Improvement of Stoilov algorithm based on adaptive mean filtering

In the process of using the mean filtering algorithm to repair the singular points in Stoilov algorithm that depend on the acquired image light intensity, the details such as the edge of the phase shift will be lost, resulting in the calculated phase errors. To solve this problem, the mean filtering algorithm is improved. Firstly, the size of the filtering window is adjusted adaptively according to the number of singular points in the statistical window, then the weights of each non-singular point in the filtering window are calculated adaptively, and finally the singular points are repaired by weighted mean. This method can not only remove the singularity effectively, but also keep the phase distribution details, and restrain the phase error introduced by the singularity well. Experimental results show that the precision of phase error can be increased by more than 2 times.