CN-121977411-A - Rapid quantification method and system for width of wood structure crack

CN121977411ACN 121977411 ACN121977411 ACN 121977411ACN-121977411-A

Abstract

The invention relates to the technical field of wood structure detection and computer vision, in particular to a method and a system for rapidly quantifying the width of a wood structure crack. The method comprises the steps of firstly placing a standard ruler on one side of a target crack, then collecting a shot image containing the standard ruler and the target crack, carrying out target detection and example segmentation on the standard ruler and the target crack on the shot image to obtain pixel sizes of the standard ruler and the target crack, calculating pixel proportions of the shot image based on the measured sizes of the standard ruler and the pixel sizes, including pixel area proportions, pixel length proportions and pixel width proportions, and calculating the actual area and equivalent length of the target crack by combining the pixel sizes and the pixel proportions of the target crack, wherein the actual area divided by the equivalent length is used as the detection width of the target crack. According to the invention, the characteristics of the wood cracks are fully considered, an iteration process during crack width recognition is avoided, the problems of complex extraction, poor robustness and the like of the traditional skeleton are solved, and the recognition efficiency is remarkably improved.

Inventors

WANG HANQIN
CHEN GANG
JIANG QING
ZHONG XUN
CHEN XIN
ZHU XINYU
TANG MING

Assignees

合肥工业大学

Dates

Publication Date: 20260505
Application Date: 20251223

Claims (10)

1. A rapid quantification method for the width of a wood structure crack is characterized in that a standard ruler is placed at one side of a target crack, and then a shooting image containing the standard ruler and the target crack is collected; Performing target detection and example segmentation on a standard ruler and a target crack on a shot image to obtain pixel sizes of the standard ruler and the target crack; calculating a pixel proportion of the photographed image based on the measured size and the pixel size of the gauge, including a pixel area proportion, a pixel length proportion, and a pixel width proportion; And calculating the actual area and the equivalent length of the target crack by combining the pixel size and the pixel proportion of the target crack, and taking the division of the actual area by the equivalent length as the detection width of the target crack.
2. The method for rapidly quantifying the width of a crack in a wood structure according to claim 1, wherein the pixel size comprises a pixel area, a pixel length, and a pixel width; Pixel area ratio = real area of standard scale/pixel area of standard scale Pixel length ratio = physical length of standard scale/pixel length of standard scale Pixel width ratio = physical width of gauge/pixel width of gauge; the actual area of the target crack is the product of its pixel area and the proportion of the pixel area.
3. The method for rapidly quantifying the width of a crack in a wood structure according to claim 1, wherein the equivalent length is the actual length of a diagonal line of a mark frame of a target crack on a photographed image or the actual length of the mark frame.
4. The method for rapidly quantifying the width of a wood structure crack according to claim 3, wherein the captured image is captured with a positive pose, wherein when the wood crack skeleton line is parallel to the axis of the wood member, the length of the marked frame is used as the equivalent length, and when the wood crack skeleton line is inclined to the axis of the wood member, the diagonal length of the marked frame is used as the equivalent length of the wood crack.
5. The method for rapidly quantifying the width of a wood structural crack according to any one of claims 1 to 4, wherein the training example segmentation algorithm performs object detection and example segmentation on a gauge and a target crack on a photographed image, and the training dataset is denoted as { photographed image, object position, category and segmentation mask }, the object including the gauge and the wood crack.
6. The method for rapidly quantifying the width of a wood structure crack according to claim 5, wherein the data set is obtained by searching for the wood structure crack on site in the wood structure, setting a standard ruler with a known actual size, performing on-site shooting to obtain an image sample containing the wood crack and the standard ruler, marking a target position, a category and a segmentation mask, and then expanding the image sample by data enhancement to form the data set.
7. The method of claim 6, wherein the image enhancement comprises one or more of random cropping, panning, rotating, changing brightness, and adding noise.
8. A rapid quantification system for the width of a crack in a wood structure, comprising: the camera module is used for acquiring a shooting image comprising a standard ruler and a wood crack; The example segmentation module is used for detecting and segmenting the shot image by adopting an example segmentation algorithm to obtain the pixel sizes of the standard ruler and the target crack, wherein the pixel sizes comprise pixel areas, pixel lengths and pixel widths; The proportion calculating module is used for obtaining the measuring size of the standard ruler and calculating the pixel proportion based on the measuring size of the standard ruler and the pixel size, wherein the pixel proportion comprises the pixel area proportion, the pixel length proportion and the pixel width proportion; And the crack calculation module is used for calculating the actual area and the equivalent length of the target crack by combining the pixel size and the pixel proportion of the target crack, dividing the actual area by the equivalent length to obtain the detection width of the target crack, and outputting the detection width.
9. A rapid wood structure crack width quantification device, comprising a memory and a processor, wherein the memory is stored with a computer program, the processor is connected with the memory, and the processor is used for executing the computer program to realize the rapid wood structure crack width quantification method according to any one of claims 1-7.
10. A storage medium, characterized in that a computer program is stored, which computer program, when executed, is adapted to carry out the method for rapid quantification of the width of a crack in a wood structure according to any one of claims 1-7.

Description

Rapid quantification method and system for width of wood structure crack Technical Field The invention relates to the technical field of wood structure detection and computer vision, in particular to a method and a system for rapidly quantifying the width of a wood structure crack. Background Cracks are widely distributed in infrastructures such as buildings, roads and bridges, are one of key indexes for evaluating the degradation degree of a structure, and have remarkable influence on bearing capacity of a component and an overall structure. Thus, timely and effective crack detection and accurate width quantification are critical to ensure structural safety and extend its useful life. The traditional crack detection technology mainly comprises manual detection which is time-consuming and labor-consuming, and a machine detection method such as infrared detection, ultrasonic detection or laser detection which is high in cost and complex in operation. With the development of artificial intelligence technology, computer vision-based deep learning technology has been widely applied to the field of engineering crack detection. However, the existing research results mainly aim at materials such as concrete, asphalt, rock and the like, and the crack forms of the materials are mostly irregular. Thus, these methods typically require complex iterative searches along the fracture skeleton line during the identification process to determine the maximum width of the fracture. For wood components, cracks mainly develop along the grain direction, the form is relatively regular, and crack skeleton lines usually represent straight lines or inclined straight lines along the axial direction of the component, and particularly, reference can be made to fig. 1 and 2. This morphological feature determines that the crack width does not change drastically in certain areas. In view of the above background, the conventional method for iteratively searching the maximum width based on the skeleton line is not efficient when applied to the wood structure crack, and is easily limited by the skeleton extraction precision. Disclosure of Invention In order to solve the problems of complex extraction of the traditional skeleton, poor robustness and the like in the quantification of the width of the wood structure crack in the prior art, the invention provides a rapid quantification method of the width of the wood structure crack, which fully considers the characteristics of the wood crack, avoids the iterative process when identifying the width of the crack, and obviously improves the identification efficiency. The invention provides a rapid quantification method for the width of a wood structure crack, which comprises the steps of firstly placing a standard ruler on one side of a target crack, and then collecting a shooting image containing the standard ruler and the target crack; Performing target detection and example segmentation on a standard ruler and a target crack on a shot image to obtain pixel sizes of the standard ruler and the target crack; calculating a pixel proportion of the photographed image based on the measured size and the pixel size of the gauge, including a pixel area proportion, a pixel length proportion, and a pixel width proportion; And calculating the actual area and the equivalent length of the target crack by combining the pixel size and the pixel proportion of the target crack, and taking the division of the actual area by the equivalent length as the detection width of the target crack. Preferably, the pixel size includes a pixel area, a pixel length, and a pixel width; Pixel area ratio = real area of standard scale/pixel area of standard scale Pixel length ratio = physical length of standard scale/pixel length of standard scale Pixel width ratio = physical width of gauge/pixel width of gauge; the actual area of the target crack is the product of its pixel area and the proportion of the pixel area. Preferably, the equivalent length is the actual length of the diagonal line of the marking frame of the target crack on the photographed image or the actual length of the marking frame. Preferably, the shooting image adopts posture correction shooting, when the wood crack skeleton line is parallel to the axis of the wood component, the length of the marking frame is used as the equivalent length, and when the wood crack skeleton line is inclined to the axis of the wood component, the diagonal length of the marking frame is used as the equivalent length of the wood crack. Preferably, the training instance segmentation algorithm performs object detection and instance segmentation on the standard ruler and the object crack on the shot image, and the training dataset is denoted as { shot image, object position, category and segmentation mask }, and the object comprises the standard ruler and the wood crack. The method for acquiring the data set comprises the steps of searching wood structure cracks on site of a wood struc