A New Spatial Approach to Image Restoration

Study in restoring images from their degraded states has been an important research topic in image processing and has potential applications in complex pattern recognition. We propose in this paper a new adaptive image restoration method using the concept of random-function realizations in geostatistics. This conceptual framework allows us to derive the model means and variances in the context of spatial statistics. Experimental results demonstrate the superior performance of the proposed approach to other image restoration algorithms.