基于图像的植物种类识别与植物三维建模

项目名称： 基于图像的植物种类识别与植物三维建模

项目编号： No.61501464

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 刘佳

作者单位： 北京信息科技大学

项目金额： 19万元

中文摘要： 本项目研究基于图像的植物种类识别与植物三维建模，对于提高植物分类效率、普及植物学知识、认识和保护植物资源具有重要意义。主要研究内容包括：基于相似性的植物图像分割、基于植物分类学的叶片分类和冬树识别、基于视觉方法的植物参数提取，以及植物器官重建和基于生长路径优化的群体植株建模。关键科学问题为：前景与背景颜色相似、细小结构多、空洞多的叶片或植株图像分割、叶片有效特征提取以及冬树特征提取、立体视觉用于计算植物参数的可行性、符合植物学机理的群体植株建模。项目的创新性在于笔画引导与形态先验约束相结合的优化分割、基于植物分类学的叶片特征提取、基于树木树枝整体形态特征的冬树识别，以及基于局部重建和整体形态模拟的植物群体建模。

中文关键词： 图像分类；叶片识别；三维建模；图像分割

英文摘要： The research of image-based plant recognition and modeling have important significance for improving plant classification efficiency, popularizing botanic knowledge, recognizing and protecting plant resources. In this project, the main contents include plant image segmentation based on similarity, leaf classification and winter tree recognition based on phytotaxonomy, parameter measurement based on vision technology, and organ reconstruction and group of plants modeling based on growth path optimization. Key problems are as follow: leaf and plant image segmentation with complex background and tiny structure, feature extraction for leaf and winter tree, the feasibility of using stereo vision for plant parameter measurement, and the modeling of group of plants based on botany theory. The innovativeness exist in combing sketch guidance and prior shape restrict for optimized segmentation, leaf feature extraction based on phytotaxonomy, winter tree recognition based on whole branch structure, and group of plants modeling through reconstructing parts and simulating the whole structure.

英文关键词： image classification;leaf recognition;3D modeling;image segmentation