مقاله انگلیسی رایگان در مورد روش آنالیز تصویر برای توزیع شکاف سازه های بتن – الزویر 2018

Crack observation is an important aspect of most reinforced concrete (RC) structural experiments and structural safety evaluation. The crack patterns, angles, and distribution density may reveal the failure modes, damage levels, and stiffness degradation for concrete models [1,2]. The effect of cracks on structural strength or water permeability has been discussed in many studies [3,4]. The reasonable limits of crack widths and repair methods have been recommended by standard codes or regulations [5,6]. In addition, for the containment vessel of a nuclear reactor, cracks indicate the risk of radiation leak [7]; therefore, regulations have been established [8]. While many numerical models of concrete materials adopt the concept of smeared cracks to estimate crack-induced strength degradation (e.g., [9]), some methods involve simulation of concrete cracks and numerical analysis of crack widths for better prediction of crack-induced structural behaviors [10,11]. Since the advancement of digital image technology, image analysis methods have been utilized for crack detection as they provide more advantages and feasibility for structural health monitoring (SHM) applications. Image analysis offers a cost-effective, alternative solution to concrete crack observation and has potential in SHM applications. Using image analysis, we can not only record the overall visual appearance of an RC surface but also analyze vibrations [12], deformation [13,14], and terrain models [15] as well as assess construction quality such as welding quality [16] or loosened bolts [17]. Structural deformations, mode shapes, natural frequencies, and motion magnification can also be estimated using image analysis [18]. Image analysis has been also employed to detect cracks. Yu et al. [19] analyzed infrared images to detect tunnel lining surface cracks. Hutchinson and Chen [20] conducted image analysis to evaluate concrete damage of bridges induced by cracks and spalling. Zakeri et al. [21] developed an approach to interpret and classify pavement cracks. Chen et al. [22] recognized cracks through analyzing hundreds of photos in a bridge management database. Li et al. [23] recognized bridge cracks through image edge detection and noise reduction. Dinh et al. [24] proposed a method to extract concrete cracks based on the image gray-scale histogram. Machine-learning-based computer vision has been applied and trained to detect concrete cracks under various environmental conditions, e.g., Prasanna et al. [25] and Cha et al. [26]; it is capable of detecting a wide variety of concrete surface defects as well as reducing the effects of uncontrollable ambient lights, provided that sufficient training data are available. Most of the existing crack detection methods are based on edge detection, which extracts the dark shadow lines or crack regions.