【摘要】 利用霍夫变换抗噪声能力强和能分离出属于不同直线附近点的特性,研究离散数据点集M中xi,yi满足幂函数关系时的曲线检测问题。首先,对离散数据点集M中的数据xi,yi作对数变换,得到新的数据点集M(xi,yi),此时,xi,yi具有线性关系;其次,用霍夫变换检测M中的直线,可得直线参数;然后,利用霍夫变换所得的直线参数,计算图象中的边缘点到直线的距离dki,并与给定阈值dk比较,从而将分布在不同直线附近的点分离出来,同时剔除数据点集M中的干扰点或噪声;最后,用最小二乘法拟合直线,得到剔除干扰点或噪声后的拟合曲线方程参数a和b。更多还原

【Abstract】 Hough transform is always used to detect line.In this paper,a new application of Hough transform is pro-posed,which is used to extract curve from many discrete points.The approach can detect curve from the points x i ,y i fit power function.The algorithm is as follows :Firstly,the new data points M  (x i ,y i )are gotten through logarithmic transform-ing to the discrete data points,the relation between x i ,y i is linear at that time ;Secondly,line is detected by using Hough transform and parameters b,cof lines are also gotten;Sequence,the distance d ki between points of edge and the lines is calculated by using parameters b,c,by comparing threshold d k with d ki ,points in the vicinity of different lines are extracted and interference points and noise of data points are deleted too;Lastly,line is fitted by using a method of least square and accurate parameters a and b of lines are obtained.The algorithm is insensitive to noise be-cause Hough transform can resistant noise and separate points from the vicinity of different lines.更多还原