Study of camera calibration process with ray tracing

Abstract

Camera calibration is essential for any optical system used to obtain 3D measurements from images. The precision of the 3D depth estimation relies on an appropriate camera model and the accurate estimation of model parameters. These parameters are sensitive to environmental conditions and it is well established that a vision system should be calibrated in operating conditions. This is not always possible since the calibration process is often tedious and time-consuming. Unfortunately, the use of poorly estimated calibration parameters for 3D reconstruction and measurements may lead to suboptimal performance of the system and inaccurate depth estimation. This paper presents a technique using an existing camera model and optical design software to perform calibration simulations. This virtual calibration technique allows for a study of the impact of environmental conditions on the calibration parameters. Using this procedure, it is also possible to predict the statistical behavior of the calibration parameters considering the chosen fabrication processes and tolerances. It can assist vision scientists in the choice of the optical system that best meets the requested precision of the 3D reconstruction. This technique could eventually be integrated in the lens design process to create more reliable optical systems that could be calibrated and used in a range of environmental conditions with a very small variation of their calibration parameters.

Mitigating thermal‐induced image drift for videogrammetric technique in support of structural monitoring applications

“…Furthermore, the contraction/expansion of image may become notable in some situations, for example, in low temperature conditions. 19,22 Therefore, this study models the thermal-induced displacement field of the image plane via a first-order polynomial, which is sufficient for simulating the rigid-body motion and the uniform strain. Define a coordinate system that is parallel to the pixel coordinate system in which the x axis is leftward while the y axis is downward, and the upper-left corner of the image plane coincides with its origin initially.…”

“…It is an outcome attributed to the inclusion of the displacement components induced by the normal strain. Recalling that the contraction/expansion of image may become more significant in some situations, for example, low temperature conditions, 19,22 it is generally suggested to include the displacement components contributed by the normal strain in the first-order polynomial model. In this test, the temperature variation was realized with the aid of an air conditioner.…”

“…The thermal effect of videogrammetric technique has been reported in many previous studies. [15][16][17][18][19][20][21][22][23][24][25][26][27] Particularly, a few works have been devoted to exploring the source of the thermal effect. Among the camera parameters, the coordinates of principle point have been found to be prone to temperature variation.…”

“…Meanwhile, the focal length has also been shown to have a systematic relationship with temperature. 19,21,22,26 The variation of the focal length leads to inward/outward image drift (contraction/expansion of image). In addition, the camera-to-object distance shows temperature correlated variation as well.…”

“…It changes by some 0.1 pixel/ C. 19,21,22 It is also worth noting that the focal length is more sensitive to temperature variation at low temperature. 19,22 This helps explain why it has been taken into consideration mainly in aerial applications where the lowest temperature can be as low as À30 C, and the temperature difference before and after taking off can approach up to 70 C. Similarly, the variation of the camera-to-object distance is also insignificant. Particularly, the relative variation diminishes as the initial camera-to-object distance increases.…”