Sensor Integrated Data Compression

State of the art and coming optical sensors in surveillance applications are complicated in the way data is represented and treated. Often the amount of data is very large and automatic treatment is necessary. One example is Automatic Target Recognition (ATR), frequently used in military applications and a coming technique for civilian surveillance applications. As soon as sensors work together in a network the size of data to communicate between them becomes a problem. Automated analysis has different demands on data quality than a human observer. Existing image compression methods are optimized on human vision, not automated analysis. In such a system compression must be done in a way that considers the analysis method used. This project takes advantages of the results from Sensor Integrated Data Compression and provides deeper theory and solutions for this new approach of data compression for automated analysis. This project will test detection methods on data compressed using COTS coders. When the project is finished there will be knowledge of how compression affects the reliability of the detection result. First year's experiments on IR data and SAR data are presented in this report. In collaboration with Linköping university initial studies of compression of Hyper spectral data is performed.