Automatic Camera Calibration Method for Solving the Problem of Moving Object Detection

Abstract

Camera calibration is a key task in computer vision and plays an important role in many applications related to visual surveillance. A calibrated camera allows correcting perspective distortions of an object's image. In addition, it provides the ability to extract metric information from images that is robust to changes in viewing angle. Using calibrated cameras, it is possible to use pre-known 3D models to estimate the spatial position of objects, which increases the resilience of detection and tracking systems to noise and partial occlusions. This paper presents an automated method for restoring camera parameters from traffic video recordings. With only camera installation height data, the proposed algorithm allows fully determining both internal and external camera parameters based on the appearance and dynamics of objects in the video. The method has been tested under various conditions, and the results obtained confirm its effectiveness and practical applicability in moving object detection tasks.