基于机器视觉与随机方法的作物行提取算法.pdfVIP

2 0 0 8 年 11 月 农 业 机 械 学 报 第 39 卷 第 11 期 基于机器视觉和随机方法的作物行提取算法 姜国权 　柯 　杏 　杜尚丰 　陈　娇 　　【摘要】　根据农田作物图像特点 ,采用基于垂直投影的窗口移动方法 ,得到代表作物行的定位点 ,提出了基于 机器视觉和随机方法的作物行提取算法 。该算法从定位点中随机选取 2 个不同点 ,决定一条候选直线 ,再根据阈 值规则 ,进一步判断候选直线的真实性 。实验结果表明 ,该算法可以提取不同作物的作物行 ,处理一幅 640 ×480 像素的彩色图像平均需要 220 ms ,正确识别率达 98 % 。 关键词 : 作物行提取 　机器视觉 　随机方法 　Hough 变换 中图分类号 : TP242 6 + 2 文献标识码 : A Detection Al gorithm of Crop Rows Based on Machine Vision and Randomized Method J iang Guoquan 　Ke Xing 　Du Shangfeng 　Chen J iao ( Chi na A g ricul t u ral Univers ity , B eij i ng 100083 , Chi na) Abstract One way of navigating autonomous vehicles to perform field operations is using a vision of crop row s detection system . A new algorit hm for row s recognition was p resented based on machine vision and randomized met hod . During image p rep rocessing , a vertical p roj ection met hod was employed to estimate t he po sition of crop row s. Moreover , t he randomized met hod was used for crop row s detecting. A candidate line was determined by two different localization point s , which were selected from dat a sp ace randomly . The t hreshold rule was adopted to furt her determine whet her t he candidate line is t he desired one . The test result s p rove t hat t he row s detection met hod is suit able for different crop s. The average time required for a 640 ×480 pixels color image is 220 ms and t he recognition rate reaches 98 %. Key words 　Crop row s detection , Machine vision , Randomized met hod , Hough t ransform 它的主要缺点是内存消耗大 、处理速度慢 。尤其是 　　引言