A Research to the Autoregressive Model-Based Efficient Interpolation Algorithm

In order to improve the quality of interpolated images, this thesis proposes the concept of the autoregressive-based adaptive interpolation and the respective model. The basic idea is to construct the AR model based on the pixel structures of the image; to employ the idea of iteration, having got the initial interpolated values by applying the AR model, to consider the multiple constraints that the to-be-interpolated pixel is involved, respectively to construct multiple AR models, and to weightedly summarize their results, the process of iteration can be implemented many times, until the best result is achieved. Simulation results demonstrate that this interpolation algorithm can generate interpolated images with excellent subjective quality as well as high PSNRs.