GradientShop: A gradient-domain optimization framework for image and video filtering
In this paper a simple but effective operator for the reduction of blocking artifacts is presented. The method is based on the Rational Filter approach: the operator is expressed as a ratio between a linear and a polynomial function of the input data. Such filters proved to outperform other conventional methods in other applications, such as noise smoothing , thanks to their capability of adapting gradually to the local image characteristics. The filter is capable of biasing its behaviour in order to achieve good performance both in uniform areas, where linear smoothing is needed, and in textured zones, where nonlinear and directional filtering is required. A detector of activity is embedded in the expression of the operator itself so that the biasing of the behaviour of the filter is smooth and not based on fixed thresholds. The proposed method has been originally designed as a post-processing tool for frames of sequences coded at medium-low bitrate with the dynamic coding scheme proposed in , but gave good results also when applied to JPEG coded images.