| Summary: | 碩士 === 國立清華大學 === 資訊工程學系 === 91 === Due to the rapid growth of technologies for network and multimedia, digital input devices have become more and more popular, and the cost of the digital image is much lower now. Various applications which using facial features like personal identify and facial expression capture are coming to the mature period. However, face detection plays an important role in these issues. A good face detection algorithm could slash the computational time in verifying probable face candidates, and improve the comparing efficiency.
In this paper, we propose a face detection algorithm. In our method, after smoothed the picture to reduce thin color region effects, we can decide which regions are in the skin-tone scope by the color threshold, Morphology operations help us to exclude the parts which not fit in the right size and proportion. Then we divide the probable face candidate into three parts, and try to find the major objects in each divided area. Consider their color characteristic and position relations; we can get the correct face region.
|
|---|
