An Improved Camera Intrinsic Parameters Calibration Method Based on Point and Line Features

Real-time and robust calibration of camera intrinsic parameters is a prerequisite for visual sensors based algorithms and systems to obtain high-quality positioning results and maps. Although traditional DLT methods meet the requirements of real-time performance, they are difficult to maintain accuracy in the case of input data corruption. To better meet the needs of high-accuracy and real-time performance at the same time, an improved DLT method based on the combination of point and line features was presented. The proposed method has three main contributions. First, uniformly filtered point features are adopted to reduce the impact of point features distribution on calibration accuracy. Second, the average coordinates of several 3D points corresponding to 2D line feature are used to reduce the error. Finally, by combining point and line features to construct constraints for calculating camera intrinsic parameters, we obtain better results than traditional methods. In order to verify the proposed method, we compared the accuracy of the calibration results calculated by the proposed method and the reference methods. And the experimental results demonstrated the effectiveness of the proposed method.