Abstract
Abstract. Visual perception is regularly used by humans and robots for navigation. By either implicitly or explicitly mapping the environment, ego-motion can be determined and a path of actions can be planned. The process of mapping and navigation are delicately intertwined; therefore, improving one can often lead to an improvement of the other. Both processes are sensitive to the interior orientation parameters of the camera system and mathematically modelling these systematic errors can often improve the precision and accuracy of the overall solution. This paper presents an automatic camera calibration method suitable for any lens, without having prior knowledge about the sensor. Statistical inference is performed to map the environment and localize the camera simultaneously. K-nearest neighbour regression is used to model the geometric distortions of the images. A normal-angle lens Nikon camera and wide-angle lens GoPro camera were calibrated using the proposed method, as well as the conventional bundle adjustment with self-calibration method (for comparison). Results showed that the mapping error was reduced from an average of 14.9 mm to 1.2 mm (i.e. a 92 % improvement) and 66.6 mm to 1.5 mm (i.e. a 98 % improvement) using the proposed method for the Nikon and GoPro cameras, respectively. In contrast, the conventional approach achieved an average 3D error of 0.9 mm (i.e. 94 % improvement) and 6 mm (i.e. 91 % improvement) for the Nikon and GoPro cameras, respectively. Thus, the proposed method performs more consistently, irrespective of the lens/sensor used: it yields results that are comparable to the conventional approach for normal-angle lens cameras, and it has the additional benefit of improving calibration results for wide-angle lens cameras.
Highlights
An average human eye has a vertical field-of-view (FOV) of approximately 135 degrees (°) and a horizontal FOV of 160°
While wide-angle lens cameras are beneficial in robot vision, the larger FOV introduces some new challenges that need to be addressed before these cameras can be used for ego-motion estimation and structure from motion
A calibration room with targets arranged in concentric circles was used and ground truth target locations were required since single photo resection was used rather than bundle adjustment
Summary
An average human eye has a vertical field-of-view (FOV) of approximately 135 degrees (°) and a horizontal FOV of 160°. With binocular field the horizontal FOV extends to 200° and beyond This implicitly assists in our day-to-day human activities such as navigation, path-planning, object recognition and tracking. Such a wide visual field is beneficial for survival in the animal kingdom because it allows more information to be gathered and analysed from a single viewpoint without exerting energy to turn our heads. While wide-angle lens cameras are beneficial in robot vision, the larger FOV introduces some new challenges that need to be addressed before these cameras can be used for ego-motion estimation and structure from motion. A typical 35 mm camera with a 50 mm focal length lens yields vertical and horizontal FOVs of approximately 27° and 40°, respectively. No expert knowledge is necessary since machine learning approaches are used to automatically adapt the tuning hyperparameters and decide on the model complexity based on the dataset
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
