Research on the spatial quantity distribution of fibers in ultra high performance concrete slabs based on image recognition method

Xueming Fan,Kun Zhao,Honglin Wu,Xiangdong Sun,Yuquan Ma,Jiabo Xu,Chunhua Chen,Yueqiang Tian
DOI: https://doi.org/10.1177/13694332241263867
2024-07-23
Advances in Structural Engineering
Abstract:Advances in Structural Engineering, Ahead of Print. In this paper, the detection of fiber distribution in six UHPC slabs with fiber content of 1%, 2%, 3%, and horizontal pouring and vertical pouring is studied by image recognition method. The results show that: firstly, the fiber distribution can be identified by a series of processing steps, such as grayscale and binarization of fiber source image, extraction of initial detection area, extraction of fiber contour and segmentation of fiber adhesion area. Furthermore, it is proposed that the scaling curve of n-e detection area per pixel of fiber number can anchor the effective recognition area range of fiber. Secondly, the statistical analysis of the identification results can visually draw the fiber distribution maps of different types of UHPC slabs in different directions, and quantitatively characterize the fiber distribution uniformity of UHPC slabs by quoting the evaluation index of fiber distribution coefficient. Finally, the cross-sectional fiber Scatter Density Plot can perceptually show the degree of fiber dispersion and its position distribution. The more uniform the color, the more uniform the fiber distribution in the cross-section of the specimen.
construction & building technology,engineering, civil
What problem does this paper attempt to address?