Fast face detection method adapted for mobile platforms

For the original AdaBoost algorithm´s long searching time, this paper improves the original face detection method which is based on AdaBoost face detection method using a simple rectangular features. The paper also proposes a method for rapid detection of human faces. First of all, we do HSV space transformation to the pretreatment of image, then add skin color segmentation to the image, and finally do its morphological operations and contour extraction. This method effectively reduces the time require for the face detection process, and gets higher speed and accuracy than previous algorithms. The method based on Opencv comes true under the open source vision library. In order to overcome the limitations which can only be achieved the method in the C/C++ environment, we use the native method interface provided by SUN Company and transplant it to the Java platform. We can apply this method to the Android etc. or the embedded operating system using Java as the developing language after improved. The proposed method improves the speed and maintains a higher availability at the same time. We conduct experiments on the PC, the results of the detection rate reaching to 94.28%.