Comparison of an Iterative Deconvolution and Wiener Filtering for Image Restoration

Linear deconvolution techniques usually suffer from a number of short comings such as generation of negative pixel values and artefacts which lead to poor quality restored images. This paper discusses the application of a nonlinear iterative deconvolution algorithm to both synthetic and real images, and the results are compared with the Wiener inverse filter. Experimental res11lts are presented which show that the nonlinear algorithm always produces restored images with much better quality than the Wiener filter. In particular the Wiener filter shows poor performance when applied to images with a low signal-to-noise ratio.