中文摘要：

研究了基于视频的储粮害虫检测与基于粗糙集的储粮害虫图像识别技术，通过对采集的视频图像序列进行运动目标检测分割，提取静态储粮害虫图像；通过对储粮害虫图像进行预处理，直接将图像像素值作为条件属性构造决策表，通过基于遗传算法的属性约简，有效压缩规则集，依据计算得到的规则隶属度输出图像的匹配结果，实验表明采用粗糙集方法识别储粮害虫图像能够取得比较好的效果。

英文摘要：

The vedio detection and rough set was introduced into the detection and recognition of grain pests. The static picture of grain pests was distilled by detection and division of moving object from a series of real time pictures. Pixels were regarded as condition attributes directly to setup decision table, and reduce regula-tion set and condition attributes greatly by using Genetic algorithms. At last, correct results were obtained by calculate subjection functions for the regulations. The results of experiments showed that it has a good affection by using rough set.