Dynamic features based stroke recognition system for signboard images of Gurmukhi text
- PDF / 2,949,663 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 58 Downloads / 149 Views
Dynamic features based stroke recognition system for signboard images of Gurmukhi text Jasleen Kaur Bains 1
2
& Sukhdeep Singh & Anuj Sharma
1
Received: 10 September 2019 / Revised: 24 June 2020 / Accepted: 18 August 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The computation of correct features is an essential phase for efficient data representation and benchmarked accuracy in text recognition systems. The offline text lacks dynamic information regarding the writing order or nature of trajectories of stroke. Recovery of drawing order technique helps to retrieve trajectory of a stroke. This information aids in computing dynamic feature vector based on chain codes or trajectory points for text recognition. The present work proposes a dynamic feature extraction approach based on recovery of drawing order to understand scene text in Indic script Gurmukhi. An inhouse dataset of strokes was obtained from 820 real time Gurmukhi signboard images. Stroke recognition was performed using Conv1D, SVM and HMM classifiers. Best recognition results were achieved using SVM and Conv1D as 82.88% and 84.67%. The major objective of present study is to propose dynamic features based recognition scheme for Indic scripts signboard images suitable for real-life applications. Keywords Drawing order . Gurmukhi signboard . Chain codes . Trajectory point . Offline text recognition . Deep learning
1 Introduction The state of art in text recognition has achieved praise worthy milestones in last few decades in the field of pattern recognition. It is categorized into online and offline subject to the kind of
* Jasleen Kaur Bains [email protected] Sukhdeep Singh [email protected] Anuj Sharma [email protected]
1
Department of Computer Science and Applications, Panjab University, Chandigarh, India
2
D. M. College, Moga, Punjab, India
Multimedia Tools and Applications
data that is dealt in recognition. The online text recognition techniques are used for recognizing handwritten text where data is captured using stylus on a tablet or personal digital assistant. Offline text recognition techniques use data which is either scanned or captured using a camera, mainly from scanned handwritten or printed documents, signboard images containing text or any image which contains text written on any kind of surface like book cover or cargo boxes. The scene text recognition has gained enormous interest of the research community in past decades due to the vital information it can provide such as names of places, shops and organizations. It has led to the development of various vision based applications such as travel assistant, map navigation, traffic surveillance, aid for helping blind people [5, 19]. These applications provide the end user with translated version of the text contained in images in textual or speech form [5, 19, 34, 35]. The signboard text recognition is one of the promising areas of research in scene text recognition as it can help to develop useful real life applications. The signboards carry ve
Data Loading...
