Improving camera parameter estimation using an adaptive genetic algorithm

In this paper, we propose an adaptive genetic algorithm designed to address the camera calibration problem. This approach facilitates the resolution of a complex optimization challenge. Our objective is to refine the camera calibration results estimated by the analytical method. For this purpose, a study was conducted on the type and probability of crossover, the probability of mutation and on the adaptation of the initialization intervals. This adaptation consists of adjusting the length of the initialization intervals. The main objective is to find an optimal solution for the camera calibration parameters by minimizing the cost function. This function is reformulated from the relationship between the points of the 3D target and their 2D projection in the image. Experimental tests and evaluations were conducted to validate the proposed approach. The results indicate that our algorithm is robust and can achieve very satisfactory calibration results.