Target detection through image processing and resilient propagation algorithms

This paper deals with target detection studies using the image processing m ethod as well as resilient propagation-based neural network paradigm. In th e resilient propagation-based algorithms, the pre-processing operation to e xtract features of relevance is done using the moment invariance method. Th ese features are then fed as input to the resilient propagation neural netw ork. RPROP (resilient propagation) is an adaptive technique based on the st andard backpropagation algorithm. This RPROP algorithm is also implemented in ADSP-21062 assembly language, since a digital signal processor (DSP) exe cution is much faster than the normal PC execution, as speed is desirable i n real time. It is observed that the resilient propagation-based target det ection is better compared to the image processing method of target detectio n. The main objectives of the paper are the demonstration of the applicabil ity of moment invariant features to neural network-based target detection m ethod and implementation of the technique using a DSP chip, ADSP-21062. (C) 2000 Elsevier Science B.V. All rights reserved.