Hide/Show Apps

Through the wall target detection using textural features

Gedik, Emre
Through the wall imaging(TWI) is a important technology for detection and identification of human or other objects behind the wall especially for military and civilian applications. However, the clutter, multipath ghost, and wall distortions decrease the accuracy of current TWI-based human detection and identification methods. Various algorithms use various techniques such as shape-based, image-based, and physicsbased techniques. In this thesis, we summarize all the techniques used for TWI and analyze the GPRbased detection techniques and evaluate their performances with simulated data. Images are obtained by simulation using gprMax software. After, the flash effect removal, image segmentation and feature extraction methods are applied. Finally, obtained features are classified with SVM as targets or other objects. 108 different scenarios are generated and used in the evaluation of the algorithms.