抄録
The paper presents the improvements in color appearance by Retinex vision model. The image is processed based on space-to-pixel relations not on the conventional pixel-to-pixel. Two different models were tested. One is center-surround and other is multi-resolution model. Both models worked to eliminate the spatial non-uniformities in the scene. Especially the multi-resolution Retinex resulted in the natural renditions in color and tone better than the center-surround Retinex.