嵌入式植物自动识别系统的设计与实现

Design and implementation of an embedded automatic plant recognition system

采用嵌入式技术和图像处理技术,通过提取叶片相对稳定的形状特征实现对植物种类的识别分类。系统基于Ubuntu 10.04,采用Qt Creator1.3.1在北京博创公司的UP-NETARM2410-S开发板上进行了实现。功能包括植物叶片的采集和图像拍摄、图像预处理(图片灰度处理及轮廓提取)、图像特征提取(包括叶片的圆形度、偏心率等特征)、图像识别这4个步骤。实验结果表明:该系统可以比较准确地实现对银杏Ginkgo biloba,樟树Cinnamomum camphora,无患子Sapindus saponaria等9种植物叶片的识别分类。

To identify plant categories based on steady characteristics of leaves with embedded techniques and image processing technology, a system was designed and implemented based on Ubuntu 10.04, Qt Creatorl.3.1, and UP-NETARM2410-S board developed by Bochuang Company. The system includes four steps： （ 1 ） collecting leaves and taking images, （2） preprocessing images （transform them into grayscale images and extracting the contours）, （3） extracting shape features （including leaf shape, complexity, and eccentricity）, and （4） matching and recognizing the leaves. The results showed that this system can accurately recognize Ginkgo biloba, Cinnamomum camphora, Sapindus saponaria, and plant leaves from nine other species. [Ch, 6 fig. 1 tab. 13 ref. ]