矢量多边形模式识别的小波方法

A Wavelet Method for Vector Polygon Pattern Recognition

提出了一种基于小波描述子的矢量多边形的模式识别方法,首先分别计算目标多边形与模板多边形的小波系数矩阵,再通过两个矩阵求取两多边形之间的非相似度,最后通过非相似度来确定是否匹配成功。并且,由所选用的小波的性质,可针对性地计算能够体现多边形特征的系数进行比较,从而使识别效果更好。实验结果表明该方法识别效果好,运算效率高,对平移、旋转、缩放等变换不敏感,是一种有效的矢量多边形模式识别方法。

This paper proposes a vector polygon pattern recognition method based on the wavelet descriptor.In order to minimize the amount of calculation we calculate wavelet coefficient matrix of the target and the template polygons.Using this matrix,we calculate the similarity between these two polygons.We can determine through the similarity,whether or not these two polygons can be matched.Due to the nature of the selected wavelet,the comparison can be made by using the coefficients which can reflect the characteristic of polygon,so that the recognition effect is more effective.Experimental results show that the method is efficient and not sensitive to translation,rotation,and zoom transform.