植物学报 ›› 2005, Vol. 22 ›› Issue (05): 599-604.

• 技术与方法 • 上一篇    下一篇

植物叶形的计算机识别系统

朱静 田兴军 陈彬 吕劲紫   

  1. 1(南京大学生物系 南京 210093) 2(南京大学教务处 南京 210093)
  • 收稿日期:2004-09-20 修回日期:2005-04-15 出版日期:2005-10-30 发布日期:2005-10-30
  • 通讯作者: 田兴军

Computer Recognition System of Plant Leaf-shape

ZHU Jing TIAN Xing-Jun CHEN Bin LV Jin-Zi   

  1. 1(Department of Biology, Nanjing University, Nanjing 210093) 2(Department of Biology, Nanjing University, Nanjing 210093)
  • Received:2004-09-20 Revised:2005-04-15 Online:2005-10-30 Published:2005-10-30
  • Contact: TIAN Xing-Jun

摘要: 植物叶形是识别植物的重要和常用形态特征, 建立计算机自动识别系统对于认识和正确识别植物十分重要。本文论述了植物叶形图像识别系统设计中的图像处理、特征提取及分类识别等问题。本系统采用Visual Basic.Net编程工具设计, 在Windows 2000/XP平台上通过叶片图像的输入、变换、平滑和分割等识别过程,实现了叶片图像的形状和叶缘特征的结果输出。实验结果表明, 该系统能够很 好地识别植物的叶形, 对14种植物337份叶片样本的叶形测试准确率达93.2%。为植物识别的进一步研究奠定了基础。

Abstract: Leaf-shape is a significant feature for plant recognition. Establishing computer recognition system is an essential topic for identifying plants. Image processing, feature abstraction and classification of plant leaf shape with computer recognition system were discussed in this paper. The system has been developed with Microsoft Visual Basic.Net under Windows 2000/XP system. It can accomplish leaf image input, transform, smoothness, segmentation and finding leaf margin, etc. The performance of the system to recognize leaf shape has been demonstrated with 93.8% precision rate through testing 337 samples which belong to 14 species of plants. This work is very essential for further studies.