Computer-aided detection and identification of mine-like objects in infrared imagery using digital image processing

Abstract

Numerous image processing algorithms are available for detection and identification of buried landmine objects from infrared IR images. There is no universal solution available for this challenging worldwide problem. While developing an algorithm for a specific database, the external factors are not taken into consideration which implies that an algorithm which gives best results for a particular database may not be effective for another database acquired from a dissimilar scenario. Nevertheless, in order to increase the robustness of the detection system, a hybrid of intelligent algorithms and newly proposed methodologies are being used in many situations yielding better results. In that view, an effort is taken to identify the feasibility of some of the available image processing algorithms for their effective application in landmine detection using IR images. The results of outdoor measurements in two different soil types are presented and the implementations of these algorithms are qualitatively and quantitatively compared.