Fitting planar points with two lines

There are various researches on linear regression and it has become a systematic and thus an important subject of statistics. However, almost all of the researches focused on regression with only one line, and only a few of the researchers are concerning about fitting the scatter points with more than one line, for example, two lines. The key point of this problem is that we have to find a method to generate two lines that at least locally minimize the sum of squared distances from the points to the nearest line. The method is given by fitting regression lines for two updating sets of points.