【摘要】 设计一种基于无人机影像的水稻群体监控系统，预处理获取图像后使用多种模型计算植被覆盖度，对比模型准确度。结果表明：Otsu灰度阈值分割、逻辑回归和朴素贝叶斯的模型算法误差在4%～5%之间，误差度在可接受范围内；KNN算法中，K取不同值时模型准确度有较大差别，K=4时误差最小为3.96%;Kmeans算法中，K取不同值时模型准确度也有较大差别，K=4时误差最小为2.56%。更多还原

【Abstract】 A rice population monitoring system based on UAV images was designed, and after pre-processing and obtaining images,multiple models were used to calculate vegetation coverage and compare the accuracy of models. The results show that the error of Otsu gray threshold segmentation, logistic regression and naive Bayes is between 4%～5%, within the acceptable range; In KNN algorithm, the model accuracy is different when K takes different values, the minimum error is 3.96%; In Kmeans algorithm, the model accuracy is different with different values, and the minimum error is 2.56% when K=4.更多还原