In this research, hyperspectral and terahertz (THz) imaging were used for segmentation and detection of desired objects. In the hyperspectral part of the research, the performance of hyperspectral image classifying algorithms were tested and enhanced by parameter modification and postprocessing for detection of prescribed objects in hyperspectral datasets. Both supervised and unsupervised classifiers were used for enhancing the accuracy of detection of features such as roads, buildings, trees and land in hyperspectral imagery. Postprocessing of classifier output was used to further enhance the accuracy of detection and segmentation of desired objects. In the THz part, concealed objects camouflaged under cloth/paper/leather/plastic were detected using the joint transform correlation algorithm. With THz imaging technology, it is possible to spot concealed objects which cannot otherwise be revealed through traditional imaging techniques. Here, a novel technique for the detection of concealed objects in noisy low resolution THz imagery is proposed. The detrimental effects of other artifacts are reduced by utilizing the nonzero-order fringe adjusted joint transform correlation (NFJTC). Test results using real-life THz imagery confirm the effectiveness of the proposed technique.
展开▼
