Comparative analysis of RADAR-IR sensor fusion methods for object detection

This paper presents the Radar and IR sensor fusion method for objection detection. The infrared camera parameter calibration with Levenberg-Marquardt (LM) optimization method is proposed based on the Radar ranging data represented by Cartesian coordinate compared with 6 fusion methods. The proposed...

Full description

Saved in:
Bibliographic Details
Published in2017 17th International Conference on Control, Automation and Systems (ICCAS) pp. 1576 - 1580
Main Authors Taehwan Kim, Sungho Kim, Eunryung Lee, Miryong Park
Format Conference Proceeding
LanguageEnglish
Published Institute of Control, Robotics and Systems - ICROS 01.10.2017
Subjects
Calibration
Cameras
Charge coupled devices
Extrinsic parameter
Image resolution
Infrared
Intrinsic parameter
Optimization
Radar
Radar cross-sections
Radar imaging
Online AccessGet full text

Cover

More Information
Summary:This paper presents the Radar and IR sensor fusion method for objection detection. The infrared camera parameter calibration with Levenberg-Marquardt (LM) optimization method is proposed based on the Radar ranging data represented by Cartesian coordinate compared with 6 fusion methods. The proposed method firstly performs the estimation of the intrinsic parameter matrix of infrared camera with some optical trick. Then the method searches the extrinsic parameters using the generative approach. The initial angle and translation of the extrinsic parameters are optimized by the LM method with the geometrical cost function. In the experiments, the performance of proposed method outperforms by a maximum 13 times the performance of the other baseline methods on the averaged Euclidian distance error. In future work, the angular noise of the Radar information will be improved and the proposed method will provide the effective proposals for the deep neural network.
DOI:10.23919/ICCAS.2017.8204237