基于梯度Hough变换的遮挡苹果目标定位  

Location of occluded apples based on gradient Hough transform

作　　者：吴庆岗[1] 张卫国[1] 常化文[1] 金保华[1] 刘朝霞[2] 

机构地区：[1]郑州轻工业学院计算机与通信工程学院,河南郑州450001 [2]大连外国语大学软件学院,辽宁大连116044

出　　处：《浙江农业学报》2017年第6期1009-1016,共8页Acta Agriculturae Zhejiangensis

基　　金：国家自然科学基金项目(61502435;61401404;U1404623);河南省教育厅科技攻关项目(14A520034);河南省高等学校重点科研项目(16A520028);郑州轻工业学院博士基金项目(2014BSJJ077;2013BSJJ041);郑州轻工业学院校青年骨干教师项目(13300093);郑州轻工业学院研究生科技创新基金资助项目

摘　　要：为实现自然环境中被枝叶或其他果实遮挡的苹果目标定位,提出一种基于图像边缘信息的梯度Hough变换的目标定位方法。该方法首先在Lab空间中利用K-means聚类算法对自然环境下苹果图像进行分割,然后对分割结果进行形态学操作以去除小区域,接着采用Sobel算子提取苹果目标的边缘,最后利用梯度Hough变换获取苹果目标的圆心及半径,实现遮挡苹果目标定位。实验结果表明,该方法能够有效定位遮挡苹果,定位重合度高达93.17%。In order to accurately locate the apples occluded by branches or other apples in natural environment,a method based on gradient Hough transform was proposed in the present study. Firstly,K-means clustering algorithm was used in Lab space to segment apples in natural environment. Secondly,morphological operations were made based on the segmented results to remove the influence of small area. Then,Sobel operator was adopted to extract the edge of apples. Finally,gradient Hough transform was applied to estimate the center and radius of target apples.Thus,the occluded apples could be automatically located. Experimental results showed that the method could effectively locate the occluded apples with the precision as high as 93. 17%.

关 键 词：梯度Hough变换 自然环境 目标定位 苹果图像 K-MEANS算法 形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]