Face detection in color images

摘要: Human face detection plays an important role in applications such as video surveillance, human computer interface, recognition, and image database management. We propose a algorithm for color images the presence of varying lighting conditions well complex backgrounds. Based on novel compensation technique nonlinear transformation, our method detects skin regions over entire then generates candidates based spatial arrangement these patches. The constructs eye, mouth, boundary maps verifying each candidate. Experimental results demonstrate successful wide range facial variations color, position, scale, orientation, 3D pose, expression from several photo collections (both indoors outdoors).