Title:
Detection of AAR Deterioration Patterns in Concrete using Wavelets for Multiscale Texture Analysis
Author(s):
S. Kabir, P. Rivard, and G. Ballivy
Publication:
Symposium Paper
Volume:
234
Issue:
Appears on pages(s):
127-146
Keywords:
alkali-aggregate reaction (AAR); artificial neural networks (ANN); damage analysis; grey level co-occurence matrix (GLCM); image analysis; multi-resolution analysis (MRA); texture analysis; wavelets
DOI:
10.14359/15933
Date:
3/22/2006
Abstract:
Imaging methods for the quantification of cracking in concrete structures caused by alkali-aggregate reaction (AAR) are being employed increasingly due to the development of advanced Non-destructive testing techniques. However, more efficient image interpretation methods need to be developed in order to extract accurate information. This research proposes the application of an enhanced method of texture analysis on concrete imagery using the signal processing technique of Haar’s Wavelet Transform in combination with first-order histogram and second-order Grey Level Co-occurrence Matrix (GLCM) statistical approaches. This wavelet multiscale texture analysis technique provides increased discrimination of deterioration features in concrete images. For the detection and classification of the cracks, an Artificial Neural Network (ANN) method was used; the resulting classifications were used to extract surface information, such as the length and width of the cracks, as well as the total damage. This method was applied on grey scale images of outdoor exposed concrete blocks exhibiting various levels of AAR damage. The resulting levels of damage quantified through the image analysis approach correlate well with damage parameters obtained through in-situ data of the blocks, such as expansion measurements, impact-echo velocities, as well as laboratory data, such as mixture proportions.