机器视觉技术在玉米苗期杂草识别中的应用
|
摘要
杂草严重影响我国的玉米产量,化学除草剂是防治杂草的有效方法,但通常采用粗放的喷洒方式,防治效果较差。精准喷洒极大地降低了除草剂的用量,提高了使用效率。除草剂实现精准喷洒的前提是识别玉米田中杂草位置和种类等信息,而机器视觉是一种开始广泛应用于识别作物田间杂草的方法。为此,设计了一种基于机器视觉的玉米苗期杂草识别方法,采集的图片经过畸变矫正、HSI颜色空间转换和阈值分割后,根据形状和颜色特征提取并识别杂草。田间试验结果表明:在不同播种方式下对玉米和杂草的识别率超过85%,处理单张图片的平均耗时为67ms,能为除草剂的精准喷洒提供信息。
Weed seriously affect the yield of corn in China,chemical herbicides is an effective way to control weeds,but usually by extensive spraying methods,control less effective. Precise spraying greatly reduced the amount of herbicides,improve the efficiency. Herbicides to achieve precision spraying on the premise that the identification of corn fields in the weed position and type of information. Machine vision is a widely used method to identify crop weeds. This paper designs a weed recognition method for maize seedling based on machine vision. After distorted correction,HSI color space transform and threshold segmentation,the collected images are extracted and identified according to their shape and color characteristics. Field experiments showed that the recognition rate of maize and weed was over 85% under different sowing methods,and the average time spent on processing single leaf was 67 ms,which could provide information for the accurate spraying of herbicides.
Weed seriously affect the yield of corn in China,chemical herbicides is an effective way to control weeds,but usually by extensive spraying methods,control less effective. Precise spraying greatly reduced the amount of herbicides,improve the efficiency. Herbicides to achieve precision spraying on the premise that the identification of corn fields in the weed position and type of information. Machine vision is a widely used method to identify crop weeds. This paper designs a weed recognition method for maize seedling based on machine vision. After distorted correction,HSI color space transform and threshold segmentation,the collected images are extracted and identified according to their shape and color characteristics. Field experiments showed that the recognition rate of maize and weed was over 85% under different sowing methods,and the average time spent on processing single leaf was 67 ms,which could provide information for the accurate spraying of herbicides.
引文
[1]李少昆,王崇桃.我国玉米产量变化及增产因素分析[J].玉米科学,2008,16(4):26-30.
[2]张凤海.杂草对夏玉米危害损失率测定[J].江苏农业科学,1998(4):42.
[3]金小俊,陈勇,孙艳霞.农田杂草识别方法研究进展[J].农机化研究,2011,33(7):23-27.
[4]毛文华,王一鸣,张小超,等.基于机器视觉的苗期杂草实时分割算法[J].农业机械学报,2005,36(1):83-86.
[5]纪寿文,王荣本,陈佳娟,等.应用计算机图形处理技术识别玉米苗期田间杂草的研究[J].农业工程学报,2001,17(2):154-156.
[6]龙满生,何东健.玉米苗期杂草的计算机识别技术研究[J].农业工程学报,2007,23(7):139-144.
[7]侯晨伟,陈丽.基于概率神经网络的玉米苗期杂草识别方法的研究[J].农机化研究,2010,32(11):41-43.
[8]李慧,祁力钧,张建华,等.基于PCA-SVM的棉花出苗期杂草类型识别[J].农业机械学报,2012,43(9):184
[9]张红旗,刘宇,王春光.改进田间杂草识别图像预处理方法的研究[J].农机化研究,2015,37(6):70-73.
[10]毛文华,王一鸣,张小超,等.基于机器视觉的田间杂草识别技术研究进展[J].农业工程学报,2004,20(5):43-46.
[11]刘波,方俊永,刘学,等.基于成像光谱技术的杂草识别研究[J].光谱学与光谱分析,2010,30(7):1830-1833.
[12]邓巍,陈立平,孟志军,等.大豆田间杂草的光谱识别研究[J].农机化研究,2012,34(6):118-121.
[13]毛文华,王辉,赵博,等.基于株心颜色的玉米田间杂草识别方法[J].农业工程学报,2009,25(s2):161-164.
[2]张凤海.杂草对夏玉米危害损失率测定[J].江苏农业科学,1998(4):42.
[3]金小俊,陈勇,孙艳霞.农田杂草识别方法研究进展[J].农机化研究,2011,33(7):23-27.
[4]毛文华,王一鸣,张小超,等.基于机器视觉的苗期杂草实时分割算法[J].农业机械学报,2005,36(1):83-86.
[5]纪寿文,王荣本,陈佳娟,等.应用计算机图形处理技术识别玉米苗期田间杂草的研究[J].农业工程学报,2001,17(2):154-156.
[6]龙满生,何东健.玉米苗期杂草的计算机识别技术研究[J].农业工程学报,2007,23(7):139-144.
[7]侯晨伟,陈丽.基于概率神经网络的玉米苗期杂草识别方法的研究[J].农机化研究,2010,32(11):41-43.
[8]李慧,祁力钧,张建华,等.基于PCA-SVM的棉花出苗期杂草类型识别[J].农业机械学报,2012,43(9):184
[9]张红旗,刘宇,王春光.改进田间杂草识别图像预处理方法的研究[J].农机化研究,2015,37(6):70-73.
[10]毛文华,王一鸣,张小超,等.基于机器视觉的田间杂草识别技术研究进展[J].农业工程学报,2004,20(5):43-46.
[11]刘波,方俊永,刘学,等.基于成像光谱技术的杂草识别研究[J].光谱学与光谱分析,2010,30(7):1830-1833.
[12]邓巍,陈立平,孟志军,等.大豆田间杂草的光谱识别研究[J].农机化研究,2012,34(6):118-121.
[13]毛文华,王辉,赵博,等.基于株心颜色的玉米田间杂草识别方法[J].农业工程学报,2009,25(s2):161-164.