A Discussions on the Least-square Method in the Course of Error Theory and Data Processing

This paper introduces the principle of least squares briefly. We can get optimal results by least squares when conduct parameter estimation, in the case of the pending data does not contain gross errors. But the least squares method cannot avoid the influence of gross errors. When teachers teach this part, they only teach the least squares principle generally. They do not show students how to avoid the influence of gross errors. This paper discusses robust estimation method, when the pending data contain gross error and least squares method is not applicable, i.e. the least-square method combined with RANSAC algorithm. And take a linear parameter estimation problem as an example to illustrate.