A vehicle-mounted camera external parameter dynamic self-calibration method

Abstract

The invention discloses a vehicle-mounted camera external parameter dynamic self-calibration method, relates to the technical field of intelligent vehicle information processing. According to the method, a regular hexagon black and white calibration reference object is arranged on a carrier vehicle; Firstly, parameters of a monitoring camera on a static carrier vehicle are calibrated through a tensor calibration method, initial external parameters participated in the camera are obtained, and then the initial three-dimensional camera coordinates of all calibration reference points are calculated in combination with the three-dimensional world coordinates of all the calibration reference points in a reference object; And then the dynamically changed external parameters are regularly detectedand updated in the working process of the monitoring camera. Accuracy and reliability of external parameters of the monitoring camera are kept, and a calibration accuracy premise is provided for subsequent related machine vision application.

Description

technical field [0001] The invention relates to the technical field of intelligent vehicle information processing, in particular to a dynamic self-calibration method for external parameters of a vehicle-mounted camera based on the constraints of a specific geometric reference object. Background technique [0002] With the continuous advancement of my country's smart city pilot work process, the intelligent transportation system has become an inevitable choice for the development of the transportation industry. One of its key technologies is the research on vehicle safety and assisted driving systems. As an important part of the intelligent transportation system, the main purpose of research on intelligent vehicles is to reduce the incidence of increasingly serious traffic accidents and improve the efficiency of the existing road transportation system. Machine vision is one of the key sensing technologies in the related research of intelligent vehicles. The realization of aux...

AI Technical Summary

Problems solved by technology

Static calibration methods such as Zhang’s calibration method, RAC classic two-step method, etc. usually have the advantages of simple solution and high calibration accuracy, but the calibration process is relatively complicated, requiring high precision in the production of calibration objects, and the placement of calibration objects is relatively large. Limits its application occasions, it is difficult to apply to the calibration of the dynamic external parameters of the vehicle camera

Among the dynamic calibration methods, methods such as Caprile geometric parameter estimation method and three-orthogonal hidden point calibration method are mainly based on the information of parallel line hidden points and hidden line information in buildings or roads, which weakens the constraints on the scene. , but in the application, its algorithm robustness and calibration accuracy usually need to be further optimized; another type of dynamic calibration method, such as layered step-by-step method, Triggs quadratic surface calibration method, etc., is mainly based on the absolute quadratic surface between image sequences It does not need the constraints of the external scene, but its calculation process is usually complicated, which is not conducive to the wide application in the vehicle environment.

In the process of working with the vehicle, the existing vehicle-mounted surveillance camera has dynamic changes in the external parameters of the camera due to small deflection.

Images