An Intelligent Face Detection Technique by Using Facial Geometric Features

• Main Authors: Chi-wei Chen, 陳麒為
• Other Authors: Wen- Chen Huang
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/46997884574807548643

碩士 === 國立高雄第一科技大學 === 資訊管理所 === 95 === Because the handy equipments such as digital cameras and digital camcorder with camera function provide a convenient way to capture images. It becomes a major topic for research or application to extract content from these images. We propose an automatic face detection method which can be used to extract facial features from a color image with complex background in a still image. First, lighting compensation is the preprocessing in our method to ensure that the input image keeps original color. Next we use NRGB and YCbCr color space transformation to detect the location of skin color and employ dilation and erosion operation of morphology to delete noise of image. Then, we utilize Fuzzy logic method to analyse facial ratio. If it doesn’t match ratio, we use Integral Projection Method to segment connected area. In addition, we detect eyes, mouth and facial outline. Finally, we compare our method and CMU system. Our detection rate is higher than CMU system. CMU system is supervised system, but our method is non-supervised system. We don’t need to train image beforehand and detect objective image. The results show that the proposed algorithm certainly works well to validate multiple face regions under complex background.