Ứng dụng các bộ lọc có hướng Gaussian trong việc phát hiện các khuyết tật trên bề mặt kết cấu

Nội dung text: Ứng dụng các bộ lọc có hướng Gaussian trong việc phát hiện các khuyết tật trên bề mặt kết cấu

1. 14 Hoàng Nhật Đức / Tạp chí Khoa học và Công nghệ Đại học Duy Tân 5(48) (2021) 14-18 5(48) (2021) 14-18 Applications of Gaussian Steerable Filters in detecting structural damages Ứng dụng các bộ lọc có hướng Gaussian trong việc phát hiện các khuyết tật trên bề mặt kết cấu Hoàng Nhật Đứca,b Hoang Nhat Duca,b* aInstitute of Research and Development, Duy Tan University, Da Nang, 550000, Vietnam aViện Nghiên cứu và Phát triển Công nghệ Cao, Đại học Duy Tân, Đà Nẵng, Việt Nam bFaculty of Civil Engineering, Duy Tan University, Da Nang, 550000, Vietnam bKhoa Xây dựng, Trường Đại học Duy Tân, Đà Nẵng, Việt Nam (Ngày nhận bài: 23/5/2021, ngày phản biện xong: 29/5/2021, ngày chấp nhận đăng: 12/10/2021) Abstract Periodic structural health survey is very crucial to guarantee the safety and serviceability of civil engineering structures. This study aims at developing a computer vision tool to detect defects on surface of civil engineering structures by means of Gaussian steerable filters. This tool has been developed with Visual C#. NET to facilitate its implementations. The developed software programs have been tested with images containing various defects such as crack, pothole, and spalling. Key words: Gaussian Steerable Filter; Structural health survey; Structural damage; Computer vision; Software development. Tóm tắt Khảo sát định kỳ là một nhiệm vụ quan trọng để đảm bảo sự an toàn và khả năng làm việc của kết cấu. Nghiên cứu của chúng tôi phát triển một công cụ thị giác máy tính để phát hiện các khuyết tật trên bề mặt kết cấu dân dụng sử dụng các bộ lọc có hướng Gaussian. Công cụ này đã được chúng tôi phát triển với ngôn ngữ Visual C# .NET và xây dựng thành phần mềm để tăng tính ứng dụng của công cụ. Chương trình phần mềm đã được kiểm chứng với các mẫu ảnh chứa các khuyết tật trên bề mặt kết cấu bao gồm vết nứt, hố trên đường, và vết lở trên tường bê tông. Từ khóa: Bộ Lọc Gabor, Khảo Sát Trạng Thái Kết Cấu; Hư Hỏng Kết Cấu; Thị Giác Máy Tính. 1. Introduction the data regarding structural health. Accurate The acceptable level of a structure’s and timely recognition of structural have serviceability is crucial to ensure the safety of become an integral part of the people. Accordingly, maintenance agencies building/infrastructure maintenance system. need to perform periodically survey and collect The reason is that early detection of structural *Corresponding author: Hoang Nhat Duc, Institute of Research and Development, Duy Tan University, Da Nang, 550000, Vietnam; Faculty of Civil Engineering, Duy Tan University, Da Nang, 550000, Vietnam; Email: hoangnhatduc@duytan.edu.vn
2. Hoàng Nhật Đức / Tạp chí Khoa học và Công nghệ Đại học Duy Tân 5(48) (2021) 14-18 15 defects can help to establish cost-effective has been successfully employed in other tasks rehabilitation methods and prevent reduction in of the computer vision field [26-31]. service life of various structures [1-3]. It is noted that in the GSF algorithm, a linear Crack is a widely encountered form of combination of Gaussian second derivatives is surface degradation for buildings and asphalt used as a basic filter. For an image I(x,y), a 2-D pavements [4-7]. The detection of crack in Gaussian at a certain pixel coordination is pavement is crucial for road maintenance. It is expressed as follows: because if cracks are recognized timely, the 1 (xy22 ) G( x , y , ) exp[ ] (1) required cost of maintenance can be saved up to 22 22  80% [8]. Besides crack, various forms of where  denotes a tunable parameter of the structural damages can be found such as pothole Gaussian function variance. for pavements and spall or bughole for concrete elements [9-14]. Thus, periodic structural health The first order derivatives used to compute o o survey is mandatory to detect these forms of the filters at 0 and 90 are given by [22, 32]: damages early to preclude accidents caused by G(,,)() x y x x22 y G exp[ ] (2) structural degradation [15-20]. 0 x 22 42  In recent years, computer vision has proved 22 to be a capable tool for automatic structural G(,,)() x y y x y (3) G90 42exp[ ] heath survey. The computer-based approach y 22   has significantly enhanced the productivity and A filter at an arbitrary orientation β is given objectiveness of the structural surveying by [22]: process. With such motivation, the current G cos( ) G sin( ) G (4) study aims at developing a software program  0 90 based on computer vision to analyze digital It is worth noticing that when the value of images and highlight various forms of the Gaussian function variance (r) is fixed, the structural damages. Gaussian steerable filters final filter response is a combined result of GSF are used to automatically analyze the image and with a set of orientation . The value of is detect edges representing damages on concrete selected from a set of angles such as 23 surface including spall, crack, bughole, and {0, , , }and pothole. The edges revealed by the Gaussian 4 4 4 2345 steerable filters can be subsequently employed {0, , , , , , }. 6 6 6 6 6 for further damage categorization and measurement. The final SF response at the pixel location of (x,y) in the image I is computed as follows: 2. Gaussian Steerable Filters for image processing R(x, y) F(x, y, ,)* I(x, y) The Gaussian Steerable Filter (GSF) [21, 22] (5) is essentially an image enhancement technique where ‘*’ is the convolution operator. that employs orientation-selective convolution 3. Applications of the newly developed kernels. As demonstrated in the previous works program [23-25], this image enhancement technique is particularly useful to differentiate the crack The performance of the newly developed patterns and the background texture of asphalt software program based on GSF used for pavement. In addition to crack detections, GSF structural damage detection is demonstrated in
3. 16 Hoàng Nhật Đức / Tạp chí Khoa học và Công nghệ Đại học Duy Tân 5(48) (2021) 14-18 four categories of defects. The image samples successfully highlighted the edges caused by have been collected by the Cannon EOS M10 damages in the surfaces of various structures: (CMOS 18.0 MP) and Nikon D5100. The size (a) concrete surface with crack, (b) asphalt of an image sample is 128x128 pixels. The pavement with crack, (c) wall with spalling, (d) images processed by GSF are demonstrated in asphalt pavement with a pothole, and (e) Fig. 1. As can be seen from the figure, the concrete column with a bughole. integrated computer vision model has (a) (b) (c) (d) (e) Fig. 1 Image analyses with images containing: (a) wall crack, (b) pavement crack, (c) spall in wall surface, (d) pavement pothole, and (e) concrete bughole
4. Hoàng Nhật Đức / Tạp chí Khoa học và Công nghệ Đại học Duy Tân 5(48) (2021) 14-18 17 4. Concluding Remarks pavements from RGB camera imagery and classification using circular Radon transform," Periodic survey of structural heath is Advanced Engineering Informatics, vol. 30, pp. 481- 499, 2016/08/01/ 2016. important in building or infrastructure [9] N.-D. Hoang, "Image Processing-Based Spall maintenance. To improve the productivity of Object Detection Using Gabor Filter, Texture this process, this work has developed a Analysis, and Adaptive Moment Estimation (Adam) computer vision model based on the Optimized Logistic Regression Models," Advances in Civil Engineering, vol. 2020, p. 8829715, applications of the GSF to detect structural 2020/11/30 2020. defects. Experiments show that the GSF based [10] N.-D. Hoang, Q.-L. Nguyen, and X.-L. Tran, approach can effectively identify edge features "Automatic Detection of Concrete Spalling Using Piecewise Linear Stochastic Gradient Descent caused by various types of distress. To facilitate Logistic Regression and Image Texture Analysis," the application of this model, a software Complexity, vol. 2019, p. 14, 2019. program has been constructed. The capability [11] N.-D. Hoang and Q.-L. Nguyen, "Metaheuristic of the newly developed program has been Optimized Edge Detection for Recognition of Concrete Wall Cracks: A Comparative Study on the tested with five cases including the detections Performances of Roberts, Prewitt, Canny, and Sobel of crack, spalling, pothole, and bughole objects. Algorithms," Advances in Civil Engineering, vol. 2018, p. 16, 2018. References [12] B. Liu and T. Yang, "Image analysis for detection of bugholes on concrete surface," Construction and [1] N.-D. Hoang and Q.-L. Nguyen, "A novel method Building Materials, vol. 137, pp. 432-440, for asphalt pavement crack classification based on 2017/04/15/ 2017. image processing and machine learning," Engineering with Computers, April 18 2018. [13] G. Yao, F. Wei, Y. Yang, and Y. Sun, "Deep- Learning-Based Bughole Detection for Concrete [2] N.-D. Hoang, "Classification of Asphalt Pavement Surface Image," Advances in Civil Engineering, vol. Cracks Using Laplacian Pyramid-Based Image Processing and a Hybrid Computational Approach," 2019, p. 12, 2019. Computational Intelligence and Neuroscience, vol. [14] N.-D. Hoang, "An Artificial Intelligence Method for 2018, p. 16, 2018. Asphalt Pavement Pothole Detection Using Least Squares Support Vector Machine and Neural [3] Y. Liu and J. K. W. Yeoh, "Automated crack pattern Network with Steerable Filter-Based Feature recognition from images for condition assessment of Extraction," Advances in Civil Engineering, pp. 1- concrete structures," Automation in Construction, 12, 2018. vol. 128, p. 103765, 2021/08/01/ 2021. [15] N.-D. Hoang, "Image processing based automatic [4] M. Gavilán, D. Balcones, O. Marcos, D. F. Llorca, recognition of asphalt pavement patch using a M. A. Sotelo, I. Parra, et al., "Adaptive Road Crack metaheuristic optimized machine learning Detection System by Pavement Classification," Sensors, vol. 11, p. 9628, 2011. approach," Advanced Engineering Informatics, vol. 40, pp. 110-120, 2019/04/01/ 2019. [5] N.-D. Hoang, "Detection of Surface Crack in Building Structures Using Image Processing [16] C. Koch, S. G. Paal, A. Rashidi, Z. Zhu, M. König, and I. Brilakis, "Achievements and Challenges in Technique with an Improved Otsu Method for Machine Vision-Based Inspection of Large Image Thresholding," Advances in Civil Concrete Structures," Advances in Structural Engineering, vol. 2018, p. 10, 2018. Engineering, vol. 17, pp. 303-318, 2014. [6] N.-D. Hoang, Q.-L. Nguyen, and V.-D. Tran, "Automatic recognition of asphalt pavement cracks [17] S. C. Radopoulou, I. Brilakis, K. Doycheva, and C. Koch, "A Framework for Automated Pavement using metaheuristic optimized edge detection algorithms and convolution neural network," Condition Monitoring," in Construction Research Congress 2016, ed, 2016. Automation in Construction, vol. 94, pp. 203-213, 2018/10/01/ 2018. [18] N.-D. Hoang and V.-D. Tran, "Image Processing- Based Detection of Pipe Corrosion Using Texture [7] Q. Chen, Y. Huang, H. Sun, and W. Huang, Analysis and Metaheuristic-Optimized Machine "Pavement crack detection using hessian structure Learning Approach," Computational Intelligence propagation," Advanced Engineering Informatics, and Neuroscience, vol. 2019, p. 13, 2019. vol. 49, p. 101303, 2021/08/01/ 2021. [19] Y.-S. Yang, C.-l. Wu, T. T. C. Hsu, H.-C. Yang, H.- [8] Y. O. Ouma and M. Hahn, "Wavelet-morphology J. Lu, and C.-C. Chang, "Image analysis method for based detection of incipient linear cracks in asphalt
5. 18 Hoàng Nhật Đức / Tạp chí Khoa học và Công nghệ Đại học Duy Tân 5(48) (2021) 14-18 crack distribution and width estimation for IEEE Transactions on Pattern Analysis and reinforced concrete structures," Automation in Machine Intelligence, vol. 26, pp. 1007-1019, 2004. Construction, vol. 91, pp. 120-132, 2018/07/01/ [27] S. Li, Y. Cao, and H. Cai, "Automatic Pavement- 2018. Crack Detection and Segmentation Based on [20] H. N. Nguyen, T. Y. Nguyen, and D. L. Pham, Steerable Matched Filtering and an Active Contour "Automatic Measurement of Concrete Crack Width Model," Journal of Computing in Civil Engineering, in 2D Multiple-phase Images for Building Safety vol. 31, p. 04017045, 2017. Evaluation," Cham, 2018, pp. 638-648. [28] J. Liang, X. Gu, and Y. Chen, "Fast and robust [21] W. T. Freeman and E. H. Adelson, "Steerable filters pavement crack distress segmentation utilizing for early vision, image analysis, and wavelet steerable filtering and local order energy," decomposition," in [1990] Proceedings Third Construction and Building Materials, vol. 262, p. International Conference on Computer Vision, 120084, 2020/11/30/ 2020. 1990, pp. 406-415. [29] H. Mahersia and K. Hamrouni, "Using multiple [22] W. T. Freeman and E. H. Adelson, "The design and steerable filters and Bayesian regularization for use of steerable filters," IEEE Transactions on facial expression recognition," Engineering Pattern Analysis and Machine Intelligence, vol. 13, Applications of Artificial Intelligence, vol. 38, pp. pp. 891-906, 1991. 190-202, 2015/02/01/ 2015. [23] H. Chen, H. Zhao, D. Han, and K. Liu, "Accurate [30] N.-D. Hoang, Q.-L. Nguyen, and D. T. Bui, "Image and robust crack detection using steerable evidence Processing-Based Classification of Asphalt filtering in electroluminescence images of solar Pavement Cracks Using Support Vector Machine cells," Optics and Lasers in Engineering, vol. 118, Optimized by Artificial Bee Colony," Journal of pp. 22-33, 2019/07/01/ 2019. Computing in Civil Engineering, vol. 32, p. [24] N.-D. Hoang, "Image Processing-Based Recognition 04018037, 2018. of Wall Defects Using Machine Learning [31] N.-D. Hoang and Q.-L. Nguyen, "Automatic Approaches and Steerable Filters," Computational Recognition of Asphalt Pavement Cracks Based on Intelligence and Neuroscience, vol. 2018, p. 18, Image Processing and Machine Learning 2018. Approaches: A Comparative Study on Classifier [25] N.-D. Hoang and Q.-L. Nguyen, "Fast Local Performance," Mathematical Problems in Laplacian-Based Steerable and Sobel Filters Engineering, vol. 2018, p. 16, 2018. Integrated with Adaptive Boosting Classification [32] TUM, "1D and 2D Gaussian Derivatives," Tree for Automatic Recognition of Asphalt Technische Universität München, Lehrstuhl für Pavement Cracks," Advances in Civil Engineering, Computer Aided Medical Procedures & Augmented vol. 2018, p. 17, 2018. Reality [26] M. Jacob and M. Unser, "Design of steerable filters , 2021.