基于图像处理的储粮低密度虫害监测系统的设计与验证

为解决高害虫密度储粮处理成本昂贵和现有监测系统实时性和移动性不足等问题，采用Web技术，结合自主设计的粮虫诱捕器，建立了储粮低密度虫害实时监测系统。树莓派控制诱捕器采集害虫图像并进行图像处理得到图像中害虫的数量，再将数据传至云端服务器，用户通过Web客户端获取历史以及实时的害虫图像和害虫数量。在实验室用该系统监测了赤拟谷盗密度为0.5、1、2、3、4、5头/kg的稻谷，通过系统捕获第一只害虫的时间来评价其灵敏度，24 h内对害虫的捕捉率验证系统用于低密度虫害监测的可行性，并以人工直接计数结果为参考计算了系统计数的准确率，结果表明：系统灵敏度高，在低密度害虫条件下对害虫的捕捉率高于61.98%且诱捕器捕捉的害虫数与稻谷中的害虫总数存在显著线性关系，系统计数准确率为90.26%。因此，该系统可用于低密度虫害的实时监测。

Design and Verification of Low-density Pest Monitoring System for Stored Grain Based on Image Processing

To solve the problems such as high cost of high-density pest control for stored grains and the lack of real-time and portable monitoring systems, web technology combined with self-designed grain pest traps was used to develop real-time monitoring system for low-density insect infection. The Raspberry Pi control trap collected images of insects and processed images to obtain the number of insects in each image. Then the data were transmitted to the cloud server. Users could obtain the history and real-time insect images and numbers through the web app. In the laboratory, the system was used to monitor the density of the red flour beetle in rice as 0.5, 1, 2, 3, 4 and 5 heads/kg. The sensitivity and feasibility of the system were evaluated based on the time required for capturing the first red flour beetle and the capture rate of the insects within 24 h, respectively, while the accuracy of the counting system was calculated based on the results obtained by direct manual counting as the reference. The results showed that the sensitivity of the system was high, with the capture rate higher than 61.98% under low-density pest conditions. The number of the insects captured by was the trap highly correlated with the total number of insects in grains. The counting accuracy of the system was 90%. Therefore, the system can be used for real-time monitoring of low-density pests during grain storage.