论文部分内容阅读
A practical neural network model formorphological opening and closing filters with its opti-mal parameters training algorithm is proposed in thispaper.For application to motional image target detec-tion,dynamic training algorithm is applied to the de-signing process using asymptotic shrinking error andappropriate network weights adjusting.Experimentalresults show that the algorithm has invariant propertywith respect to shift,scale and rotation of moving tar-get in continuing detection of moving targets.