Department of Statistics, The Ohio State University
Statistics and Biostatistics Colloquium Series

An image can be regarded as a surface of the image intensity function which may have jumps at the outlines of objects. So edge-preserving image restoration in image processing is essentially the same problem as jump-preserving surface estimation in statistical regression analysis. In this talk, a new methodology of jump-preserving surface estimation is discussed, which is based on local smoothing techniques. Its surface estimator has an explicit mathematical formula like most conventional surface estimators in the literature. Therefore it is easy to use, simple to compute and has profound statistical theory. This method can be applied directly to other application problems including mine surface estimation in geology and equi-temperature surface estimation in meteorology. Numerical examples show that it works well compared to some existing procedures for the same purpose.