Bayer Pattern Demosaicking Using Local-Correlation Approach

A new Bayer pattern demosaicking scheme for single-sensor digital cameras is introduced. The raw output from a sensor, mostly a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) sensor, with a Bayer filter represents a mosaic of red, green and blue pixels of different intensity. To interpolate the two missing color components in each spatial location and constitute the full color, camera output, the proposed method utilizes edge-sensing interpolation and correction steps. Since the correction step is suitable only for the image regions with high spectral correlation, otherwise is counter productive, the scheme is adaptively controlled through the comparisons between the correlation coefficient and the pre-determined parameter. The proposed method yields excellent performance, in terms of subjective and objective image quality measures, and outperforms previously developed CFA interpolation solutions.