GPU Accelerated Computing for Human Skin Colour Detection Using YCbCr Colour Model
Ganesh B. Solanke, Sonal A. Gore · 2017
Human skin detection is a significant initial stage to improve the performance of other areas of object detection or recognition such as skin segmentation, human face detection, human-computer interface, video surveillance and recognition of hand gesture. Fast recognition of skin colour in images is a significant task for image processing. In this work, a new GPU accelerated computing for detecting skin colour has been introduced. The proposed approach uses YCbCr colour space, which is not popular in image processing, and skin detecting algorithm, but resulted as a good choice. YCbCr properties make YCbCr colour space the most suitable for skin colour detection. An image is made of millions of pixel and every pixel information is independent of its neighboring pixel. Hence this work focuses on the capability of Graphics Processing Unit (GPU) to compute in parallel against the millions of pixel calculations involved in image processing. Every pixel processing is independent from other; therefore GPU can be effectively used by using high programming interfaces. In this paper we have used CUDA as parallel programming platform. It helps to speed up the human skin detection process. The proposed YCbCr skin colour model for skin colour detection was evaluated on a SFA database. The results analysis of this work proves achievement of significant speed up in well-defined test environment. The result showed that parallel computing technology can improve the speed of processing greatly and the performance of human skin detection using YCbCr colour space.