An efficient human body contour extraction method for mobile apps.

Abstract

With the increasing prevalence of powerful mobile technology, many applications involve human body measurements, such as online cloth shopping. Aiming at the application of non contact human body measurement and modeling system, this paper presents a new method for extracting human contours in complex background environment. Since the H component on the HSV color space is independent of the light changes, the hair and the lower garment can be divided. Therefore, a method of using the elliptical skin model on YCbCr color space is proposed to detect the skin color, then, automatically extract the skin samples to determine the threshold segmentation range. The combination of the two improves the skin color point recognition rate, gradually separating the body skin, clot and the hair by using these binary values of the linear fusion operation to get the final body contours. Our experiments demonstrates that this algorithm effectively reduces constraints on background environment when extracting contours.