An demosaicing algorithm based on edges

Abstract

Digital cameras have become the most important imaging devices in all kinds of Hand-held devices. In these devices, the quality of image reconstruction is an important indicator of the overall quality of the end-user experience image, and the image resolution is a mainly parameter of Digital cameras. Usually, only incomplete color information of the scene is measured at each pixel location by the CCD sensor, so that the Digital camera will have a low production cost. A true color image can be outputted by estimating all unknown color components, and we named the estimation process Demosaicing. A single chip image sensor is consisted of different color-pass filtering unit that is named CFA (Color Filter Array). Because there is only a kind of color sensor at the pixel position in CFA, the mosaic phenomenon of an entire image is serious. In order to remove the mosaic phenomenon, the unknown other two kinds of color information need to be reconstructed in each pixel. In the paper, our demosaicing algorithm is proposed to deliver high quality image based on CFA edge information. The algorithm is used to reconstruct a three-color image by analyzing local spatial gradient information of a CFA. The test results show that our algorithm is effectiveness for improving the phrase Peak Signal to Noise Ratio (PSNR) and the quality of image reconstruction.