A Novel Approach to EO, IR, SAR, and Hyperspectral Sensor Fusion

Target detection and classification is challenging because target detection and classification cannot be effectively carried out by pure visual inspection and must rely on some automatic algorithms. Intelligent Automation, Inc. (IAI) and its subcontractor, Prof. C. Chang of the University of Maryland at Baltimore County (UMBC), propose a new hybrid framework for target detection and classification using airborne images from two perspectives: spatial and spectral. The spatial approach focuses on texture analysis. The idea is motivated by the fact that the image texture will be disturbed if targets are present. This approach is applicable to EO, IR, hyperspectral, and SAR images. The spectral approach and has two steps. First, a new automatic target generation process (ATGP) is used to generate a set of potential targets from the image data in an unsupervised fashion without using any prior knowledge. Second, an Automatic Target Detection and Classification Algorithm (ATDCA) is used to identify the potential targets. The algorithm can be used to detect anomalies (new and unknown targets) in blind environments. The algorithm is suitable for surveillance operations where the objective is to detect the presence of any potential targets.