基于机器视觉的玉米虫害区域SIFT识别仿真  

SIFT Recognition Simulation of Corn Pest Areas Based on Machine Vision

作　　者：刘岩 张宁宁[2] 海玲 王斌虎 LIU Yan;ZHANG Ning-ning;HAI Ling;WANG Bin-hu(Xinjiang Institute of Engineering School of Control Engineering,Urumqi Xinjiang 830023,China;School of Electrical Engineering Xinjiang University,Urumqi Xinjiang 830000,China)

机构地区：[1]新疆工程学院控制工程学院,新疆乌鲁木齐830023 [2]新疆大学电气工程学院,新疆乌鲁木齐830000

出　　处：《计算机仿真》2023年第7期215-219,共5页Computer Simulation

基　　金：2020年新疆维吾尔自治区高校科研计划(XJEDU2020Y043)。

摘　　要：图像中的噪声点会对图像感兴趣区域提取结果造成影响,不利于获取准确的图像识别结果。为了准确识别玉米航拍图像的虫害区域,提出一种机器视觉下玉米航拍图像虫害区域识别方法。通过无人机采集玉米图像,通过中值滤波和K-means聚类算法对玉米航拍图像实施去噪与分割处理,有效保留虫害区域的边缘特征和颜色特征。通过Gentle AdaBoost算法筛选最佳特征,构建强分类器和弱分类器,基于训练好的分类器构建级联识别器,快速排除背景值。利用SIFT模板对识别结果二次筛选,最终确定识别目标,完成虫害区域识别。仿真结果表明,所提方法可以获取高精度的玉米航拍图像虫害区域识别结果,且识别率较高。The noise points in image may affect the extraction result of the region of interest,which is not conducive to obtaining accurate image recognition result.In order to accurately identify the insect pest region of corn aerial image,this paper put forward a method for recognizing the insect pest areas of corn aerial image based on machine vision.Firstly,corn images were collected by UAV,and then these corn aerial images were denoised and segmented by median filtering and K-means clustering algorithm in order to effectively retain the edge features and color features of the insect pest region.Secondly,Gene AdaBoost algorithm was adopted to screen out the best feature,and thus to build a strong classifier and a weak classifier.Based on the trained classifier,a cascade recognizer was constructed to quickly eliminate background values.The SIFT template was used for the secondary screening of identification results.Finally,the identification target was determined,and the identification of pest region was completed.Simulation results show that the proposed method can obtain high-precision recognition result of insect pest area in aerial image of corn,with high recognition rate.

关 键 词：机器视觉 玉米航拍图像 虫害区域识别 级联识别器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]