A fast algorithm for YCbCr to perception color model conversion based on fixed-point DSP

Abstract

Real-time video processing heavily relies on the color space conversion. Due to the real-time requirement, the traditional conversion methods often suffer from the moderate conversion speed, inaccuracy and low video quality. In order to meet the efficiency requirement of color space conversion in real-time video systems, we present a fast algorithm for color space YCbCr to perception color model conversion, which is based on the simplified shift and look-up table (SSLUT). The approach can be divided into two steps. At first, the simplified fixed-point shift method is used to convert YCbCr to RGB. Then look-up table assists with the YCbCr to perception color model conversion, such as HSV, HSI and HSL. To validate the proposed fast algorithm, the conversion speed and accuracy are compared with the traditional methods based on Code Composer Studio (CCS) test platform. Moreover, we make the video quality evaluation by using the peak signal-to-ratio (PSNR) and the structural similarity (SSIM). Experimental results illustrate the real-time, robustness and accuracy of the fast algorithm.