IMPROVING TRAFFIC SIGN DETECTION BY COMBINING MSER AND LUCAS KANADE TRACKING

Soetedjo, Aryuanto (2019) IMPROVING TRAFFIC SIGN DETECTION BY COMBINING MSER AND LUCAS KANADE TRACKING. IMPROVING TRAFFIC SIGN DETECTION BY COMBINING MSER AND LUCAS KANADE TRACKING, 15 (2). pp. 1-13. ISSN 1349-4198 (Submitted)

[img] Text
Jurnal Aryu-2019-1 Improving Traffic.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (495kB)
[img] Text
Jurnal_Aryu-2019-1 Cek plagiat Improving Traffic.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (3MB)

Abstract

This paper presents the combination of Maximally Stable Extremal Region (MSER) and Lukas Kanade Tracking (LKT) for detecting traffic sign. The proposed ap- proach employs the MSER to find the red circular sign in an image. Once the traffic sign is detected, it is tracked by LKT to predict the position in the next frame. The detected sign and predicted position are treated as the traffic sign candidates. To reduce the false positive, the traffic sign candidates are validated using a template matching technique based on the Histogram of Oriented Gradient (HOG) feature. Instead of using MSER and LKT individually, our approach combines both methods to find the best candidate ac- cording to the matching score obtained by the validation stage. The experimental results show that our proposed method provides the efficient detection with the Recall of 0.9353, the Precision of 0.9747, the Accuracy of 0.9331, and the frame rate of 47.950 fps. Keywords: Traffic sign detection, MSER, Lukas Kanade tracking, HOG

Item Type: Article
Uncontrolled Keywords: Traffic sign detection, MSER, Lukas Kanade tracking, HOG
Subjects: Engineering > Electrical Engineering
Divisions: Fakultas Teknologi Industri > Teknik Elektro S1
Depositing User: Ms Nunuk Yuli
Date Deposited: 10 Feb 2021 03:20
Last Modified: 10 Feb 2021 03:20
URI: http://eprints.itn.ac.id/id/eprint/5310

Actions (login required)

View Item View Item