Implementing real-time RCF-Retinex image enhancement method using CUDA

Xiaomin Yang, Lihua Jian, Wei Wu, Kai Liu, Binyu Yan, Zhili Zhou, Jian Peng

Abstract: RCF-Retinex is a novel Retinex-based image enhancement method which can improve contrast, eliminate noise, and enhance details simultaneously. It utilizes region covariance filter (RCF) to estimate the illumination. However, RCF-Retinex encounters time-consuming problem, since the region covariance filter is computationally intensive, which restricts the practical application in real-time systems. Therefore, it is necessary to decrease the computational complexity by parallelization. This paper proposes a GPU-based RCF-Retinex, which can accelerate region covariance filter using CUDA. It is feasible to use CUDA to parallel the region covariance filter due to its consecutive convolution operations, thus we can obtain the illumination image fast. Experiments have proved the improvement of running time and the enhancement results are similar with those using the unaccelerated RCF-Retinex method.