Proceedings of 1994 28th Asilomar Conference on Signals, Systems and Computers (1994)
Pacific Grove, CA, USA
Oct. 31, 1994 to Nov. 2, 1994
Y.T. Zhou , HNC Software Inc., San Diego, CA, USA
R. Hecht-Nielsen , HNC Software Inc., San Diego, CA, USA
Unlike traditional automatic target recognition (ATR) approaches, the foveal vision based approach uses a foveated sensor to dynamically allocate visual resolution spatially and temporally to objects relevant to the task. It mimics human eye movements to find "attractive" regions quickly and accurately. A software demonstration system has been built based on the foveal vision approach and tested on a variety of sensor data including TV, infrared (IR), and laser radar (LADAR). Test results indicated that the foveal vision based approach not only supports a decrease of several orders of magnitude in the amount of data processing but also provides an increased ability to discriminate targets.<
computer vision, military systems, image sampling, eye, multilayer perceptrons, feedforward neural nets, radar target recognition, image resolution, military computing
Y. Zhou and R. Hecht-Nielsen, "A foveal vision based approach to ATR," Proceedings of 1994 28th Asilomar Conference on Signals, Systems and Computers(ACSSC), Pacific Grove, CA, USA, 1995, pp. 874-878.