A study of shape characterization framework for planar objects from computer vision perspective

Abstract

Shape is perceived as a unique characteristic of an object which is intuitively understood newlineyet difficult to define formally. The human perception of shape seems to be an abstract newlineentity for which all kinds of mathematical frameworks fall short to find a proper crisp newlinedescription in representing the unique characteristic of a shape. Over the years, researchers newlinehave been trying to explore multifarious frameworks to characterize the shape of a planar newlineobject represented as digital bi-level image. In digital geometry, a shape of an object newlineis usually viewed as a set of connected dots located within a two-dimensional grid. newlineRegardless of how a shape is viewed, it has always been important visual information newlineattracting the attention of researchers over the decades for dealing with the problems newlineof computer vision. With the ever-growing demand for emulation of a human vision newlinesystem, research interest among today s scientists in the computer vision domain has been newlinegradually driven towards obtaining maturity in automated image analysis and recognition. newlineSince many aspects of images in our world are strongly determined by shape based newlinecharacteristics, the applications of shape characterization extend over almost every applied newlinescientific and technological area. In spite of its ubiquitous role in multiple applications, newlinenot many comprehensive shape characterization frameworks are found to be effective newlinefor offering a desirable solution to the computer vision problems. Existing studies do newlinenot fill in the vacuum in a comprehensive fashion particularly in developing effective newlineshape characterization frameworks for the kind of solutions required for today s computer newlinevision problems. Our effort in this doctoral proposal has been primarily conceived as newlinedocumenting a study of a few shape characterization frameworks and their abilities in newlineproviding solutions to various computer vision problems. newline