Extensible PDE and Wavelet Tools for Tactical

The military is increasingly making use of various types of imaging sensors, such as SAR, ISAR,IR, and optical, for tasks such as surveillance, precision strike, automatic target recognition, and rapid retargetting. Traditional techniques for image enhancement and segmentation have a limited utility for military applications using the kinds of sensors. In this research, we will develop an open and extensible software library of multiresolution and multiscale image enhancement and image segmentation techniques that overcome many of the problems with traditional methods. Multiresolution methods based on adaptive wavelet analysis have been shown to be very useful for a variety of image analysis problems such as compression, edge detection, and denoising. Similarly, multiscale methods based on variational methods and partial differential equations (PDE) have also shown much promise for image enhancement and segmentation. The specific goals of the Phase~I research will be to evaluate the state of the art in multiscale and multiresolution algorithm in terms of the key features required for target-based image classification. We will develop key multiscale and multiresolution algorithms for image enhancement and segmentation incorporating several novel ideas and evaluate their efficacy for classification of tactical images.