Suppression of the negative effect of abnormal data based on the Hough transform and application to the magnetic compensation of airborne optically-pumped magnetometer data.

Abstract

A magnetic sensor of the optically pumped magnetometer may enter a dead zone during an aeromagnetic survey, resulting in unavoidable abnormal data and seriously affecting the effect of aeromagnetic compensation. We propose a fast discrimination and culling method based on the Hough transform to prevent abnormal data from participating in the coefficient estimation. In the proposed method, the parameter space is partitioned into small buckets and the most frequently passed region of sinusoidal curves is detected to eliminate abnormal data. Although the conventional method performs similarly to the proposed method when there is only 1% abnormal data, it is theoretically shown that the proposed method has a better goodness of fit of 0.9518, compared with a value of 0.1956 for the conventional method, in the presence of 45% abnormal data. Furthermore, we construct an experimental platform and conduct a flight test in which the proposed method has an improvement ratio of 4.11 compared with a value of 0.34 for the conventional method.