Camera calibration based on an improved exponential trigonometric optimization algorithm

Camera calibration is essential in computer vision, yet achieving high accuracy with low-resolution cameras remains challenging. The traditional nonlinear optimization methods are computationally intensive and often suffer from large reprojection errors. This paper proposes a hybrid differential evolution-exponential trigonometric optimization (DETO) algorithm. DETO improves the differential evolution through adaptive parameter control and integrates the best individual information to enhance search efficiency. Initial parameters are obtained using Zhang's method, and a reprojection error model is constructed for optimization. Experimental results demonstrate that DETO achieves faster convergence, improved stability, and lower reprojection errors compared with conventional algorithms, reducing the average reprojection error to 0.07 pixels.