基于机器视觉的田间飞行害虫自动检测方法

害虫数量的精准统计，对害虫的综合治理有重要的意义。传统统计害虫的方法是在固定植株上数害虫数量，难以统计受惊飞走的害虫。采用固定位置放置粘虫板捕捉害虫并自动识别，在害虫正常生活习性下，可有效解决飞行类害虫难统计、信息不准确的问题。同时利用自动阈值分割、目标粘连处理、目标识别和利用生物特征干扰去除等机器视觉方法，有效统计田间飞行害虫数量，识别准确率＞85%，为病虫害防治提供依据。 

Automatic Detection Method of Field Flying Pests Based on Machine Vision

Accurate statistics of pest number is of great importance to comprehensive pest management.It is difficult to calculate flushing away pests by traditional method of statistical pest which is to count number of pests on fixed plant.Place sticky board to catch insects and automatically identify in a fixed position，recognition in the normal life habit of pests，which could effectively solve problems of inaccurate statistics and inaccurate information of flying pests.At the same time，methods of automatic threshold segmentation，target adhesion processing，target recognition and biological feature interference removal were adopted to effectively calculate number of insect pests in the field.Recognition accuracy was above 85%，which provided basis for pest control.