作物种植行自动检测研究现状与趋势
陈鹏飞,马啸

Research Status and Trends of Automatic Detection of Crop Planting Rows
CHEN PengFei,MA Xiao
表1 不同作物种植行提取方法的基本原理、优缺点及适用范围
Table 1 Principles, advantages and disadvantages, and application scope for different crop planting row detection method
分类方法
Class methods
基本原理
Principle
优缺点
Advantage and disadvantage
适用范围
Application scope
Hough变换法
Hough transform method
将图像空间的点变换为参数空间中的线,统计参数空间的峰值点,最终将峰值点按一定映射关系在图像空间中形成相应的种植行
Transform the points in the image space into lines in the parameter space, and then count the peak points in the parameter space, and finally form the corresponding planting rows in the image space based on peak points according to the certain relationship between the two spaces
能同时进行图像中多条种植行的提取,但也存在计算量大、容易受影像中杂草像元干扰等缺点
It has advantage that can extract multiple planting rows in the image at the same time, but it also has disadvantages such as long calculation time and easy to be affected by weed pixels
杂草压力小、数据量小的影像中作物种植行提取;可同时在无人机影像及田间农机拍摄影像中应用
It is suitable for planting row detection in images with a low weed pressure and small data volume; It can be used with both drone images and images taken from agricultural machinery
最小二乘法
Least square method
基于获取的种植行特征点,利用最小二乘拟合的方式获取种植行
Based on the obtained represent points of planting row, the least squares fitting method is used to make the planting row
具有算法简单、计算速度快等优势,但存在容易受图像中杂草像元干扰的缺点
It has advantages of simple algorithm and fast calculation speed, but it also has disadvantage of being easily affected by weed pixels
杂草压力小的影像中作物种植行提取;可同时在无人机影像及田间农机拍摄影像中应用
It is suitable for planting row detection in images with a low weed pressure; It can be used with both drone images and images taken from agricultural machinery
绿色像元累加法
Green pixel accumulation method
假定沿作物行方向上会有最大的植被绿色像元累积,据此提取作物种植行
The crop planting row is detected according to the assumption that there will be the largest accumulation of vegetation green pixels along the direction of the crop row
具有算法简单、计算速度快等优势,但存在容易受图像中杂草像元干扰的缺点
It has advantages of simple algorithm and fast calculation speed, but it also has disadvantage of being easily affected by weed pixels
杂草压力小的影像中作物种植行提取;可同时在无人机影像及田间农机拍摄影像中应用
It is suitable for planting row detection in images with a low weed pressure; It can be used with both drone images and images taken from agricultural machinery
Blob分析法
Blob analysis method
假设图中一个植被的连通区域即代表一个作物行,可以通过Blob分析确定每一个作物行连通区域的重心与长轴,从而生成种植行
Assuming that a connected area of vegetation in the image represents a crop row, Blob analysis can be used to determine gravity and long axis of each connected area to generate planting row
能耐受一定的杂草压力,但要求种植行必须呈直线分布
It has advantage that can endure a certain weed pressure, it also has disadvantage that require planting rows must be linearly distributed
杂草压力不大且影像中作物行呈直线的种植行提取;可同时在无人机影像及田间农机拍摄影像中应用
It is suitable for planting row detection in images with a low weed pressure and linearly distributed; It can be used with both drone images and images taken from agricultural machinery
滤波法
Filtering method
假定作物种植行呈特定的特征和图案,这些特征和图案可以用函数描述,基于函数来确定种植行位置
Assuming that the crop planting rows have specific features and patterns and these features and patterns can be described by function, the planting row can be determined based on the function
具有不易受杂草干扰的优势,但要求作物种植行按一定规律出现,当不存在这种规律时会产生误差
It has advantage of not being easily affected by weed pixels, but it also has disadvantage that require crop planting rows must distribute with a certain rule, and errors will occur if there exist no such rule
影像中作物种植行分布呈现规律分布;可同时在无人机影像及田间农机拍摄影像中应用
It is suitable for planting row detection that are distributed with a certain rule in images; It can be used with both drone images and images taken from agricultural machinery
消隐点法
Vanishing point method
田间拍摄影像近处的种植行显得间隔比较大,而远处的显得间隔比较小,如果把图像中作物种植行进行延长,它们会交于一点(即消隐点),根据这一特征可进行作物种植行的提取
In the field image, planning rows in the vicinity appear to be more spaced, while those in the distance appear to be less spaced. If extend crop rows in the field image, they will meet at one point (vanishing point). According to this feature, crop planting row can be detected
具有计算速度快、能耐受一定杂草压力的优势,但要求种植行必须呈直线分布,且图像中种植行存在“消隐点”
It has advantage that can endure a certain weed pressure and has a fast calculation speed, it also has disadvantage that require planting rows must be linearly distributed and have vanishing point in the image
杂草压力不大且影像中作物行呈直线并存在“消隐点”的种植行提取;仅适于在田间农机拍摄影像中应用
It is suitable for planting row detection with low weed pressure and the rows in image must be linear distributed and have vanishing point; It only can be used with images taken from agricultural machinery