Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 98 === With the high-speed development of technique, the camera imaging system has
been widely used. People could record their good experience or beautiful memories by
those devices easily; however, the displaying image appears to have an unequal brightness
distribution that may caused by unfamiliar with those devices or by the bad weather. It will
be hard for people to decide to drop it or not.
We can regulate those unperfect images by image processing softwares such as
PhotoShop, PhotoImpact and so on, but it will take a lot of time to regulate them manually.
This thesis presents the adaptive backlight-balancing processing algorithm to compensate
unequal brightness distribution based on face detection. We find out all the face locations
and sizes by Haar Object Detection algorithm which was published by Viola and others.
We can compensate unequal brightness distribution by the detecting results. And we
employ a brightness reducing slope to reduce the brightness nearby the face locations. By
this way, we do not need to process the whole image to regulate the unequal brightness. So
that it makes the processing efficiently and automatically. Simulations demonstrate our
algorithm can improve the unbalance brightness from still camera, specially for human
head.
|