openCv4Android之fitLine中各算法总结

在Imgproc类中有将2D点集或3D点集拟合成直线的方法 

public static void fitLine(Mat points,

           Mat line,
           int distType,
           double param,
           double reps,

           double aeps)

这类方法的本质都是（by minimizing sum_i rho(r_i) where r_i is a distance between the i^(th) point, the line and rho(r) is a distance function）就是使点集与线的误差和最小。其中rho(r_i)是距离函数，其类型有如下几种：

distType=CV_DIST_L2

rho(r) = r^2/2(the simplest and the fastest least-squares method

distType=CV_DIST_L1
rho(r) = r

distType=CV_DIST_L12
rho(r) = 2 * (sqrt(1 + frac(r^2)2) - 1)

distType=CV_DIST_FAIR
rho(r) = C^2 * ((r)/(C) - log((1 + (r)/(C)))) where C=1.3998

distType=CV_DIST_WELSCH
rho(r) = (C^2)/2 * (1 - exp((-((r)/(C))^2))) where C=2.9846

distType=CV_DIST_HUBER
rho(r) = r^2/2 if r < C; C * (r-C/2) otherwise where C=1.345

输出的Line有(vx, vy, x0, y0)或(vx, vy, vz, x0, y0, z0)，其中都是(vx, vy)或(vx, vy, vz）归一化的向量。

reps和aeps控制原点和西安之间的距离与角度的精度。

该方法的问题是未剔除掉奇异点。