Naked image detection based on adaptive and extensible skin color model

摘要

The paper presents a new naked image detection algorithm. A learning-based chromatic distribution-matching scheme is proposed to determine the image's skin chroma distribution online such that it can tolerate the chromatic deviation coming from special lighting without increasing false alarm. The texture feature, namely coarseness, is used to acquire accurate skin segmentation. The low-level but reliable geometrical constraints and the mug shot exclusion procedure are employed to further examine the skin regions. Experimental results show our method can achieve satisfactory performance for detecting naked images under special lighting conditions.