MSR (Multi-Scale Retinex) and SSR (Single-Scale Retinex) methods based on Land's Retinex model for lightness and color constancy of human vision have been applied to image processing for color correction. However, these methods are not always effective to correct simultaneously both color and contrast of negative color film taken by rear light, particularly if the RGB density distribution of film image is remarkably different each other. In this paper, we introduce a new algorithm to correct

more »

... or and tone at the same time which is modified the MSR model under the consideration of the weighting coefficient in MSR and relationship between original image and processed image by MSR. It is shown that the proposed method is significant to improve the color and tone by comparison with linear transformation , histogram equalization and MSR processing methods.