A Novel Bundle Adjustment Algorithm Based on Hybrid Multi-Objective Particles Swarm Optimization

Abstract

3D reconstruction has been developing during the last two decades, from moderate to medium and to large scale. It’s well known that bundle adjustment plays an important role in 3D reconstruction, mainly in Structure from Motion (SfM) and Simultaneously Localization and Mapping(SLAM). While bundle adjustment optimizes camera parameters and 3D points as a non-negligible final step, it suffers from memory and efficiency requirements in very large-scale reconstruction. Multi-objective optimization (MOO) is widely used for solving various engineering real-life problems. MOPSO is regarded as one of the states of the art for meta-heuristic MOO. MOPSO has adopted the concept of crowding distance as a measure that can leverage the characteristics of the distribution of solutions in the search space and provide a high level of exploration. However, this method is not sufficient to effectively explore the search space because it ignores the direction of the exploration. In addition, MOPSO starts the search from a fully randomly initialized swarm without taking any prior knowledge about the searching space into account which is considered impractical in applications where we can estimate initial values for solutions like bundle adjustment. In this paper, we introduced a  novel bundle adjustment algorithm based on hybrid MOPSO  that takes advantage of the traditional optimization algorithms like ADAM and other LSE solvers to improve the mobility of MOPSO solutions, the results showed that our algorithm can help to improve the accuracy and the efficiency in both memory and time of BA.