An Adoptive approach in Demosaicing True Color from Digital Camera having Bayer Color Filter

Single sensor with Bayer pattern color filters are used to reduce size and cost of consumer digital cameras. These cameras use various demosaicing algorithms to construct the entire image with true colors from the acquired insufficient RGB (Red,Green and Blue) mosaics. In order to estimate original true colors, iterative demosaicing algorithms are being employed. These algorithms incur high cost in computing and process all the pixels irrespective of the content of the image. The computation expense of algorithm and the quality of image are directly proportional. However, it has been observed that for homogeneous region, the simple (non-iterative) and complicated (iterative) demosaicing algorithms yield result in same image quality. In this paper, homogeneous and heterogeneous regions of images are identified using neighboring color dependency matrices namely NRDM (Neighboring Red Dependency Matrix), NBDM (Neighboring Blue Dependency Matrix) and NGrDM (Neighboring Green Dependency Matrix). The color distribution of homogeneous, heterogeneous regions is evaluated from these dependency matrices and used to identify the type of selected region. The identified region type is considered for determining a demosaicing algorithm having less computation provides the same image quality. The simple, non-iterative demosaicing algorithms are employed in homogeneous regions and iterative algorithms are used in heterogeneous regions. The image quality metrics Mean Square Error (MSE) is used to evaluate the proposed adaptive method with the existing iterative algorithms. The reduction in MSE values shows the effectiveness of the proposed adaptive approach. KeywordsDemosaicing, Color Dependency Matrix, Homogeneous & Heterogeneous regions