A modified triangle star pattern recognition algorithm

The traditional triangle star identification algorithm has a higher redundancy and a poor robustness to noise, a modified triangle algorithm based on traditional one is proposed. By the proposed algorithm, a planar normal vector formed by three angles of the observed triangle is projected to the principal axis to get a projection value. Some navigation triangles within some scope are selected by using the projection value, and the three angles between the observed triangle and the navigation triangles are compared, if exists redundancy, a normalized magnitude is used to eliminate it. Compared to the traditional triangle approach, the modified algorithm has a couple of advantages, including higher rate of correct star pattern recognition, lower redundancy and better robustness. The modified algorithm has the identification rate of 97.7% when the position error is 0.5 pixels and the noise of star magnitude is 0.5Mv, while the traditional algorithm is 91.4% when the position error is 0.5 pixels.