Indoor Pedestrian Location Based on Homography Transformation with Monocular Vision
LIN Lihua
ZHANG Hanjin
SHEN Shujian
LI Yuan
FU Wenxu
Abstract：At present,vision based pedestrian detection is mostly used for pedestrian recognition and classification,and less used for accurate pedestrian spatial positioning.In this paper,the homography transformation is introduced to solve the problem that once the camera angle changes,the external parameters of the camera need to be re-solved to achieve accurate pedestrian position-ing.This method is based on monocular vision.Pedestrian detection is realized by YOLOv5.The position of pedestrians in the world coordinate system is solved by coordinate transformation model.The camera angle is changed by combining the homography matrix.Pedestrian positioning is not required to re-solve the external parameters of the camera.The experimental results show that the maxi-mum positioning error of this method is less than 50 cm,and the average error is less than 25 cm,which meets the positioning accu-racy requirements.The average time consumed per frame is 35 ms,and meets the real-time requirements.
Keywords：homography transformationmonocular visionindoor pedestrian positioning
Publication Date：2025-07-20
Online Publishing Date：2025-09-25(First online date of this platform, not the publication date of the document)
Pages：7( 1885-1890,1939 )