Image Processing Technology For Quicklyacquiring Key Growth Parameters Of Greenhouse Tomato

With the development of my country’s economy and society,people’s living standards are gradually improving,with stricter requirements for the quality of fruits and vegetables,and more diverse types of demands.Facility agriculture,characterized by intelligence and automation,is gradually becoming a new direction for agricultural development.During the growth of facility crops,growth parameters such as fruits,flowers,leaves,etc.can fully reflect the growth status of plants.Growth parameter identification can be used to guide agricultural operations such as flower thinning to affect fruit quality and yield,and can also be applied to automated operations.This thesis takes the facility tomato as the research object,and uses image processing technology to quickly obtain the key growth parameters of tomato fruit,flower and leaf.The main innovations and work are as follows:(1)Research on the recognition and location of tomato bunch fruit number based on clustering-improved gray wolf algorithm.The purpose of the research in this section is to quickly obtain the fruit number of tomato bunches,which can be applied to the automatic picking work of tomato picking robots.In terms of algorithm,an image segmentation method combining the improved gray wolf algorithm and K-means clustering algorithm with perturbation factor is proposed to improve the clustering effect and achieve high-quality segmentation.Finally,the Hough circle transform algorithm is used to complete the conglutination of fruits in the image.Segmentation and recognition.The comparison shows that the improved segmentation algorithm has significantly improved the convergence speed,clustering quality and stability.Among them,the running time is shortened by 47%,the structural similarity is increased by 10%,the peak signal-to-noise ratio is increased by 31%,and the average fruit recognition The rate has increased by10.8%,and the recognition rate has been significantly improved,providing technical support for the automated operations of picking robots.(2)Research on the recognition of tomato blossom number and fruit maturity based on improved SSD lightweight neural network.The purpose of this section is to quickly obtain the flowering number and fruit maturity of tomato bunches,which can be used in agricultural management work such as flower thinning and fruit thinning.In terms of algorithms,it proposes an improved SSD lightweight neural network recognition algorithm,introduces the lightweight module Mobile Net V3 into the SSD,adjusts the backbone network,and combines it with agricultural operation experience and inspection robots to guide agricultural operations on the basis of identification.The comparison shows that the model has good real-time performance and robustness in the facility environment.Compared with SSD,the recognition rate is increased by 7.9%,and the recognition speed is0.079 s,which is increased by about 4 times,which meets the application requirements.(3)Study on the identification of tomato leaf geometric growth parameters.The purpose of this section is to quickly obtain tomato leaf length,leaf width,leaf area and perimeter,which can be used in the data acquisition of tomato cultivation experiments.In terms of algorithm,the leaves are spaced and image segmented,and then compared with reference objects,the leaf area is labeled,and the leaf area is predicted through the BP neural network,and the advantages and disadvantages of the two different recognition methods are compared;the outline of the leaf is extracted,And combined with the chain code algorithm to realize the blade circumference recognition;on the basis of the blade segmentation,the smallest circumscribed rectangle algorithm is used to realize the automatic labeling of the blade length and width;the MATLAB graphical user interface is used to develop a visualization system based on the identification of different blade parameters.Visual display of blade geometric parameters.