农田蛾类图像边界特征的获取  

Edge Feature Extraction about Digital Image of Moths in Agricultural Field

作　　者：顾波[1] 张红涛[1] 刘新宇[1] 邱道尹[1] 

机构地区：[1]华北水利水电学院,电力学院,郑州450011

出　　处：《农机化研究》2012年第6期17-21,共5页Journal of Agricultural Mechanization Research

基　　金：中国科学院模式识别国家重点实验室基金资助项目(NLPR2003);河南省自然科学基金项目(2008A520015)

摘　　要：介绍了农田蛾类图像的预处理方法,然后使用Otsu方法来对农田蛾类图像进行分割。对分割得到的目标区域采取旋转变换,使同一种蛾类的目标区域具有相同方向;对旋转变换得到的蛾类图像目标区域,以目标区域的边界点到目标区域的质心点距离作为目标区域的边界特征,获取各种蛾类图像的边界特征。从所获取的边界特征可知,不同种类的蛾类图像具有各不相同的边界特征,可以把这些边界特征作为各种蛾类昆虫分类识别的数据依据。Introduce the image pre-processing method for digital image of moth in agricultural field,after,using Otsu method to segment the digital image of moth in agricultural field;we can get aim area of moth from the segmented digital image.Because the orientation of these aim area of the same kind moth but different individual moth image are different,in order to get the same orientation of aim area for the same kind moth,we should circumrotate these aim areas,when we get the same orientation aim area of the same kind moth,we can extract edge feature by computing the distance of edge pixels to the centroid point for individual moth image aim area.The edge feature is different for different kind moth;we can classify a moth to a certain kind moth if the edge features of the moth resemble a certain kind moth's edge feature.

关 键 词：蛾类 图像分割 边界特征获取 分类识别 

分 类 号：S126[农业科学—农业基础科学] TP391.41[自动化与计算机技术—计算机应用技术]