| Summary: | 碩士 === 元智大學 === 資訊工程學系 === 96 === In this thesis, genetic algorithm with silhouette statistics is used to select significant features for face detection using silhouette statistics. The benefit of using silhouette statistics is simple and fast thus easy to meet real time requirement. As regards features, rectangle features are used in this study to discriminate the difference between face and non-face images.
More specifically, silhouette statistics are applied to training set, including face and non-face images, to evaluate the fitness value of feature set in the stage of feature selection using genetic algorithm. Then, the resulting optimal feature set is adopted for face detection using silhouetted statistics. Moreover, verification on face duplication and face skin has been involved to increase the accuracy of face detection. Experimental results prove the feasibility of the proposed face detection method.
|
|---|
