Artificial intelligence search application to isothermal contour detection in a thermal image

Abstract

Summary form only given. The problem of isothermal contour detection has been reduced to the problem of finding an optimal path in a weighted tree. The properties of the contour are embedded in the structure of the tree. Graph searching techniques are then used to find the optimal solution. To search the optimal isothermal contours, a heuristic search algorithm has been adopted. The algorithm has been implemented in C on the Micro-Vax II workstation. The information about the problem is given by specifying the start node, the characteristics of the goal node, and the rules for expanding a node and for computing its cost. The program has been tested by adding different amounts of Gaussian noise to a picture. It is able to draw acceptable isothermal contours even in noise-added thermal images, but the computation time for detecting an isothermal contour depends on the noise.<>