关键词: 叶片识别, 形状特征, Android, [K]近邻算法-支持向量机（KNN-SVM）

Abstract: As to the problems of traditional plant recognition methods for heavy workload, inefficiency and hard to guarantee the objectivity of the data, a plant leaf recognition algorithm based on shape features is presented and an Android application of online plant leaf recognition based on C/S model is developed in this paper. Firstly, the leaf image is preprocessed to get the leaf contour. Then, the traditional shape features and two new ones, the contour convex ratio for plant leaf and the variance of contour curvature, are extracted. Finally, the plant is recognized by using KNN-SVM. Experimental results show that the higher recognition rate can be obtained by using the proposed method compared to some existing algorithms. Also the Android application is stable and reliable for user’s requirement.

Key words: plant recognition, shape features, Android, K-Nearest Neighbor-Support Vector Machine（KNN-SVM）