Performance evaluation measures for adaboost algorithms in face detection

Abstract

The human face has unique ability to recognize all thousand of face by human itself and they will learn to remember the face based on eye, hair, nose, mouth, cheek and also physical faces. From these characteristic, human face represent the identity and emotional. The Adaboost algorithm will be use in this study. These algorithm good in performance and minimize the error to located the face in the image. In face detection process, the noise came from the image data will affect the performance. The preprocessing process in raw image involving the image enhancement, filtering and determination of textual feature will be necessary use for successfully implementation of aim in this research. Firstly, the raw image has been applied histogram equalization method in order to enhancement the image intensity. The segmentation process will also applied to partitioning an image into components using threshold concept and the noise from the image will be eliminated by using filtering. The feature extraction is important to extract the face region such as eye, mouth and skin. The output from feature extraction will be used for classification. The classification process will be determining the characteristics between face and non-face. Then, the face will be calculated in the image. The proposed technique will be study the performance in face detection.