| Rice planthoppers and other pests and diseases are important culprits in reducing rice production,seriously affecting the safety of food production in China.Real time and quantitative monitoring of rice planthoppers can not only predict the prevalence and trend of diseases and pests in a timely manner,but also provide precise guidance on pesticide use,improving rice yield and quality.Therefore,this article focuses on the problems and requirements of rice planthopper monitoring,and designs and develops a rice planthopper pest monitoring device based on image processing.The device involves mechanical structure design,hardware system design,control program design,and rice planthopper image recognition algorithm design.The mechanical structure includes insect trapping unit,killing unit,image acquisition unit,and rain blocking unit.The bait unit selects specific wavelength LED lights based on the phototactic behavior of rice planthoppers;The killing unit uses a resistance heating method with precise temperature control to kill pests;The image acquisition unit achieves linear scanning image acquisition through the cooperation of a conveyor belt and an industrial camera;The rainproof unit achieves rainproof function through lifting and lowering movement.Each unit is supported and fixed by profiles to form the mechanical body structure of the device.The hardware system design includes modules such as the main control unit,power supply,pest control,heating control,and image acquisition,respectively completing the device selection and circuit design for each module.The main controller uses an STM32 microcontroller,and the environmental parameter sensor and touch screen communicate with the main controller through RS485 interface and USART serial port,respectively.The pest control and heating control are both achieved through electromagnetic relays for opening and closing.The power module integrates a solar charging circuit and outputs 5V,12 V,and 24 V DC power to each component using the LM2595 S chip.The image acquisition unit uses Raspberry Pi 4B as an auxiliary controller,which communicates with the image sensor MV-CE200-10 UC through a USB interface.The software program consists of two parts: the main program and the auxiliary program.The former designs the driver programs for various units such as environmental parameter sensors,insect traps,stepper motors,and interactive interfaces.Under the control of the main program,it sequentially performs work such as checking working conditions,trapping insects,and killing insects.The key parameters of the control process can be configured through the human-machine interaction interface.The latter includes functions such as image acquisition,rice planthopper image recognition,and data upload.After receiving the image acquisition command from the main controller,Raspberry Pie started the camera to take pictures of the pests on the conveyor belt,and then called the target detection model to identify and count the rice planthoppers.The rice planthopper target detection model is designed based on the Yolov4-tiny target detection algorithm.The model was constructed using4201 rice planthopper sample images for training and testing.The m AP and PFS values of the model are 92% and 89.52,respectively,with a weight size of 23013 kb.After image collection and recognition,upload the images and result data to the cloud server to build a complete monitoring loop.Finally,the heating response time,power supply,and overall performance of the killing unit of the device were tested separately.The test results showed that the heating temperature of this device can reach 100 ℃,and a temperature time response curve was obtained;The average charging time of the power supply and the operating time of the device are 8.02 hours and 10.24 hours,respectively;The average monitoring accuracy of the whole machine test is 74.2%,which can quantitatively and accurately reflect the changes in the number of rice planthoppers.Each performance index of the device meets the Functional requirement.The pest monitoring design scheme proposed in this paper is correct and feasible,and has practical application value. |
