| Summary: | 碩士 === 華梵大學 === 機電工程學系博碩專班 === 97 === The color of human’s skin is an important feature for human or face detection in a surveillance system. With the correct skin segmentation, the procedures of the face detection, hand gesture analysis, and objectionable image filtering are then reaching the higher accuracy. In normal light, the range of human’s skin color is usually lying within an interval. When an input image is too dark or insufficient exposure, the range of its skin color is outside the interval of human’s skin color and it will result in the error detections for the previous algorithms. By using the image enhancement to the image of too dark or with insufficient exposure, the range of its human’s skin color is converted to the detectable range of human’s skin color. Then, the correct regions of the skin color are found. Experiments are carried out for some samples with too dark or with insufficient exposure to demonstrate the computational advantage of the proposed method. Therefore, the proposed algorithm can be easy used in the intelligent image surveillance systems.
|
|---|
