Blur identification in image processing

The aim of this study is to achieve a blur identification task in still images. In fact, in photographic camera, the optical lenses may be set in a way to clearly distinct two areas in the image: the blurry one and the non blurry one. An automatic segmentation coupled to specific descriptors allow first to describe any region of the image. Then, a supervised learning processes permits to build a classifier able to decide for each unknown region the label "blurry" or "sharp". We discuss here precisely the overall process, from the objective choice of the segmentation algorithm to the presentation of the different introduced descriptors. Finally, some results are presented validating such an approach.