![](https://csdnimg.cn/release/download_crawler_static/88485529/bg2.jpg)
Abstract 硕士学位论文
II
Abstract
Crack is a common and important disease in building construction and it is of
great significance for structural inspection, identification and maintenance of
dangerous buildings to detect. With the help of digital image processing technology
and computer vision technology, the recognition algorithm, crack image calibration
method and crack parameter calculation method of building construction crack image
are researched, and the building of on-site crack measurement system is realized,
which can provide a lot of reference data for housing quality inspection and
evaluation. The main research contributions in this paper are as follows.
(1) The overall process of digital image processing technology applied to the
image recognition of cracks in buildings is proposed. Through the processing steps of
graying, median filtering, maximum entropy threshold segmentation,information
screening and crack connection on the collected crack images, the black-and-white
binary image containing only crack targets and background can be obtained.
(2) An information screening method based on binary image is proposed. On the
basis of the target segmentation, the morphological processing is used to preliminarily
screen the binary image of cracks, filter out small areas of noise and close small
cracks. Combined with the characteristics of connected domains such as the ratio
between long and short axes and area, the interference information is eliminated. A
crack fracture connection algorithm based on convex hull model in computational
geometry is proposed. When the distance between the fracture point is small, good
recovery results can be obtained by using a straight line fitting method.
(3) Because of the stable identification and positioning function of the
two-dimensional bar code, QR code is used as the crack observation point to arrange
on the building surface, so that the problem of crack numbering in images can be
resolved. A pixel calibration method based on two-dimensional code is proposed. By
using the position detection graphic feature of two-dimensional code, the calculation
of the conversion ratio is achieved under the pinhole imaging model, which can obtain
the conversion relationship between the pixel size and the physical size. Experiments
of QR code image calibration show that the calibration results are satisfactory when
the distance is appropriate and the camera plane is substantially parallel to the crack,
and recognition deficiencies in common targets are addressed.