Computer Automated Image/DEM Feature Classification and Extraction

The significance in providing a solution for rapid map production lies in the cost savings for the entire mapping community, the benefits to resolving environmental problems on a local and global scale, and to assist in resolving problems of a military nature for the purpose of securing peace worldwide. The key to successfully developing a rapid map production system utilizing high resolution, high accuracy data, is to properly identify the engineering technologies required and to integrate these technologies into a robust and expandable system. The developed rapid map production capability should be portable and transferable to the mapping community at large. This proposal describes an approach for establishing the basis for developing a rapid map production system. A majority of the Phase I work will be to develop, validate, and integrate automated computer algorithms for feature classification and extraction. Both classical PR and neural network (NN) approaches will be investigated. Innovative approaches to the data classification problem are outlined in this proposal, and a testing/verification methodology is also presented. The two source data considered for feature classification and extraction are fine resolution SAR data, and high resolution digital elevation models (DEMs).