Issue34

Z. Jijun et alii, Frattura ed Integrità Strutturale, 34 (2015) 590-598; DOI: 10.3221/IGF-ESIS.34.65 593 value of ( , ) f i j means the uneven brightness of the image, and bigger the interference of the background information. Therefore, we need to weigh in accordance with needs, and take weighting variables that are most suitable for effect of filtering. In this article, we take u as 0.6, and thus can get the ideal effect. Fig. 3 tells us that the uneven brightness and shadowing are eliminated after spatial homomorphic filtering and the marginal information is strengthened. Extraction of Wire rope Edge is the basic feature of the image. Generally speaking, at the location of edge, the gray level will be different from that of the adjacent area, consequently it will draw the outline of the target object [7-9]. This article uses Roberts operator to detect edge [10-11], and get the edge image. Roberts uses the performance of the partial difference operator to find out the operator of the edge. It sets ( , ) f i j as the input image and ( , ) g i j as output image. Roberts edge gradient is calculated by the formula below. 2 2 ( , ) {[ ( 1, 1) ( , )] [ ( , 1) ( 1, )] } g i j f i j f i j f i j f i j         Figure 3: Effect of holomorphic filtering. Figure 4: Roberts’s operator to calculate edge. It can be seen from the experiment that a certain quantity of noisy points have appeared during the edge detection of the image with Roberts operator, as shown in Fig. 4, which surely affecting the extraction of wire rope, conducting filtering is a necessity. It can be observed that the noise points are relatively separate and the quantity is small, resulting in a distinctive comparison with the wire rope itself at the vertical direction. This paper lists the statistical filtering to make statistics of the quantity of efficient points of each line of the image, and sets threshold value. Then the wire rope itself and the blending points of the background will be differentiated based on the results of statistics and conduct filtering, as shown in Fig. 5. In this way the extraction of the image of wire rope can be achieved, as shown in Fig. 6. Figure 5: Denoised edge image. Figure 6: A Sketch after background subtraction.