Color Image Pre-processing Algorithm Research

Chip development ordinarily requires a long development cycle. Taking into account the cost and life cycle constraints, the effectiveness of chips to develop should be fully verified a prior. The research of video image processing algorithm is aimed at this feature of chips development. The software is used to simulate the video image collecting and processing. At first, the video is disassembled into frames, then each frame is processed in sequence, and finally the frames are assembled as output video. The frame processing includes several typical algorithms, which contain three major categories. They are denoising, color interpolation and white balance respectively. This paper mainly discusses color interpolation and denoising.Denoising is the first and also an important step in the image processing. If the noise can not be removed, the subsequent interpolation will be affected seriously. Denoising algorithm in this paper is used to remove video image noise, taking into account not only the spatiality of the image frame but also its temporality. The influence of the former frame to the current is considered. The method removes the noises well and guarantees the accuracy of the interpolation.In order to obtain full-color images, some algorithms are used to retrieve the information of the two missing colors at each pixel. This process is called color interpolation. In the existing literature, a large number of interpolation algorithms have been proposed, ranging from simple linear interpolation algorithm to mature adaptive interpolation. The paper improves the dual-related interpolation algorithm. The improved algorithm has good results, but because of its complexity, it is somehow difficult to implement with hardware. This paper also proposes an interpolation algorithm based on the Taylor formula and an improved CFA post-processing algorithm. Both have produced good results in the tests. Last discussed are the evaluation measures of current typical interpolation algorithms, which are used to comprehensively evaluate and analyze the performances of interpolation algorithms introduced in this paper.