摘要:
[目的]探讨基于普通数码影像的单株桃树估产方法,为实现桃树产量的高精度、低成本估测提供参考.[方法]以安徽省滁州市张山桃园为试验区,利用数码相机分别从西北、东南两个方向对抽取的成熟期单株桃树进行拍摄,通过改进的分割算法和特征空间的优化进行果实信息提取,进而选取斑块数量、周长和面积作为特征参数估测单株桃树产量.[结果]采用双方向拍摄的斑块数量之和的建模(y=0.9748x+0.3995)精度最高,预测值与实际值间的决定系数(R2)达0.9049,均方根误差(RMSE)达0.21,而仅采用西北方向拍摄的斑块数量作为建模参数的效果最差,R2仅0.0687,RMSE为0.64.[结论]可以利用普通数码影像提取斑块数量作为特征参数构建估产模型对成熟期果树进行产量预测,其精度较高,该方法具有一定科学性和应用前景.
Abstract:
[Objective]This paper proposed a method for estimating per plant yield of peach based on ordinary digital images to provide reference for realizing high-accuracy and low-cost peach production estimation.[Method]Zhangshan peach orchard in Chuzhou,Anhui was taken as the experimental area.A digital camera was used to shoot selected peach individual plants from the northwest and southeast directions at mature period.The fruit information was extracted through improved partitioning algorithm and characteristic space optimization,and plaque number,perimeter and area were se-lected as the characteristic parameters to estimate the yield of peach per plant.[Result]The results showed that the mode-ling of the sum of plaques photographed bi-directionally(y=0.9748x+0.3995)had the highest accuracy,the determinant co-efficient between the predicted value and the actual value(R2)was 0.9049,and root mean square error(RMSE)was 0.21. But the number of plaques shot from the northwest was the worst for modeling parameters with R2=0.0687 and RMSE=0.64.[Conclusion]The results prove that the number of plaques can be extracted by ordinary digital images as feature pa-rameters to construct an estimation model,and the yield of mature fruit trees can be predicted with high precision.This method is scientific and enjoys application prospects.
百度学术
下载: