文摘
High-spatial resolution and wide field of view (FOV) can be satisfied simultaneously with a dual-sensor camera. A special kind of dual-sensor camera named dual-resolution camera has been designed and manufactured; therefore, a high-resolution image with narrow FOV and another low-resolution image with wide FOV are captured by one shot. To generate a high-resolution image with wide FOV, a fast super-resolution reconstruction is proposed, which is composed of wavelet-based super-resolution and back projection. During wavelet-based super-solution, the high-resolution image captured is used to learn the co-occurrence prior by a linear regression function. At last, low-resolution image is reconstructed based on the learnt co-occurrence prior. Simulation and real experiments are carried out, and three other common super-resolution algorithms are compared. The experimental results show that the proposed method reduces time cost significantly, and achieves excellent performance with high PSNR and SSIM.