基于傅里叶变换的彩色滤波阵列插值新算法与实验验证

A novel algorithm of color filter array interpolation based on Fourier transform and its verification

在数码摄像系统中,CCD彩色滤波阵列(CFA,color filter array)被用于对颜色信息进行采样,然而每个像素只能对一种颜色进行采样,为了得到全彩色图像,必须对采样数据进行处理以估计其它颜色的像素值,这个过程通常利用插值来完成。CCD彩色滤波阵列插值算法是决定全彩色图像质量的最重要的因素之一,如果插值算法不够完善,图像的色彩就会出现失真。目前已有多种插值算法被提出,针对双线性插值图像在高频处失真严重及Gunturk自适应投影迭代法计算代价较高等问题,本文从傅里叶变换域分析彩色滤波阵列采样图像,根据采样图像在频率域上可以分为处于基带的亮度信息和高频段的色差信息的特点,提出一种色差调制插值算法。通过多组实验结果对比,本文插值算法在CPSNR(color peak-signal-to-noise ratio)值与ΔEs(S-CIELABdelta E)值以及还原图像的视觉效果上均要优于一般的插值算法。

In digital camera systems, the color filter array （CFA） is used to sample the colors. However, each pixel represents only one of the color bands. To produce a full-resolution color image,the recorded image must be processed to estimate the values of the pixels for all the other color bands. This restoration process is often implemented by interpolation. The interpolation algorithm for the charge-coupled device （CCD） color filter array is one of the most important factors for the quality of the restored full color images. If interpolation is not performed appropriately,images suffer from serious visible color artifacts. There have been many solutions to the problem, in order to avoid serious distortion in high frequency caused by bilinear interpolation and expensive computation cost caused by the interpolation using alternating projection, and the images sampled from the color filter array are analyzed in Fourier domain. The CFA images could be interpreted as a luminance component located in baseband and the color difference components modulated in high bands. According to this, a new interpolation algorithm based on color difference modulation is proposed in this paper. Compared with some other algorithms, the proposed method outperforms all them,both visually and in terms color peak-signal-to-noise ratio （CPSNR） and △Es.