Issue34

Z. Jijun et alii, Frattura ed Integrità Strutturale, 34 (2015) 590-598; DOI: 10.3221/IGF-ESIS.34.65 594 Extracting of the wire rope strand based on canny To judge whether there is any defect in the wire rope after separating it from its background, we need to extract the slant texture of the wire rope strand. If the wire rope is undamaged, the texture will be complete and regular, based on which, we firstly adopt the homomorphic filtering technique to filter the image by a high pass filter to strengthen the edge information of the wire rope strand as in Fig. 7. The gray level change based on the reflected light image can strengthen the contrast of image. The process of grey level transformation is as below: ( , ) [ ( , )] g x y T f x y  It turns the gray level, ( , ) f x y , of each pixel, ( , ) x y , in the input image into the gray level, ( , ) g x y , in the output image. This article uses imadjust to transform the gray level, and the results are shown in Fig. 8. Figure 7: Sketch after high-pass filtering. Figure 8: Sketch after gray level transformation. After we get the function of contrast enhancement, we need to position the edge point accurately on the pixel point that transformed to a higher grey level to restrain false edge to the best extent and reduce peripheral point. So we use Canny operator [12] to conduct the extraction of wire rope strand. The procedure is as below: (1) Smooth the image with Gaussian filter. (2) Calculate the gradient magnitude and direction with first-order partial finite difference. (3) Restrain the gradient magnitude to the biggest extent. (4) Detect and connect edges with dual threshold algorithm. The effect is shown as Fig. 9. Angle adaptive integral transformation The projection feature of image obtained by integral projection [13-15] has translation invariance and can well express the distribution pattern of pixels in the space. ij F is the gray level of the image at (i,j). The formulas to solve the horizontal integral projection, i PY , of image in [ 1 , m x x ] and the vertical integral projection, j PY , of the image in [ 1 , m x x ], are as below: 1 1 2 1 , , Y ,..., Y m X i ij n i X PY F j Y m     1 1 2 1 , , ,..., n Y j ij m i Y PY F j X X X n    