The systems currently can search for designated ground objects, selectively transmitting an image for human inspection and making quicker and more selective analysis possible. Based on integration of mathematical, computational, and knowledge-based approaches, the techniques provide quick analysis of photographic images. It can classify, measure, and locate objects as well as compress images to a 200:1 ratio in seconds.Radiant Tin provides a library of algorithms to perform sensor fusion, integration of geospatial data with imagery, autonomous image alignment, and potentially reduced classification products. It also provides image retrieval and regeneration. It supports sight model generation, change detection, man-made feature detection, object editing/removal, and some model generation. Radiant Tin provides automated indexing of images using an object oriented approach in which objects are classified by an operator initially pointing out three instances of an object. The Radiant TIN algorithm might benefit from the use of simple representations of texture that were not deterministic. Radiant TIN Image Analysis (TIA) research focuses on development of a system to analyze imagery and graphics products from disparate sources with varied format, quality, contrast, resolution, and detail level in order to align different images of the same area automatically; to detect differences in these images; to identify and classify objects found in the images; and to provide SENSORREP and other reports for photo-interpreters and tactical users. Radiant TIN Image Dissemination (TID) work focuses on development of image compression (lossy) and related software to disseminate imagery and graphics products via existing low-data-rate tactical communications paths to combat units that rely on low-end data processors to receive and process operational intelligence. Typically, lossy image compression is achieved through transformation, quantization, and coding. The Radiant TIN technique, on the other hand, decomposes the original representation of the pixel image into a set of attributed symbols that faithfully represent the image. These symbols are separated from the surrounding textures and encoded with considerable compression-over 1:1000. The remaining textures are compressed, avoiding the artifacts that usually appear in images that are compressed and restored with texture-based methods such as wavelets. By separating edges from textures, a higher compression rate can be achieved, since the data concerning edges is absent and the resulting sub-band coefficients are greatly reduced. Symbols generated by Radiant TIN algorithms can also be used to identify edge-based objects such as buildings and texture-based features such as forested areas based in symbol-based descriptions previously stored in the knowledge base. The Radiant TIN algorithms are designed to retain and use as much information about man-made objects as possible.