Mine Detection Techniques Using Multiple Sensors

This work has four main objects, to survey previous research about mine detection, to collect signature data for future research, to experiment with the surveyed methods and collected data, and to propose an idea for an advanced method. Mainly this report is about the surveyed image processing methods and their implementation. Also, a method to find mine targets from a set of candidates is proposed. Sensor technology is introduced prior to the image processing sections to help understand the character of the data signal. The intention of the solution is to analyze a set of infrared data sequences, which is called dynamic thermography. The necessary image processing methods are introduced as four topics, filtering, feature extraction, gray-scale morphology, and segmentation. In the detailed description, the Karhunen-Loeve transformation is introduced as a feature extraction topic. Several operators of gray-scale morphology are introduced for the purpose of filtering, gradient, and segmentation. The alternating sequential filter is especially used for the filtering purpose. Finally, the markerbased watershed algorithm is introduced for the segmentation method. These methods are tested with actual mine data. Table of Content Mine Detection Techniques Using Multiple Sensors.............................................................................................. 1 Chapter