Reducing feature dimensionality for cloud image classification using local binary patterns descriptor

Reducing feature dimensionality for cloud image classification using local binary patterns descriptor

Cloud images
Feature ranking
Feature selection
Local binary patterns
Support vector machine

Clouds play a crucial role in precipitation and weather prediction. Identifying and differentiating clouds accurately poses a significant challenge. In this paper, we present a novel approach that utilizes the local binary patterns (LBP) feature descriptor to extract color cloud images. We employ feature fusion to combine LBP features from the independent channels of the RGB color space. Furthermore, we apply five well-known feature selection methods, namely ReliefF, Ilfs, correlation-based feature selection (CFS), Fisher, and Lasso, to select relevant and useful features. These selected features are then fed into a support vector machine (SVM) classifier. Experimental results demonstrate that our proposed approach achieves superior performance by significantly reducing the number of features while maintaining prediction accuracy.

Thongchai Surinwarangkoon1, Vinh Truong Hoang2, Kittikhun Meethongjan3

Journal homepage: http://telkomnika.uad.ac.id

Sep 6, 2024

PERI IRAWAN

PDF

ENGLISH

TEXT