Research On The Key Technologies Of Real-time Detection Device For Pesticide Droplets

Pesticides are an important means to control pests and diseases,but the backward application of pesticides and technology has led to a pesticide utilization rate of only 35%.Effective evaluation of spray quality can help improve the application of machinery and technology.Two important evaluation indicators are droplet size and distribution uniformity.In the production practice,water-sensitive paper is generally used to detect pesticide droplets,but there are problems such as repeated arrangement,high labor intensity,vulnerable to environmental humidity and cannot be reused,and the droplets on the surface of the water sensitive paper are prone to sticking and will decrease droplet detection effect.Therefore,the development of a high-efficiency,accurate and low-cost pesticide droplet real-time detection device has practical significance such as simplifying operation,improving detection accuracy and reducing cost.In this paper,image processing technology is used to realize real-time detection of pesticide droplets.The mist drop collecting device was fabricated,and the fog drop detection system was developed.The concave point method and the iterative open operation were combined to complete the optimization design of the adhesion segmentation algorithm.The optimization algorithm,the droplet detection system and the acquisition device are integrated to complete the prototype development of the pesticide droplet real-time detection device.The experimental results show that the optimization algorithm improves the accuracy of the adhesion segmentation.The measurement results of the laser particle analyzer are used as reference.The error rate of the droplet diameter is less than 9%.And verified the test results through field experiments.The specific research work is as follows:(1)Design of the mist collection device.The droplet collection device includes a quick release bracket,an image sensor,a cabinet,and a target.The UG12.0 and AutoCAD2014 were used to design the 3D model and the 2D engineering drawing respectively,and the 3D printing and laser cutting technology were used to complete the processing of the device.Quick release bracket can be adjusted by adjusting the knob to change the Angle and height,the height can reach 70 cm.The image sensor is OV2710 industrial camera,which can meet the image quality requirements of the device.The box body includes a black box and a packaging box,which are made of acrylic and PLA respectively.The target is a transparent acrylic sheet of size 130×79×2(unit: mm).Through the droplet collection test,it is verified that the device can adapt to different collision angles and droplet sizes,and the collection effect is good.(2)Development of the fog drop detection system.The system consists of a lower computer,a data transmission module and a host computer.The lower computer is a 3B Raspberry Pi.The development environment of OpenCV 2.4.9 and Qt5.4.1 is used to develop the fog image processing program.The hardware part of the data transmission module includes controllers such as CC2530,ESP8266 and STM32F103RCT6.Keil5 and IAR were used to write software,including serial communication program,UDP client program and ZigBee communication program.The data transmission module receives the droplet characteristic parameter data from the lower computer and forwards it to the upper computer;the upper computer is a Lenovo Z40 computer.Qt5.10.0 is used to develop the upper computer program,including human-computer interaction interface,UDP server program and MySQL database operation program,which are respectively used for parameter display,sending and receiving data and database management.(3)Optimization design of the adhesion segmentation algorithm.An iterative opening operation is introduced on the basis of the pit method to segment the weakly adhering droplets and enhance the degree of complex adhesion droplets.The positional relationship between the midpoint of the two points adjacent to the current point and the contour of the droplet is the basis of the pit search.The representative point is selected by the angle threshold and the Euclidean distance from the remaining pit to be matched is calculated.The nearest pit is selected.Temporary matching object,if the two matching points form a straight line in the area where the representative point is connected with its previous point and subsequent point,the matching is considered to be correct.The droplet counting test was carried out using a selfmade droplet collecting device and water sensitive paper as artificial targets,respectively.Among them,the device count error rate is less than 6%,and the water sensitive paper count error rate is less than 8%.The adhesion segmentation effect is better than the Dajiang droplet analysis software and the Deposit Scan software.Finally,with the spray test platform,under the hydraulic pressure of 0.3MPa,the three types of nozzles F110-01,F110-03 and F110-015 produced by TEEJET were used to carry out multiple sets of tests.The error rate of the droplet diameter detection was lower than that.9%.Field experiments under different light intensities were carried out and it was concluded that the detection was best in the morning or evening.