论文部分内容阅读
本文提出了一种选择性注意相似度量、并构造了一种用于模板匹配分类的网络模型（SelectiveAttentionalTemplateMatchingNetwork）.该网络由匹配子网和竞争子网络成：匹配子网络完成输入模式与样本模式的选择注意性比较；竞争子网络则选择最佳匹配的模式作为分类结果。本文提出的选择性注意相似性度计不便叫用于二值模式问的比较，也可用于连续模式问的比较实验证明该相似性度量较若干种常用相似性度量更能突出反映模式问的差异．本文将给出网络实现方案及实验结果
This paper presents a selective attention similarity measure and constructs a network model for template matching classification (SelectiveAttentionalTemplateMatchingNetwork) .The network consists of matching subnetworks and competing subnetworks: the matching subnetworks complete the input mode and the sample mode Select the noteworthy comparison; competitive sub-network is to choose the best matching model as the classification result. The inconvenience of the selective attentional similarity measure proposed in this paper is that it is used to compare the binary patterns and also can be used to compare the continuous patterns to prove that the similarity measure is more prominent than the common measures of similarity . This article will give the network to achieve programs and experimental results