Automobile location determination algorithm based on fundamental matrix and triangulation approaches

Abstract

GPS has certain limitations, such as difficulty functioning indoors due to weak signal transmission and obstruction by building structures. It also faces challenges in urban areas with tall buildings blocking the view between the receiver and satellite. Additionally, certain applications require a higher level of accuracy than what is typically provided by GPS receivers. In this paper, computer vision technologies were utilized to develop an algorithm that calculates the location of an automobile using only two images as input. The fundamental matrix and triangulation methods were employed to determine the location of the target automobile. An NVIDIA Jetson TX2 Development Kit camera was used to capture two consecutive images of the automobile. The location of the targeted automobile was then determined through the use of the normalized 8-point algorithm and the triangulation approach. This algorithm is effective in both indoor and outdoor environments.