According to anomaly characteristics and with the help of Mapgis software, the measuring points affected relatively insignificantly by the local anomaly in measured data were chosen through sieving, and the curve or curved surface represented by a n-order polynomial expression was used to approximate the regional anomaly contained in these measuring points. The coefficient of this polynomial expression was used to calculate the polynomial value of every measuring point in the whole surveyed area, and the value was used as the regional anomaly of the measuring point. On such a basis, the local anomaly can also be obtained.