether the candidate region is a breast feature or not. Then the result of image recognition is gotten through the existence of breast. Experiments show that the developed recognizing algorithm can accurately confirm whether there are breast parts in images or not. Therefore, it can be applied in practical adult image recognizing system effectively.