Abrupt changes in gravity data anomaly is edges. Edge detection methods in gravity data can be found in the basement of the location and boundaries of a mass crime. The purpose of this study, Improvement of uniqueness and avoiding premature convergence in inverse modeling of gravity data using boundary position information resources. To evaluate the proposed method, firstly gravimetric data modeling with irregular geometry stepped on a synthetic model were implemented without edge information. In the second phase edge information in inverse modeling gravimetric data were used artificial model. The results showed that using these constraints, we were able to limit the search space, As a result ant colony optimization algorithm for solving this problem increased and less time to reach a reliable conclusion. Finally the method on data Gotvand area were also implemented.