Color Casts Detection and Adjustment

This paper presents a n ew method for detection and adjustment color cast. Using the neural network to detect color cast and classify images into three subsets: no cast, real cast, and intrinsic cast (image presenting a cast due to a predominant color that must be preserved). We have a database of 700 images which are downloaded from internet or acquired using various digital still cameras. We randomly select 350 images from the database for the neural network learning, and the others are for testing. From each training image, we can calculate 13 statistical parameters as input to the neural network. The second part is the white balance algorithm which is applied to the image while a real cast is found by the color cast detector. The test image is divided into m blocks. For each block, the output weighting can be obtained by a fuzzy system and the luminance weighted value is also calculated. Finally, we can obtain the new amplifier gains of the R, G, and B channel to adjust the color cast. If the input image be classified as no cast or intrinsic cast, white balance algorithm is not applied.