It is known that cancelling the noise without blurring the image details is a very difficult task for any image denoising technique. The availability of metrics for accurate evaluation of filtering distortion is thus of paramount importance for the development of new filters. Peak signal-to-blur ratio PSBR is a recently introduced measure of detail preservation that overcomes the limitations of the sole peak signal-to-noise ratio (PSNR) and other metrics in evaluating the performance of image denoising filters. Formally, the PSBR is the PSNR component that deals with the detail blur, so the method that is adopted for blur estimation plays a key role. This paper presents a novel algorithm for PSBR computation that offers significant advantages over the first method: it is simpler, more robust and much more accurate. Furthermore, this paper presents new validation tools for evaluating the accuracy of this kind of metrics when some well known classes of linear
and nonlinear filters are considered. Results of many computer simulations dealing with images corrupted by different combinations of Gaussian and impulse noise show that the proposed PSBR algorithm outperforms the most effective metrics in the field.
