基于残差和高频替换的SONY-RGBW 阵列 彩色重构方法

RGBW filter arrays are often used to improve the imaging quality of detectors at low illuminance, but the corresponding color reconstruction methods are still designed according to the characteristics of Bayer array, and do not make full use of the advantages of luminance information, resulting in poor reconstruction results. To overcome this defect, guided filtering is applied to delve the correlation between luminance information and color information first. Then, multi-step residual-based interpolation algorithm with increasing sampling rate is designed, based on the smoothness of residual, which is applied to the spatial characteristics of the SONY-RGBW color filter array. Moreover, combined with the interpolation results in the orthogonal directions, the iterative process is introduced and the pixel-wise evaluation factor is improved to improve the edge adaptability of the algorithm. Finally, the high-frequency component of the luminance image is used to enhance the color image. Experimental results on the Kodak data set and images in real scenes show that the proposed method can reduce color aliasing, reconstruct clear image detail, and the objective criteria with or without reference images are better than the existing color reconstruction methods for RGBW color filter array.