| Summary: | 碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 93 === Face is only a part of image for most of face recognition. We have to find the face location from the image, then choose the location and do face recognition. The process for finding face location called face detection. The previous job for face detection is face recognition. If the percentage of face detection is high, not only can make face recognition easier, but also raise the correct percentage.
Recently, Support Vector Machines has become a remarkable machine learning way. The recognition speed, percentage make people feel satisfy. This article uses Support Vector Machines to build a face detection system. It can detect face location from input color image, and shows people’s face, and eyes. There are four steps for face detection. Previous work, faces located, feature select, and face test.
The test result shows the correct percentage is 93.8% in complicate environment, and more faces image. Average each page takes 1.15 second. The above proves the way in this article works. The detect ratio, and speed are pretty good, and better than other face detect way.
|
|---|
