Markov random field processing for color demosaicing
Pattern Recognition Letters
A generic variational approach for demosaicking from an arbitrary color filter array
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Nonlocal-means image denoising technique using robust M-estimator
Journal of Computer Science and Technology
Effective demosaicking algorithm based on edge property for color filter arrays
Digital Signal Processing
Linear demosaicing inspired by the human visual system
IEEE Transactions on Image Processing
Color demosaicking via directional linear minimum mean square-error estimation
IEEE Transactions on Image Processing
Practical Poissonian-Gaussian Noise Modeling and Fitting for Single-Image Raw-Data
IEEE Transactions on Image Processing
Hi-index | 0.00 |
In this paper, a refinement scheme considering channel correlation is presented for a color filter array (CFA) sensor with the white (W) channel. Differently from the Bayer CFA, which has the red, green, and blue (R, G, and B) channels only, the R, G, B, and W channels can be alternately assigned to each grid to form new CFA patterns. However, the resolution degradation of the interpolated results in the pattern with the R, G, B, and W channels is more prominent than that in the Bayer pattern, because channel correlated errors in the high frequency, which originate from errors during the color interpolation process, such as false color and aliasing artifacts along the edges and in the details, are magnified. The proposed refinement scheme is applied to the three CFA patterns, which contain the W channel of the quincuncial structure. The interpolated W channel has more high frequency than the other interpolated R, G, and B channels, because the W channel occupies the largest pixel samples in the patterns. Thus, the W channel is eventually utilized to improve the R, G, and B channels by applying the edge adaptive refinement considering channel correlation based on high frequency reconstruction. First, the CFA patterns are partially interpolated to generate quincuncial patterns with the same structures of the Bayer pattern. Then, the edge adaptive color interpolation is applied to each pattern. Finally, the smoothing filter based on robust estimator is adopted for the refinement of the color difference image between the degraded color channel and the high resolution W channel. Experimental results of the proposed scheme are shown in comparison with the results of the conventional methods.