Object-based human face detection

We propose an object-based human face detection method. The proposed method consists of two steps: region segmentation and facial region detection. In region segmentation, the input image is segmented by a genetic algorithm into several initial regions. Then, to merge the regions forming an object, we use the spatial similarity between two regions because the regions forming an object share some common spatial characteristics. In facial region detection, the facial regions are identified from the results of the region segmentation step using a skin-color model. This approach can detect multiple faces of multi-views and small size in a complex background environment.