Multi-dimensional Multi-resolution Fourier Analysis Method in Image Processing

Signals obtained by equal space interval sampling technique actually imply unobservable information through traditional analysis methods. Compared with state of the art methods to accomplish multi-resolution analysis, the Multi-dimensional Multi-resolution Fourier Analysis (MMFA) approach proposed in this paper is capable of recovering hidden frequency information in multi-dimensional signals. When applied our method in degraded image restoration processing, it can effectively decrease the parameter estimation error so that it can consequently increase the quality of restored image. Experiments show that the new method, compared with direct measuring, can yield a more accurate result both in visual quality and calculated PSNR (Peak Signal to Noise Ratio).