An Effective Modified Retinex Algorithm for Adaptive Contrast Enhancement of Image
Pei Li · Ceshi jishu xuebao · 2009
A modified Retinex algorithm for contrast enhancement of image is proposed.And nonlinear transformation functions are introduced to modify illumination and reflection components of the color image.The global contrast enhancement function can stretch illumination component to improve whole visual appearance.The nonlinear sigmoid function changes little for higher or lower reflection component value,and more for median of reflection.Finally local contrast of image is enhanced.Comparisons between the proposed method and several image enhancement algorithms show that the image processed by our algorithm has better appearance,and the visible detail in bright and dark region is improved remarkably.Classical Retinex algorithms are all influenced by multi scale convolution,so operation complexity is high.Our method overcomes limitation of classical algorithms.The method can not cause visible color distortion for RGB and some chroma-luminant color spaces.And operation cost is low and parameter adaptivity is well.