Abstract: Morphological characteristics of leaves could be diversified, and the leaf shape is significant to botanical research. In this research, contour characteristic extraction is used to store leaf shape into one-dimensional sequence so that the similarity between different contours of leaf shape can be estimated. Combined with statistics, Pearson product-moment is adopted to calculate correlation coefficient of two characteristic distributions. It’s suggested that the difference of correlation coefficients among different leaf shapes are significant, and the leaf shape recognition test is with high accuracy. This method can be applied as an essential measure to estimate leaf shape similarity.

Key words: leaf shape, contour, characteristic extraction, similarity

摘要： 叶片的形态特征多样，它的叶形特征对植物的研究具有重要的意义。为了判断不同叶形之间的轮廓相似性，利用轮廓特征提取，将叶子的叶形存储在一维序列中。结合统计学，引用Pearson相关性进行特征分布的相关系数计算。实验结果显示，不同叶形的相关系数之间有较为显著的差异，测试的叶子叶形判别准确率较高，可作为判断叶形相似性的重要依据。

关键词: 叶形, 轮廓, 特征提取, 相似性