Journal of Engineering and Applied Sciences
ISSN: Online 1818-7803ISSN: Print 1816-949x
112
Views
1
Downloads
Abstract
Thanks to its simplicity and computational efficiency, Local Binary Pattern (LBP) has been widely utilized in texture classification. Traditional LBP codes the local difference. It also, uses the binary code histogram to model a given image. However, the directional statistical information is not taken into consideration in LBP. In this study, researchers present the Improved Circular Difference and Statistical Directional Patterns (ICDSDP). It is a new textual approach for texture classification accuracy. It is a combination of the circular difference of the directional information with oriented standard deviation. This approach aims at improving the texture classification. Experiments done on Outex and Curetgrey, large texture databases have shown that the application of the proposed texture feature extraction and classification approach can significantly ameliorate the classification accuracy of LBP. Compared to other methods, the proposed scheme could remarkably improve the classification accuracy. It could also, reduce classification.
How to cite this article:
Randa Boukhris Trabelsi, Alima Damak, Masmoudi and Dorra Sellami Masmoudi. Learning Improved Circular Difference and Statistical Directional Patterns for Texture Classifiaction.
DOI: https://doi.org/10.36478/jeasci.2014.147.152
URL: https://www.makhillpublications.co/view-article/1816-949x/jeasci.2014.147.152