基于标准差加权的色偏图像校正方法

针对色偏图像校正问题,提出了1种标准差加权的色偏图像校正方法。首先,利用灰度世界算法计算色偏图像红、绿、蓝各通道的校正系数；其次,对色偏图像进行超像素分割,利用像素平均值和方差计算每1个超像素的颜色一致性,从而计算各通道的校正系数权重；最后,对校正系数进行加权处理,并进行色偏图像校正。实验表明,算法能够对色偏图像进行有效校正,并克服单一色块和校正后图像变亮问题。相比于灰度世界、完美反射和动态阈值3种算法,该算法的峰值信噪比平均提升36.46%,色差均值平均降低3.12%。

Color Cast Correction Method Based on Weighted Standard Deviation

To solve the color cast problem, this paper proposes a color cast correction method based on weight.ed standard deviation. In this method, correction coefficient of R, G, B channel is firstly obtained by grayworld algorithm. Then, the superpixel is acquired with superpixel segmentation. For each superpixel, the colorconsistency is further calculated through its pixel average value and standard deviation. The weights are thencalculated based on the color consistency. Finally, the weighted correction coefficient is get and applied to cor.rect the image. The experimental results show that the proposed method is effective. And it can not only over.come the influence of single color blocks, but also can solve the problem of brightness increase after correc.tion. In comparison to gray world, perfect reflector, dynamic threshold, the presented method can raise PeakSignal to Noise Ratio by 36.46%, and reduce color difference average value by 3.12%.