A simple and efficient color recovering system for content sharing website

Lossy image data compression is commonly used to reduce the original data size on traditional lossy compression fields, such as MPEG-x and H.26x including User-Created Content. This paper proposes a simple and efficient method for correcting online user-created contents¿ distortion in color. The proposed system consisted of two main phases: 1) in the training step, the working space was converted to CIE Lab color space. Next, the training images were classified to estimate the correction matrices. The correction matrices were calculated and optimized through a Newton iteration-based approach between the original and distorted images; and 2) in the correction step, the optimized correction matrices were applied to the distorted images for the color reproduction. In order to evaluate the performance of the proposed scheme, the quantitative measures were defined in terms of the peak signal-to-noise ratio (PSNR) and correction ratio. The experimental results showed that the proposed approach can reduce significantly the color distortion from a video sharing website. The proposed method can be a useful tool for correcting the distortion caused by image processing, such as lossy compression and color improvement in the consumer electronics-related field.