节点文献
一种基于不变距和PNN分类网络的图形识别方法
A Recognition Method of Graphic Pattern Based on Invariant Moments and Probabilistic Neural Network
【摘要】 图形识别是模式识别的一个重要分支。通过对待识别图形提取7个不变矩,然后将其送入概率神经网络加以识别,以达到对图形进行分类的目的。试验结果表明,该方法不仅能对通常的图形进行分类,而且能对有缺损、畸变的图形进行分类。
【Abstract】 Recognition of graphic pattern is a branch of pattern recognition. In this paper, firstly, seven invariant moments of graphics as well as its features are obtained. Then, these features are sent to probabilistic neural network for recognition. Experiments show that this method not only can classify ordinary graphic patterns, but also can classify defective or lopsided graphic patterns.
【关键词】 概率神经网络;
不变矩;
图形识别;
【Key words】 probabilistic neural network; invariant moments; graphic recognition;
【Key words】 probabilistic neural network; invariant moments; graphic recognition;
【基金】 湖南省教育厅科研基金资助项目(01C244)
- 【文献出处】 长沙交通学院学报 ,Journal of Changsha Communications University , 编辑部邮箱 ,2003年01期
- 【分类号】TP391.41
- 【被引频次】11
- 【下载频次】216