ABSTRACT

Landmine detection with passive infrared images can depend quite heavily on the environmental conditions, and there are cross over periods when the thermal contrast is negligible and the mines may be undetectable. Conventional antipersonnel mine detection has not evolved the perfect betterment in the methodological process. Here a generic framework is proposed using most adaptable methods and techniques. The IR captured image is being used for the detection and identification of buried targets by adopting high efficiency and better reliability image processing techniques for landmine detection. Results on diverse landmine data, collected using IR sensors show that the adopted method can identify meaningful and coherent clusters and that different expert algorithms can be identified for the different contexts. The initial experiments have also indicated that the need of preprocessing the images will highly increase the individual classifier performance. In future, the sensing technology can also be combined with processing power and wireless communication to make it profitable for various types of security threats.