Deep learning-based road surface condition detection using multi-sensor signal processing and evolutionary optimization
Abstract
Reliable recognition of road surface conditions is a key requirement for autonomous vehicles, advanced driver-assistance systems, and intelligent transportation infrastructure. This study introduces a unified classification framework that integrates multiscale texture analysis, transformer-based feature learning, statistical texture representation, and evolutionary feature selection. Initially, road images are processed through the stationary wavelet transform (SWT) to preserve texture information across multiple scales. A vision transformer (ViT) incorporating Squeeze-and-Excitation attention modules learns discriminative global visual patterns, while gray-level co-occurrence matrix (GLCM) features provide complementary local texture characteristics. The combined feature space is refined using mutual information (MI) and further optimized through enhanced grey wolf optimization (EGWO) to eliminate redundant information. Finally, a support vector machine (SVM) performs multi-class classification. Experimental evaluation demonstrates superior robustness and classification performance compared with recent deep learning (DL) and hybrid approaches, highlighting the effectiveness of the proposed framework for intelligent road surface analysis.
Keywords
Enhanced grey wolf optimization; Road surface classification; Stationary wavelet transform; Support vector machine; Vision transformer
Full Text:
PDFDOI: https://doi.org/10.11591/eei.v15i6.11956
Refbacks
- There are currently no refbacks.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Bulletin of Electrical Engineering and Informatics (BEEI)
ISSN: 2089-3191
,
e-ISSN: 2302-9285
This journal is published by the
Institute of Advanced Engineering and Science (IAES)
.