The ever-increasing requirements for reliability and quality of supply suggest to enable the self-healing features of modern distribution networks. Within the context of Active Distribution Networks (ADNs), new, fast and efficient restoration strategies can be developed relying not only on switching devices to realize load transfer operation, but also on the contribution of Distributed Generators (DGs) while controlling their set points. In this paper, a global optimization method is proposed for the restoration problem in a DG-integrated distribution network. The objective is to restore a maximum of loads with a minimum number of switching operations. In order to check the technical constraints of the resulting configuration (ex., line current and bus voltage constraints), a recently published method for exact convex formulation of the OPF problem is incorporated making the restoration problem robust particularly in the case of high nodal injections. The proposed restoration model is formulated as a mixed-integer second-order cone programming (MISOCP) problem. Two test cases are used to quantify the DG effects on the quality of the restoration strategy and to demonstrate the feasibility of the identified solution in cases where general relaxation methods for OPF problem leads to inexact solutions.