Determining the size distribution of concrete and asphalt aggregates using Gabor feature extraction and neural networks

Document Type : Research Paper

Authors

1 PhD Student of Mining Engineering, Mining Engineering Department, Shahid Bahonar University of Kerman, Kerman, Iran

2 Department of Mining Engineering, Shahid Bahonar University of Kerman,

3 Associate Professor of Mining Engineering, Department of Mining Engineering, Shahid Bahonar University of Kerman, Kerman, Iran

4 Professor of Electrical Engineering, Department of Electrical Engineering, Shahid Bahonar University of Kerman, Kerman, Iran

Abstract

Concrete and asphalt aggregates size distribution is one of the most important parameters in concrete and asphalt mix design that can affect the quality, durability, and strength of both concrete and asphalt. For evaluating the aggregates mix design, digital image analysis is a fast, reliable and indirect technique. In this study, based on one of the visual feature extractions methods (Gabor filters) and the neural networks, an algorithm was developed to determine the size distribution of digital images of concrete and asphalt aggregates. 100 images of concrete and asphalt aggregates were applied to train the neural network. Then, the results were compared with the results obtained by automatic estimation of aggregates size distribution by Split-Desktop software and sieving analysis. The results showed a general improvement in evaluating concrete and asphalt aggregates size distribution. Also, by using the proposed method, compare to automatic estimation of Split-Desktop, a reduction of 67% in error estimation was observed. Furthermore, this method showed also an improvement of 63% in evaluating of F10 to F100.

Keywords

Files

Share

How to cite

Statistics