基于超像素分割的田间小麦穗数统计方法

杜颖,蔡义承,谭昌伟,李振海,杨贵军,冯海宽,韩东

Field Wheat Ears Counting Based on Superpixel Segmentation Method

DU Ying,CAI YiCheng,TAN ChangWei,LI ZhenHai,YANG GuiJun,FENG HaiKuan,HAN Dong

表3 施氮样本分类训练结果

Table 3 Classification training results of nitrogen application samples (n=10 000)

分类器类型 Classifier type 总体准确率 Overall accuracy (%) 穗部准确率 Ear accuracy (%) 背景准确率 Background accuracy (%) linSVM 87.1 88.0 86.0 quaSVM 89.2 91.0 87.0 cubSVM 84.0 77.0 91.0 finGSVM 89.2 89.0 90.0 medGSVM 89.5 91.0 88.0 coaGSVM 88.0 90.0 86.0 finKNN 87.4 87.0 87.0 medKNN 89.4 90.0 89.0 coaKNN 89.1 93.0 86.0 cosKNN 88.8 88.0 90.0 cubKNN 88.9 89.0 88.0 weiKNN 90.2 93.0 88.0