Fast rotation invariant face detection in color image using multi-classifier combination method

A robust and real-time facial detector has been designed using multi-classifier combination method. The new detector composes of three classifiers: Skin color detector, AdaBoost detector based on haar-like features, and eye-mouth detector, a semi-serial architecture is designed to combine the three detectors, which set up the division and cooperation system and draw on each other´s merits to implement the quick and efficient facial detection. An illumination independent skin color detector has been designed. Retinex algorithm based on the advanced adaptive smoothing filter is exploited to remove illuminant in color images and then skin color pixels are classified through there color information. This method increase efficiency and ability of skin color detection in images contaminated by illuminant. Besides this, the advanced adaptive smoothing algorithm adopts adaptive windows to replace the iteration, which guarantees real-time performance of skin color detector. Based on the acquirement and conditions of the new facial detector, a simple efficient eye-mouth classifier has been designed to reduce miss detection caused by rotary faces. The detector chooses color feature of eyes and mouth and geometric feature between eyes and mouth. Experimental result proves that the new face detector implements real-time face detection and improves detection robustness.