谷物害虫检测与分类识别技术的研究及应用

Research and Application of Grain Pests Detection and Classification Technology

阐述了在仓储物害虫检测与分类识别的方法,提出了检测系统的硬件组成,对检测到的谷物害虫图像的一阶灰度值直方图和图像的目标区域,自动提取静态仓储物害虫图像的数理统计特征、纹理特征和几何形状特征的主要技术和方法。试验结果表明,该方法可为谷物害虫的计算机自动模式识别(快速分类)提供稳定的特征参数值,有效提高了识别率,为仓储物害虫的快速鉴定和分类研究开辟了新途径。

This paper presents the technique of the detection and classification of stored-product pest and provides the hardware composition of the detection system. With reference to the first order gray histogram of pest images and their graphic target zones, a new technique and method are provided to extract the mathematical and statistic feature, texture feature and geometrical configuration features of the pests are acquired. As experiments demonstrates, this digital image processing technique can reveal a range of reliable eigenvalue and effectively increase the efficiency of identification. This technique has become a new vehicle to the classification and identification of stored-product pest.