An Approach to Enhance the Character Recognition Accuracy of Nepalese License Plates

March 2023

Domains

Optical Character Recognition Low-Resource Languages Computer Vision

Authors

Abstract

A methodology is proposed in this study to reduce character recognition flaws in the extracted License Plate (LP) region. To achieve this, the LP region must first be cleaned, and all noise that does not correspond to LP characters is filtered out using various image processing techniques. Later, the structure of LP characteristics will be used based on prior knowledge about the numbers of characters that are available in each vertical segment of the LP. Each positional LP character is determined either as a letter or a digit and will be expressed in terms of vector of letter, digit, or unknown. On the test dataset, the suggested technique for accurately segmenting LP characters had a 92% accuracy for all types of LP structures, whereas labeling letter-digit pairs had a minimum accuracy of 90.5% for 1-row LP. The proposed methodology can only be used with Nepalese LP; however, the approach can be adapted to work with LP datasets from other countries.

Keywords

AI Machine Learning Deep Learning

Cite this Publication!

Pankaj Raj Dawadi, Manish Pokharel, Bal Krishna Bal (2023). An Approach to Enhance the Character Recognition Accuracy of Nepalese License Plates.. n: Khanna, A., Polkowski, Z., Castillo, O. (eds) Proceedings of Data Analytics and Management . Lecture Notes in Networks and Systems, vol 572. Springer, Singapore. https://doi.org/10.1007/978-981-19-7615-5_67

Biblography