Huangguan Pear Tree Phenophase Intelligent Recognition Technology Research

In order to improve the competitiveness of China’s agriculture,accelerate the deepening of agricultural informatization and agricultural modernization,we must speed up the breakthrough of agricultural key core technology.Fruit trees as one of the main categories of agricultural development,intelligent recognition research in fruit tree production has received more and more attention.The phenology study of fruit trees is of great significance to the development of smart orchards.In order to realize accurate and automatic discrimination of fruit trees at various phenological stages,the 11-year-old Huangguan pear planted in modern Agriculture Demonstration Base of Hebei University of Technology in Yongnian District,Handan City,Hebei Province was used as the test object in this study.The following research work was carried out:(1)Five key phenological periods of yellow crown pear in southern Hebei region were summarized by artificial field observation of phenological periods of Huangguan pear in the experimental area for two consecutive years,and the image sample data of each key phenological period were collected,which provided the discrimination basis and a large number of original image sample data of phenological period for the realization of image recognition of yellow crown pear;(2)Image processing software is used to cut the size of the collected original image data;(3)Use baidu AI Studio development platform of the processed image is cut sample image preprocessing,and then based on the image of the sample data after pretreatment,Six classical image recognition algorithms including K-proximity algorithm,naive Bayes classifier algorithm,decision tree algorithm,random forest algorithm,logistic regression algorithm and support vector machine algorithm were used to train and test the phenological phase image recognition model.Finally,the algorithm suitable for the phenological phase image recognition of Yellow crown pear tree was analyzed and evaluated.In this study,it is concluded that the random forest algorithm and support vector machine algorithm have excellent performance in the application of yellow crown pear phenology image recognition.Therefore,the two algorithms can be used to train the image automatic recognition model in the application.This study has practical guiding significance for the development and implementation of other fruit tree phenology image automatic recognition research.