In previous work, we studied the effects of a supersonic shockwave on the performance of a Shack-Hartmann wavefront sensor (SHWFS). A shockwave is a sudden change in density which then cause a sharp gradient in the wavefront. This results in dots distorting in the SHWFS dot pattern which in turn affects the accuracy of wavefront reconstruction. We used higher order statistics, such as standard deviation, skewness and kurtosis to identify and eliminate these distorted dots in Shack-Hartmann wavefront sensors, improving the accuracy of the final reconstructed wavefront. However, this is contingent on accurate calculations of the centroid, standard deviation, skewness, and kurtosis of these dots. In the first part of this work, we look at the effect of the finite pixel size on the accuracy of the Shack-Hartmann wavefront sensor, specifically on the calculation of the centroid and higher order statistics of a pixelated dot. We simulate a dot and average the intensity distribution for different pixel sizes. We show that the error is significant if the ratio of the center Dot size/pixel size is less than 2. Past this ratio of 2, the centroid and statistics error of the dots drop to near zero. We also explain some striping artifacts observed in the statistics maps in experimental work. In the second part of this paper, we investigate the centroid error of a dot approaching the edge of the Area of Interest (AOI). In cases of large tilts in the wavefronts, the dot can be pushed into or past the edge of the AOI. We show that a large AOI size/Dot size ratio allows for the most tilt and range of movement. However, by squaring or cubing the intensity distribution, the error can be further reduced for a given system.
|