Image Segmentation and Object Counting as an Instance of Travelling Salesman Problem

Abstract

Abstract The goal of this paper is to evaluate possible application of travelling salesman problem and its solving methods to image segmentation and object counting. This approach is inspired both by human skill of subitization and by ability of biological systems to solve optimization problems. Subitizing, or the ability to determine a number of objects without counting them, can be treated as a special case of image processing focused on extracting objects from background and enumerating them. This paper describes a method of generating linear image profile based on simple graph representation in which an optimal or suboptimal cycle is sought. Segmentation and counting can be then carried out by dividing the profile into coherent segments.