Classification Performance of Linear Binary Pattern and Histogram Oriented Features for Arabic Characters Images: A Review
| Main Authors: | Sungin Behram Khan, Gulzar Ahmad, Faheem Ali, Farooq Faisal,, Irfan Ahmed, Salman Elah |
|---|---|
| Format: | Article Journal |
| Terbitan: |
, 2018
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/1210109 |
Daftar Isi:
- There are millions of texts store in both off line and online forms. To utilize these documents properly, there is need of organizing these documents systematically and lots of applications are available for this purpose. Text classification is an important area of image processing deal with how the document belongs to its suitable class or category. Like other languages, Arabic language is also very rich and complex inflectional language which makes Arabic language very complex for ordinary analysis. In this review paper, we focus on the published research, especially in the field of Arabic text classification. Regard these all, three different types of feature extraction techniques are also implemented to extract features from different images of Arabic characters and presents a performance results of these techniques. From the result, it can be concluded that the combination of Linear binary pattern descriptor and Legendre moment, based moments features outperform and increase the accuracy of the LBP classifiers from 91.99 % to 93.12%