Research On Color Cast Detection And White Balance Method Based On Green Channel

With the rapid development of technology and widespread use of image acquisition equipment, there are more and more ways to acquire images. As the most direct way for us to obtain external information, image records a lot of information, such as lightness, contrast and color etc. When observing images, we will be attracted by color first. That is to say, human eyes receive color information first, so color information has great influence on image vision.Color is a kind of unstable characteristic in image. When illumination is changed, the color of object reflecting also changes. It brings great challenges to image application which relies on image color. Human eyes can see a real scene by avoiding the influence of illumination, but image acquisition equipment can't. For example, the CCD of digital camera is affected easily by the light source, so the image took by the camera always has color deviation. Therefore, color cast needs to be removed, and this process is called white balance. This paper analyzes some color cast detection algorithms and white balance algorithms, and then a new color detection algorithm and a new white balance algorithm are proposed.First, this paper embarks the colour theory, introducing some knowledge which is necessary to detect and removal cast, such as histogram, color space, edge detection operator, the relationship of color temperature and illumination, white point detection, white balance process, Lambert reflection model and Von Kries model being used to remove color cast. These knowledge is the basis of image processing, especially widely used in the field of color constancy.Then, some algorithms for color cast detection are studied in detail, then a color cast detection method based on chromaticity histograms is proposed. The methods of detecting color cast analyzed here are: Gray-world algorithm, White-patch algorithm and color cast detection method based on Lab color space. They all have their own advantages and disadvantages. Some methods are low efficiency, and some are low accuracy, so a color cast detection method based on chromaticity histograms is put forward. The detection result and efficiency of this algorithm are better, and prepare for the next section.Next, some classic color cast removal algorithms are analyzed and studied in this paper, such as Gray-world algorithm, perfect reflection algorithm and an auto white balance method based on edge detection. Then, experiments of all algorithms are listed and respective advantages and disadvantages are analyzed, preparing for the auto white balance using green channel.Finally, an effective auto white balance method based on green channel is proposed. First, the features of green channel, such as high sampling rate, high photosensitivity and sensitivity to human eye are analyzed. Moreover, green channel is more stable than other channels when color temperature changes. Therefore, white reference area which is extracted from green channel is more reasonable. For each image, this method estimates the scene illuminant according to the white reference area, and then gain coefficients are obtained. Next color cast can be removed. The experimental results show that the proposed method can achieve better results in both objective and subjective evaluative terms. The complexity of the proposed algorithm is acceptable, so color cast can be fast remove. The proposed method can be easily applied in digital cameras to obtain good results. Therefore, it can be widely used in digital camera and mobile devices.