Application of Robust Estimation in Sub-pixel Location

The least square estimation is widely used in parameter estimation. When the measurement error submits to Gaussian distribution, as a maximum likelihood estimate, the precision of the least squares estimation is high, otherwise, especially when the gross error exist, the precision of the estimation is lower. The robust estimation is a parameter estimation that some suitable estimation method is been used to relieve the impact of gross error when it is inevitable and find out the optimal parameter or the parameter nearby. The robust estimation algorithm is injected into sub-pixel locating. As a result, the impact of impulse noise could be reduced greatly and the precision of sub-pixel location could be higher.