Automatic Speed-Limit Sign Detection and Recognition for Advanced Driver Assistance Systems
Anusha  Akula1, Renuka Devi S. M2

1Anusha Akula, G Narayanamma Institute of Technology and Science, Shaikpet, Hyderabad,Telangana, India.

2Dr. Renuka Devi S.M, G Narayanamma Institute of Technology and Science, Shaikpet, Hyderabad, Telangana, India 

Manuscript received on 09 August 2019 | Revised Manuscript received on 16 August 2019 | Manuscript Published on 31 August 2019 | PP: 1-5 | Volume-8 Issue-9S2 August 2019 | Retrieval Number: I10010789S219/19©BEIESP | DOI: 10.35940/ijitee.I1001.0789S219

Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (

Abstract: In recent years, traffic accidents have become the major cause to injuries, deaths and property damages. One of the main reasons to such accidents is due to high speed of vehicles. In order to maintain proper speed limit and thus provide significant contribution to improve safety, we propose Speed Limit sign detection and recognition method which is one of the features of Advanced Driver Assistance System (ADAS). In this paper we propose two approaches, i.e., histogram oriented gradient feature with SVM classifier namely HOG-SVM and CNN based approach. In these approaches we first pre-process the image using red color enhancement method and then we detect the Region of Interest using Maximally Stable Extremal Regions (MSER). Later, we classify the image by using different classifiers. In the HOG-SVM method, we are using HOG for feature extraction and Support Vector Machine (SVM) classifier for classification. In the CNN approach, we are using Convolutional Neural Networks (CNN) both for feature extraction and classification. Performance analysis of SVM classifier and CNN classifier are first evaluated on simple German Traffic Sign Recognition Benchmark (GTSRB) dataset using 5 fold classification, we got accuracy 100% for SVM classifier and 98.5% for CNN classifier. Also Further evaluated on German Traffic Sign Detection and Recognition Benchmark datasets and the experimental results show detection accuracy upto 93.6% for SVM classifier and 85.8% for CNN classifier. 

Keywords: Speed-Limit Sign, MSER, SVM, Histogram of Oriented Gradients, Convolutional Neural Network
Scope of the Article: Security Technology and Information Assurance