用于自动驾驶系统的相机成像模型的构建

Construction of camera imaging model for vehicle automatic driving

不同的成像系统可以使用不同的模型进行更准确的描述,模型建立后,再通过标定过程计算出这些模型中的参数,最终得到对成像系统进行描述的相机内参与相机外参。模型建立的好坏,直接关系到标定计算后得到的内、外参数对相机成像系统描述的准确性。本文一改以往通过多项式来建立切变模型,而是利用旋转坐标系建立切变模型,完成了图像坐标到数字图像坐标的变化;根据成像步骤,构建了相机的成像模型。经企业实践验证,方案有效可靠。本文相机成像几何模型将用于自动驾驶系统后续图像处理。

In order to describing more accurately, different imaging systems use different models. After the model is established, the parameters of these models are calculated through the calibration process.Finally, the camera internal and external parameters which describe the imaging system are obtained.The quality of the model is directly related to the accuracy of the camera imaging system described by the internal and external parameters after calibration calculation. In this paper, instead of using polynomial to build shear model, we used rotating coordinate system to build shear model, and completed the change from image coordinate to digital image coordinate. Verified by enterprise practice,the plan is effective and reliable. The camera imaging geometry model of this paper will be used for subsequent image processing of autopilot.