Prototype neural network processor for multispectral image fusion
Joseph H. Kagel, John Reeder · 2002
Describes the design of a prototype neural network processor that can be trained to classify terrain/materials by fusing multispectral imagery. The authors have developed a system hosted by a PC, consisting of a circuit board containing a neural network chip and all associated circuitry, three input/output boards, and software. They have designed and fabricated the neural network chip and board, and are currently integrating them, the other hardware components, and the software. The authors describe the problem domain and some software simulations in support of this effort, and the test results utilizing Landsat imagery and synthetic subpixel target imagery. They also describe the architecture of the system.>