Research On Plant Growth Status Monitoring System Based On Vision And Image

In agriculture,the growth status of plants affects the subsequent fruits and the harvest in the next few years.In order to grasp its growth status in time and control it artificially,it is necessary to build a monitoring system.For a single plant,its growth status was measured by plant height,crown size,number of branches and leaf area.This paper mainly studies the leaves of plants.For the leaves of plants,their growth status is mainly reflected in the occurrence of diseases and insect pests and their severity.Citrus plants are widely planted in the world.They are widely planted in southern China.50% of them are destroyed by different plant diseases every year,which results in low yield and poor quality of citrus crops,and has a negative impact on citrus production in the next few years.In the past ten years,computer vision and image processing technology have been widely used in agriculture,and have played an important role in crop harvesting,pesticide spraying,weed identification and weeding,pest detection and other fields.With the continuous improvement of technology,intelligent agriculture industry will be built in an all-round way,which plays a key role in the development of human beings.In order to monitor the growth status of citrus plants and detect the types and severity of diseases and insect pests as soon as possible,this paper takes the leaves of Thai pomelo as the research object.Aiming at two kinds of common diseases and insect pests in citrus plants,different image preprocessing,segmentation,feature extraction and classification methods are studied,and a disease degree classification table is designed to build the whole system.The main work is as follows:(1)Image acquisition and preprocessing.Select the image acquisition equipment,acquisition equipment and transmission method,and carry out a series of pre-processing steps such as image enhancement,morphological processing,to improve the quality of the image and enhance the processability of the image.(2)Patch segmentation and feature extraction.By analyzing and comparing the two segmentation methods of maximum expectation algorithm,maximum inter-class variance method,as well as the three features of color,texture and geometry.The maximum variance method is used to segment the lesion image from the background image,and the feature values of color and texture features are used to describe the lesion image.(3)Species identification and degree classification of pests and diseases.By using the main eigenvalues extracted above,the advantages and disadvantages of four algorithms,namely,support vector machine,nearest neighbor rule classification and decision tree,are analyzed and compared,and the support vector machine method is selected as the method used in this system.A classification table of the severity of diseases and insect pests was designed,and the ratio of disease spots to leaf area was used as a criterion to describe the growth status of plants.(4)System construction and experiment.User interface is designed by Labview software,and the whole system runs by calling the program in matlab.