Detection of Region-of-Interest Based on Noise Filtering in High-Speed Images

Relatively inexpensive cameras have become common that can film even minute movements of moving objects at high speeds. This paper introduces a method of removing noise pixels abnormally generated from the ultra-high-speed images entered, and then detecting the skin color area of a person, the area of interest that represents personal information from the images of noise removed. The proposed algorithm eliminates noises by applying bilateral filters from the high-speed images that are first accepted. Then, using color models created through learning, the human skin area, which is an area of interest, is accurately extracted from the image of noise removed. The experimental results show that the proposed algorithm effectively removes noises from the high-speed images entered and then robustly detects the area of interest, the skin color area. The method introduced in this study is expected to be useful in many applications related to pattern recognition, such as image analysis and understanding, video pre-processing, noise removal, etc.