利用信号相关对BAYER格式图像色彩复原

Color Reversion of BAYER Pattern Image Using Signal Correlation

目前将单CCD采集的BAYER CFA格式图像进行色彩复原有两种通用方法:双线性插值和边缘检测.为了降低这两种方法在图像边缘引进错误颜色的数量,根据RGB彩色图像中R、G、B三个色彩通道的高度相关的特性:即在5×5像素的小区域内认为绿色减红色以及绿色减兰色为常数,提出了利用信号相关对BAYER格式图像彩色复原的方法,并通过SNR(信噪比)和NCD(归一化彩色差异规范)两种评估方法进行检测.从实验结果可以看出文中提出的方法相对于通用方法提高了图像的信噪比,锐化了图像的边缘,减少了图像高频信息的错误像素的数量,提高视觉质量.

At present, there are two conventional methods of color reversion of BAYER pattern image from signal CCD. One is bilinear interpolation method, the other is edge-sensing method. According as correlates of the R, G, B channels in color image, we propose a new interpolation method to improve the image quality and SNR , sharpen edge of the image, reduce the number of error pixels. At same time, SNR and NCD methods are used to evaluate signal correlation method and two conventional methods.