Camera Calibration with Baseline Distance Constraints

Abstract

Abstract This paper presents a distance‐constrained bundle adjustment approach which was used to calibrate a multi‐sensor system. The multi‐sensor system integrates three rigidly fixed cameras, two visible and one thermal infrared. The influence of the number of camera base constraints is empirically analysed through the process of self‐calibration bundle adjustment. The successful calibration of the multi‐sensor system requires the incorporation of a full set of baseline distance constraints into the adjustment to obtain satisfactory computations of the camera bases and relative rotations between cameras. The improvement in the precision of the calibration is outstanding; its maximum is achieved with the low‐resolution thermal camera. This constrained solution is recommended for multi‐sensor systems that integrate rigidly fixed cameras.