電子情報通信学会総合大会講演要旨
D-12-33
Pornographic Image Recognition using Eigenporn of HSV Skin Segmented Image
○I Gede Pasek Suta Wijaya(Mataram Univ.)・Keiichi Uchimura・Gou Koutaki(Kumamoto Univ.)
The research presents pornographic image recognition using eigenporn extracted by principle component analysis (PCA) of HSV based skin segmented image. In this case, HSV based skin segmented image is pornographic image which is filtered non-skin pixels using pixel-based classification on HSV color space before eigenporn extraction. The main aim of this research is to find a good solution for pornographic image recognition system, which give high accuracy and low false recognition rate. The experimental result shows that the proposed method provide by about 80% recognition accuracy less than 20% false recognition rate