مرکز منطقه ای اطلاع رساني علوم و فناوري - Human Face detection in color image By using different color space

چكيده عربي :

This paper presents a novel model to detect the human face in color image by using different color spaces (YCbCr and HSV). This work is divided into two independent stages. First stage extracts rules through using a number of images that contain skin regions and the second stage detects face regions. First stage includes using a number of color images. These images are converted into another color space (YCbCr and HSV). Then limiting only the skin regions in these images. After that compute the color values for each color space in these skin regions and draw a histogram for each color, through it, to get out the numbers that will help us to extract the rules which would be used to classify image pixels into skin or non-skin color. Second stage includes entering color image then convert it into two images each one represents different color space (YCbCr and HSV) , after that extracted rules in first stage are used to classify the pixels of each image into skin or non-skin color. Then delete the pixels that represent noise from each image by using the medium filter. After that these skin pixels combined to make meaningful regions (i.e. face, hand, etc.) in each image. Gathering the skin regions that are in the two images to get one image. Examining these regions throughout applying three tests (Compactness, Solidity and Orientation) on each region to decide whether it is a face or not. Finally the region face can be pointed out through drawing a box round it.