中文摘要：

为了准确高效地识别树木叶片,开发了一款基于Android操作平台的树木叶片设别系统。该系统提取13种树木叶片特征描述子,选择支持向量机作为分类器。该系统包括图像获取、图像处理、特征提取、分类识别和结果展示5个模块。选取来自15个树种的1 500片树叶进行了试验,结果表明,该系统的平均识别率可以达到94.44%,优于BP神经网络的91.56%,达到了令人满意的效果。该系统特征描述子的筛选、提取以及分类器算法还可以进一步优化,以更好地满足用户需求。

英文摘要：

In order to accurately and efficiently identify tree leaves, a tree leaves identification system based on Android mobile phone is proposed in this paper. 13 feature descriptors are extracted, and support vector machine（SVM） is selected as the classifier. The system includes image acquisition, image processing, feature extraction, classification and result display. 1 500 leaves from 15 tree species are used in this experiment. The results are compared with the controlled group, whose classifier is back propagation neural network（BPNN）. It shows that SVM identification rate is 94.44%, BPNN identification rate is 91.56%. Besides, SVM is more efficient than BPNN. The performance of this system is satisfactory,but feature descriptors and classification algorithm are required to be optimized.