Fast background removal of JPEG images based on HSV polygonal cuts for a foot scanner device

Abstract

Foot scanning devices aim to provide information on a patient’s foot and to help in diagnosing issues to be corrected with orthoses. The Galaxy foot scanner developed by the Aetrex company aims to provide a computer-aided framework to help physicians in their diagnoses. As numerous embedded devices, used for image processing and 3D-reconstruction, it includes cameras which provide JPEG pictures of the object to reconstruct. In this framework, an important step is the segmentation of the image, to isolate the object of interest, but the JPEG compression introduces artifacts which can lower the performance of any segmentation procedure. In this paper, we suggest a model which takes the artifacts stemming from the JPEG compression into account. The pixels are first sorted into layers of pixels with similar value V in the HSV color space, and the background is modeled by a polygon from an additional picture. Segmentation based on the knowledge of the background and the layer to be processed is then performed. Results obtained with the Galaxy foot scanner illustrate that this method provides good results for segmentation, while being sufficiently fast to be implemented for near real-time applications.