We develop a decimation-decomposition processing technique that consists of judiciously selecting certain decimation-decomposed components of an image and then performing inter-component processing. For a (k(x), k(y))-decimation decomposition, there may be up to k(x)k(y) decimation-decomposed components. The minimal surviving and maximal non-surviving lengths associated with inter-component processing algorithm allows for clutter suppression. By removing detection redundancies, one can locate the target at high speed. A "where-then-what" model is proposed for target tracking and recognition. It locates the target by-image decimation-decomposition processing first and then recognizes the target in question using a suitable image recognition technique. (C) 2001 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved. [References: 18]
展开▼
