This paper introduces a robust and computationally efficient semi-analytical approach to design regional coverage satellite constellations. By fully utilizing the characteristics of the repeating ground track orbits and multiple access intervals between the target and the satellite, the proposed methods aim to optimally design satellite constellation with the fewest number of satellites possible. Two methods are constructed by applying the circular convolution theorem based on the assumption that the seed satellite access profile and the satellite position vector can be treated as discrete signals. An analysis shows that the While-Loop method is computationally efficient while the Integer Programming method is the most optimal. Various illustrative examples are performed to demonstrate the value provided by the proposed approach.