Abstract: Considering the deficiency of conventional Euler approach in magnetic detection at present, the combining influence of geomagnetism and adjacent bodies’ anomaly is assumed as linear variation. The non-linear problem of Euler equation has been figured out utilizing multivariate linear regression with unprescribed structural index. In quality control scheme, filtering technique of Euler solutions is presented based on the relationship between structural index and object depth. Adopting the distance acceptance criteria, the dispersed Euler solutions are isolated to different clusters firstly. The modifying stage combines the already focused solutions by the previous clustering in more general clusters and makes a fusion of the clusters whose horizontal centers of gravity have no remarkable difference using t-Student distribution test. Since the clusters with small number of points are statistically not significant, we apply a filter method that eliminates the clusters with less than a given number of points. Then, the mean values are corresponding to the position of magnetic objects. The effectiveness of the suggested techniques has been illustrated by simulated sphere/ cuboid examples and real magnetic data from a collection of environmental ferro-metallic objects. The conclusion shows that Euler approach, provided by relative threshold parameters, becomes a fully automatic interpretation approach, and the calculated horizontal position together with depth has very high precision. However, the noise influence to depth’s result is more serious than it to horizontal position.