**5. Conclusions**

In this paper, we propose a methodology for automatic vehicle license plate detection and recognition. This process consists of the following steps: image acquisition, license plate extraction, character extraction, and recognition. We demonstrated through experiments on synthetic and real license plate data that the proposed system is not only highly accurate but is also efficient. We also compared this method to similar existing methods and showed that it achieved a balance between accuracy and efficiency, and as such is suitable for real-time detection of license plates. A limitation of this work is that our method was only tested on a database of Malaysian license plate images captured by the researchers and the publicly available Medialab LPR and UFPR-ALPR databases. In the future, we will explore how it performs on other license plate databases. We will also investigate the use of multi-stage deep learning architectures (detection and recognition) in this domain. Furthermore, since we proposed barrier access control for the controlled environment, the current image acquisition system was set up so that only one vehicle was visible in the field of view. As a result, the image only captured one vehicle per frame, simplifying the process of license plate detection. In future work, we will extend our methodology to identify license plates using more complex images containing multiple vehicles. In addition, we will increase the size of the training database in an effort to minimise misclassification of similar classes.

**Author Contributions:** Conceptualization, K.T.I.; Data curation, K.T.I. and R.G.R.; Formal analysis, K.T.I., R.G.R., S.M.S.I. and S.W.; Funding acquisition, R.G.R., S.M.S.I., S.W. and S.O.; Investigation, K.T.I., S.M.S.I., S.W., M.S.H. and T.R.; Methodology, K.T.I., S.M.S.I. and S.W.; Project administration, R.G.R., S.M.S.I., S.W. and S.O.; Resources, R.G.R., S.M.S.I., S.W. and S.O.; Software, K.T.I., S.M.S.I. and S.O.; Supervision, R.G.R., S.M.S.I., S.W. and S.O.; Validation, K.T.I., R.G.R., S.M.S.I., S.W., M.S.H. and T.R.; Visualization, K.T.I., S.M.S.I., M.S.H. and T.R.; and Writing—original draft, K.T.I.; Writing—review and editing, K.T.I., R.G.R., S.M.S.I., S.W., M.S.H., T.R. and S.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This Research was funded by The University of Malaya Grant Number-BKS082-2017, IIRG012C-2019 and UMRG RP059C 17SBS, Melbourne Research Scholarship (MRS), and Edith Cowan University School of Science Collaborative Research Grant Scheme 2019.

**Acknowledgments:** The authors would like to thank Md Yeasir Arafat of the Department of Electrical Engineering, Faculty of Engineering, University of Malaya, Kuala Lumpur, Malaysia, for his input on image pre-processing techniques.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Data Availability:** The datasets used in this study are available from the corresponding author upon request.

*Sensors* **2020**, *20*, 3578
