In this work the authors present a novel image abstraction and stylization framework based on the analysis of natural scene in the multi-resolution Laguerre Gauss (LG) domain. The extraction of complex LG image sketches at different resolutions and the corresponding non linear adjustment ensure the preservation of the shapes and the deleting of the unimportant edge details. The reduction of the color range within the region interiors is obtained applying the image LG synthesis formula to the smoothed simplified image sketches. The proposed method is computationally easy to implement and provides the possibility to perform a fine tune regulation of the stylization parameters, customizing the NPR process. Subjective assessment of the proposed method were designed to compare it with two low level ”anchors”; the complete set of subjective tests are in progress; preliminary expert viewing test show some effectiveness of the proposed NPR technique.