An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning

Wajahat Kazmi; Ian Nabney; George Vogiatzis; Peter Rose; Alex Codd

An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning

Wajahat Kazmi, Ian Nabney, George Vogiatzis, Peter Rose, Alex Codd

Research output: Contribution to journal › Article (Academic Journal) › peer-review

163 Downloads (Pure)

Abstract

This paper presents an industrial system to read
text on tyre sidewalls. Images of vehicle tyres in motion are
acquired using roadside cameras. Firstly, the tyre circularity is
detected using Circular Hough Transform (CHT) with dynamic
radius detection. The tyre is then unwarped into a rectangular
patch and a cascade of convolutional neural network (CNN)
classifiers is applied for text recognition. We introduce a novel
proposal generator for localizing the tyre code by combining
Histogram of Oriented Gradients (HOG) with a CNN. The
proposals are then filtered using a deep network. After the code
is localized, character detection and recognition are carried out
using two separate deep CNNs. The end-to-end system presents
impressive accuracy and efficiency proving its suitability for the
intended industrial application.

Original language	English
Pages (from-to)	1074-1079
Number of pages	6
Journal	IEEE Transactions on Intelligent Transportation Systems
Publication status	Published - 26 Aug 2019
Event	International Conference on Automation Science and Engineering - Vancouver, Canada Duration: 22 Aug 2019 → 26 Aug 2019 Conference number: 15 http://case2019.hust.edu.cn/

Access to Document

Full-text PDF (author’s accepted manuscript)
This is the author accepted manuscript (AAM). The final published version (version of record) is available online via Institute of Electrical and Electronics Engineers at [insert hyperlink] . Please refer to any applicable terms of use of the publisher.
Accepted author manuscript, 2.25 MB

Handle.net

Persistent link

Cite this

@article{9d20872e81e7489eb36a1f03c2028d55,

title = "An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning",

abstract = "This paper presents an industrial system to read text on tyre sidewalls. Images of vehicle tyres in motion are acquired using roadside cameras. Firstly, the tyre circularity is detected using Circular Hough Transform (CHT) with dynamic radius detection. The tyre is then unwarped into a rectangular patch and a cascade of convolutional neural network (CNN) classifiers is applied for text recognition. We introduce a novel proposal generator for localizing the tyre code by combining Histogram of Oriented Gradients (HOG) with a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The end-to-end system presents impressive accuracy and efficiency proving its suitability for the intended industrial application.",

author = "Wajahat Kazmi and Ian Nabney and George Vogiatzis and Peter Rose and Alex Codd",

year = "2019",

month = aug,

day = "26",

language = "English",

pages = "1074--1079",

journal = " IEEE Transactions on Intelligent Transportation Systems",

issn = "1524-9050",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

note = "International Conference on Automation Science and Engineering , CASE 2019 ; Conference date: 22-08-2019 Through 26-08-2019",

url = "http://case2019.hust.edu.cn/",

}

TY - JOUR

T1 - An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning

AU - Kazmi, Wajahat

AU - Nabney, Ian

AU - Vogiatzis, George

AU - Rose, Peter

AU - Codd, Alex

N1 - Conference code: 15

PY - 2019/8/26

Y1 - 2019/8/26

N2 - This paper presents an industrial system to read text on tyre sidewalls. Images of vehicle tyres in motion are acquired using roadside cameras. Firstly, the tyre circularity is detected using Circular Hough Transform (CHT) with dynamic radius detection. The tyre is then unwarped into a rectangular patch and a cascade of convolutional neural network (CNN) classifiers is applied for text recognition. We introduce a novel proposal generator for localizing the tyre code by combining Histogram of Oriented Gradients (HOG) with a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The end-to-end system presents impressive accuracy and efficiency proving its suitability for the intended industrial application.

AB - This paper presents an industrial system to read text on tyre sidewalls. Images of vehicle tyres in motion are acquired using roadside cameras. Firstly, the tyre circularity is detected using Circular Hough Transform (CHT) with dynamic radius detection. The tyre is then unwarped into a rectangular patch and a cascade of convolutional neural network (CNN) classifiers is applied for text recognition. We introduce a novel proposal generator for localizing the tyre code by combining Histogram of Oriented Gradients (HOG) with a CNN. The proposals are then filtered using a deep network. After the code is localized, character detection and recognition are carried out using two separate deep CNNs. The end-to-end system presents impressive accuracy and efficiency proving its suitability for the intended industrial application.

M3 - Article (Academic Journal)

SN - 1524-9050

SP - 1074

EP - 1079

JO - IEEE Transactions on Intelligent Transportation Systems

JF - IEEE Transactions on Intelligent Transportation Systems

T2 - International Conference on Automation Science and Engineering

Y2 - 22 August 2019 through 26 August 2019

ER -

An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning

Abstract

Access to Document

Handle.net

Fingerprint

Cite this