Study On The Precision Spraying Key Technology In Orchard Based On The Kinect

All kinds of diseases and insect pests often occur.The common method is chemical control.In the process of application,if the method of chemical control is inappropriate,it will cause pesticide residues,and polluting the surrounding environment.The core of target spraying technology is obtaining the target information,to achieve on-demand application,improve the pesticide spraying efficiency.There are several problems in target detection technology:(1)It is difficult to apply the lidar target detection system to practical production with its complex system;(2)The cost of ultrasonic sensor target detection technology is relatively high;(3)The detection distance of infrared light target technology cannot be too far;(4)The response of machine vision target detection technology becomes slow.In this paper,through the Microsoft Kinect equipment,the fruit video is obtained and the color and depth information is integrated based on which the LWA(Leaf-Wall-Area)is detected by the image segmentation.In this paper,an accurate spraying model consisting of LWA average distance and LWA density is proposed,which can automatically adjust the spraying distance and the spraying amount of pesticide.This paper puts forward the calculation method of average distance of the LWA region based on the wall area spline,and the path planning method based on the deviation of spraying path,therefore,the precision spraying path planning of the vineyard is achieved and ultimately the precise effect of spraying the fruit trees is achieved as well.The main work is as follows:1.To solve the problem of the detection of LWA is easy to be affected by complex background,a method of LWA detection and segmentation is proposed,which combines the advantages of color and depth information.This paper presents a design scheme of experimental condition and experimental system.Through image preprocessing,color stratification,image segmentation,etc.the image collected from the orchard is processed,which was compared to the depth image obtained by Kinect synchronization.Finally,after the image expansion and the region filling,the final LWA region image is obtained,which lays the foundation for the pesticide spraying area and precise spraying.2.To realize intelligent control of spraying dosage according to the density of leaf wall,the method of LWA density detection is proposed.A mathematical model for calculating LWA density is proposed in this paper.Dividing the LWA area which has been removed the empty area by total area(including the no-leaf area,such as the empty area and background,the LWA density was obtained,which can reflect the degree of density in a certain region of the wall,so as to adjust the intensity of spraying.By comparing and discussing 20 groups of experimental data of LWA density which obtained from the top,middle and bottom of the fruit trees of peach,apricot and grape,the density distribution of peach and apricot trees is similar,since the LWA density values of the top and middle are the maximum while the density values of the bottom are smaller;t as for the grape vine,the LWA density values of the top are relatively large,however,it is smaller at the bottom.The experimental results show that the results of the LWA density detection are more accurate,and can reflect the distribution of leaf wall density.The results can be used to adjust pesticide spraying amount and achieve the best dosage and effect.3.In order to control the spraying arm and keep the nozzle and LWA within the best range of the nozzle for achieving the best spray effect,the method of best measuring the spraying distance of LWA was put forward.In this paper,the mathematical model for calculating LWA average distance and optimum spray distance are presented,and the experimental results obtained by using the formula are analyzed.Through 20 groups of experimental data,the actual distance measured from fruit tree trunks to Kinect measurements is about the same as the average LWA distance at the bottom of the fruit tree,which means that the average distance calculated by Kinect is accurate.As for the difference value of the bottom LWA average distance of the three fruit trees of peach,apricot and grape,the maximum of the average difference is 8.8472,while the minimum is 0.5080,both of which were acceptable.Through the 20 groups of experimental data,it can be seen that the extension length of mechanical arm is 50 cm,and the adjustment value of actual spraying distance to the best spraying distance in spraying process is within the range of 50 cm,and the experimental results are acceptable.4.For the problem that spraying equipment will deviate from the best spraying position during the procession,the method of precise spraying path planning in vineyard is proposed.Combined with the depth of the image data,the paper put forward the calculation method of average distance of the LWA region based on the wall area spline,which is used to calculate the distance between the LWA and the spraying equipment,and further puts forward path deviation calculation method and path correction planning method,in order to achieve the path correction and planning when the spraying equipment deviates from the optimum spraying,so that the spraying equipment can be maintained in the best path.Through the 30 sets of experimental data,the standard deviation between the calculated value and the measured value on the left sides is 3.7cm and on the right sides is 1.9cm,and the difference is small.In order to verify the accuracy of the calculated path deviation,the variance of the measured and calculated path deviations for the 30 data acquisition points is 2.29 cm,which proves that the difference between the two values is small.