| Summary: | Images contain three primary colors at each pixel, but single sensor digital cameras capture only one of the primary channels. Process of color image reconstruction by finding the missing color component is called color image demosaicing. Various approaches have been proposed in this field of image demosaicing such as interpolation based and frequency based approaches due to sharp image edge and higher color saturation, and these techniques fail to reconstruct image efficiently. To overcome this, in this work we propose a new approach, sparse based RBF network for color image demosaicing. According to this approach a sparse model is constructed first and based on that weights are computed which are used to minimize the reconstruction error. To improve this we use optimal weight computation and RBF training for missing color component value prediction. Proposed method is implemented using MATLAB tool and experimental results show the efficiency of the proposed work in terms of color peak signal to noise ratio (CPSNR). Simulation results show 16.20% improvement in the performance in terms of CPSNR. Keywords: Demosaicing, Bayer pattern, CPSNR, RBF network
|
|---|
