A Modeling of Shape Descripting Approaches using Local Patterns of High Gradient Points

—This paper presents a brief study of different shape analysis techniques and proposes a modeling of global shapedescriptors using high gradient points.The shape signature of these gradient points de ne the local and global shape details of the region. The basic idea behind this approach lies in the fact that the performance of the boundary descriptors is highly in uenced by the region segmentation and boundary extraction procedures which are required to be carried out prior to the shape extraction operation.Also,conventional global shape descriptors often take ahuge computational time,hence becoming unsuitable for real timeapplications.Hence use of high gradient points for de ning thelocal shape is introduced here.A comparative study with threedifferent gradient selection approaches and ve different kindsof shape signatures is carried out to identify the most ef cientshape descripting procedure. Experimental results are obtainedwith a set of hand sign images consisting of different kinds ofhand gestures.The dataset used for training and testing haveconsiderable variances in lighting,viewpoint and other factors sothat the potential of the feature extractor, when subjected to anykind of variations, can be judged.