Image processing and artificial neural network based determination of surface mean texture depth on lab-controlled...

Area of Science:

• Civil Engineering
• Materials Science
• Transportation Engineering

Background:

• Pavement surface texture is critical for road safety and noise levels.
• Existing texture evaluation methods often require lane closures or are costly.
• Routine evaluation of pavement surface texture changes is essential.

Purpose of the Study:

• To develop a 2D image processing method for estimating the mean texture depth (MTD) of chip-sealed pavements.
• To assess the reliability of a surface void ratio (SVR) approach for pavement texture analysis.
• To compare the proposed method with conventional techniques like Sand Patch and Hydrotimer.

Main Methods:

• Laboratory preparation of chip-sealed pavement samples with varying aggregate properties.
• Determination of MTD using conventional Sand Patch (SP) and Hydrotimer (HT) methods.
• Application of a 2D image processing technique based on surface void ratio (SVR) analysis using captured images.
• Correlation analysis between SVR data and conventional MTD measurements using artificial neural networks.

Main Results:

• The 2D image processing method successfully estimated SVR for all pavement samples.
• A strong correlation was found between the SVR-derived data and MTD values obtained from SP and HT methods.
• The artificial neural network analysis confirmed the relationship between the proposed method and conventional techniques.

Conclusions:

• The proposed surface void ratio (SVR) approach, utilizing 2D image processing, is a reliable alternative for evaluating chip-sealed pavement surface texture.
• This method has the potential to overcome limitations of existing techniques, offering a more accessible evaluation.
• Further research could explore its application to other pavement types and conditions.