Evaluation of the Robustness of Surface Characterisation of Carbon Fibre Composites Using Wavelet Texture Analysis

The mechanical properties of advanced composites are essential for their structural performance, but the surface finish on exterior composite panels is of critical importance for customer satisfaction. Previous work by the authors established the feasibility of wavelet texture analysis (WTA) for the task of automatically classifying the surface finish of carbon fibre reinforced polymer (CFRP) samples based on computer image processing. This paper presents an evaluation of the robustness of the WTA method to common process errors that can occur in the imaging of material samples. WTA creates a rich representation of the texture in an image that includes features related to both scale and orientation. Principal components analysis was used to reduce the dimensionality of the texture feature vector to a single principal component that could be used as the basis for discrimination between grades of CRFP sample surface finish quality. The results obtained indicate that the WTA method is robust to: significant horizontal and/or vertical translations of the sample being imaged; significant rotation of the sample being imaged; and significant dilation of the sample being imaged. The results obtained suggest that as long as reasonable precautions are taken in sample imaging, then the WTA method will yield repeatable results.

The Visual Quality Recognition of Nonwovens Using a Novel Wavelet Based Contourlet Transform

In this paper, a novel wavelet based contourlet transform for texture extraction is presented. In the texture analysis section, we propose a novel wavelet based contourlet transform, which can be considered as a simplified but more sufficient for texture analysis for nonwoven image compared with version of the one introduced by Eslami in theory view. In experiment, nonwoven images of five different visual quality grades, 125 of each grade, are decomposed using wavelet based contourlet transform with ‘PKVA’ filter as the default filter of Laplacian Pyramid (LP) and Directional Filter Bank (DFB) at 3 levels and two energy-based features, norm-1 L1 and norm-2 L2, are calculated from the wavelet coefficients at the first level and contourlet coefficients of each high frequency subband at different levels and directions to train and test SVM. When the nonwoven images are decomposed at 3 levels and 16 L1s are extracted, with 500 samples to train the SVM, the average recognition accuracy of test set is 98.4%, which is superior to the comparative method using wavelet texture analysis.