Method for identifying steel bar laying structure through construction photo

Abstract

The invention relates to a method for identifying a reinforcing steel bar laying structure through a construction photograph, and the method comprises the following steps: setting reinforcing steel bar intersection type labeling data, and enabling a reinforcing steel bar picture of the labeling data to comprise three labels: intersection points at reinforcing steel bar corners, intersection points at reinforcing steel bar edges and intersection points in reinforcing steel bars; training a visual target detection model based on the steel bar intersection type labeling data to obtain a steel bar intersection detection model; acquiring an actual reinforcement picture, and inputting the reinforcement intersection detection model; forming a steel bar laying range, and connecting a middle area and an independent type area in the steel bar laying range by taking the steel bar inclination angle as an auxiliary to obtain a steel bar laying structure. Compared with the prior art, the method has the advantages of simplifying the calculation amount of recognition of the steel bar laying structure and the like.

Description

Technical Field

[0001] This invention relates to the technical field of steel reinforcement laying structure identification, and in particular to a method for identifying steel reinforcement laying structures through construction photographs. Background Technology

[0002] During construction, it is required to take photos after the steel reinforcement is laid for archiving. These photos are used to identify the structure of the steel reinforcement and analyze the results to identify problems in the steel reinforcement laying on site in a timely manner, so as to improve the quality of the project.

[0003] Traditional methods for identifying rebar laying structures in construction photos utilize computer vision edge detection algorithms to identify rebars in the image. The identified rebars are then converted into straight lines to analyze information such as quantity and uniformity. However, edge detection algorithms in photos introduce numerous interfering factors, affecting accuracy. Therefore, improved methods have emerged, primarily combining image recognition algorithms to first identify the construction area and / or individual rebar areas, then performing edge detection within these areas to enhance analysis accuracy. However, the high uncertainty of on-site image content makes these methods complex, resource-intensive, and their practical effectiveness less than ideal. Summary of the Invention

[0004] The purpose of this invention is to provide a method for identifying rebar laying structures through construction photographs, thereby simplifying the calculation required for identifying rebar laying structures.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for identifying rebar laying structures using construction photographs, the method comprising the following steps:

[0007] Set up data for labeling rebar intersection types. The data includes three types of labels on the rebar images: intersections at rebar corners, intersections on rebar edges, and intersections inside the rebar.

[0008] A visual target detection model is trained based on the rebar intersection type annotation data to obtain the rebar intersection detection model;

[0009] Obtain actual images of reinforcing bars, input them into the reinforcing bar intersection detection model, and obtain the bounding boxes of intersections at the corners of the reinforcing bars, intersections on the edges of the reinforcing bars, and intersections inside the reinforcing bars;

[0010] The boundary frames of the intersections at the corners and edges of the reinforcing bars are reduced to the middle area. Adjacent middle areas are connected to each other. The boundary frames of the intersections inside the reinforcing bars are reduced to independent type areas to form the reinforcing bar laying range. The inclination angle of the reinforcing bars is used as an auxiliary to connect the middle area and independent type areas within the reinforcing bar laying range to obtain the reinforcing bar laying structure.

[0011] Furthermore, the specific steps for setting the labeling data for the rebar intersection type are as follows:

[0012] For intersections on a rebar image, let the set of intersections be X = {xij}, = 1, ..., m, j = 1, ..., n. For xij, if i = 1 or i = m and j = 1 or j = n, then xij is marked as an intersection on a rebar corner. In the set of intersections excluding those on rebar corners, if i = 1 or i = m, or j = 1 or j = n, then xij is marked as an intersection on a rebar edge. Intersections in the set of intersections excluding those on rebar corners and edges are marked as intersections inside the rebar.

[0013] Furthermore, the central region is a concentric square or center point of the bounding box, wherein the side length of the concentric square is less than the side length of the bounding box.

[0014] Furthermore, the independent type region is a concentric square, concentric circle, or center point of the bounding box. The side length of the concentric square is less than the side length of the bounding box, and the concentric circle is located inside the bounding box.

[0015] Furthermore, using the rebar inclination angle identified by the straight-line algorithm as an aid, the intermediate area and independent type areas within the rebar laying range are connected to obtain the specific steps of the rebar laying structure:

[0016] For an independent type area, select other independent type areas or intermediate areas that are not connected to it in ascending order of distance from the independent type area. If the difference between the inclination angle of the connected line segment and the inclination angle of the corresponding reinforcement is less than a threshold, the connection is retained; otherwise, the connection is discarded. When the number of connections between the independent type area and other independent type areas or intermediate areas reaches the threshold, select the next independent type area and repeat the above steps until all independent type areas are traversed. Integrate all line segments, independent type areas, and intermediate areas to obtain the reinforcement laying structure.

[0017] Furthermore, the specific distance to this independent type of region is as follows:

[0018] The distance from the center point of an independent type region to the center point of that independent type region, or the distance from the center point of an intermediate region to the center point of that independent type region.

[0019] Furthermore, the method also includes: after obtaining the actual steel bar image, identifying the smallest spacing between the two steel bars in the actual steel bar image; if the spacing is less than the spacing threshold, then deleting these two steel bars.

[0020] Furthermore, the visual target detection model is the YOLO V8 model.

[0021] Furthermore, the inclination angle of the reinforcing bars is identified based on a straight line algorithm or by the shooting angle of the image.

[0022] Furthermore, the method also includes: after obtaining the rebar laying structure, analyzing the quantity and uniformity of the rebars in the rebar laying structure.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] This invention distinguishes between edge and internal node characteristics based on morphological differences, obtaining the bounding boxes of intersections at rebar corners, edges, and interiors, thus facilitating the determination of the laying range without requiring the training of other models. Furthermore, the machine vision edge detection in this invention serves only as an auxiliary method for determining the rebar direction; the primary detection content is the intersections obtained from target recognition. The final rebar laying structure mainly relies on these detected intersections. The target recognition method solves the rebar laying structure identification problem; the model is simple, computationally convenient, and easy to develop and implement. Meanwhile, machine vision edge detection (tilt angle), acting as a physical constraint, reduces algorithm complexity and computational load. Attached Figure Description

[0025] Figure 1 This is a flowchart of the present invention;

[0026] Figure 2 A diagram showing the type of reinforcement bar identification and labeling;

[0027] Figure 3 This is a schematic diagram illustrating the conversion of rebar identification results;

[0028] Figure 4 This is a schematic diagram of a steel reinforcement bar. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0030] To address the problems in identifying rebar laying structures, this invention proposes and implements a method for identifying rebar laying structures through construction photos. It employs image target recognition combined with graphic algorithms to reconstruct the rebar layout structure at the construction site, achieving accurate restoration of the rebar laying structure.

[0031] This invention proposes a method for identifying rebar laying structures through construction photographs, the method comprising the following steps:

[0032] Set up data for labeling rebar intersection types. The data includes three types of labels on the rebar images: intersections at rebar corners, intersections on rebar edges, and intersections inside the rebar.

[0033] A visual target detection model is trained based on the rebar intersection type annotation data to obtain the rebar intersection detection model;

[0034] Obtain actual images of reinforcing bars, input them into the reinforcing bar intersection detection model, and obtain the bounding boxes of intersections at the corners of the reinforcing bars, intersections on the edges of the reinforcing bars, and intersections inside the reinforcing bars;

[0035] The boundary frames of the intersections at the corners and edges of the reinforcing bars are reduced to the middle area. Adjacent middle areas are connected to each other. The boundary frames of the intersections inside the reinforcing bars are reduced to independent type areas to form the reinforcing bar laying range. The inclination angle of the reinforcing bars is used as an auxiliary to connect the middle area and independent type areas within the reinforcing bar laying range to obtain the reinforcing bar laying structure.

[0036] This invention employs a traditional visual object recognition model, obtaining a rebar intersection recognition model by labeling different types of rebar intersections. Then, according to... Figure 1 The process described above processes on-site images. First, the photos are fed into a model to identify various rebar intersections, and the model's output is obtained. Second, the rebar intersection location information output by the model is converted into points (or small-scale targets) in the image. Finally, the intersections are connected according to algorithm rules to obtain a rebar laying structure diagram. Based on this result, information such as the number and uniformity of rebars is analyzed.

[0037] The method for identifying rebar laying structure from construction photos described in this invention is based on a machine vision target recognition algorithm model. By identifying rebar intersections, the rebar laying structure diagram can be reconstructed using the identified rebar intersections.

[0038] The method for identifying rebar laying structures through construction photos described in this invention uses an algorithm to convert the range information of the identified target (rebar intersection) location into a point or a very small range target.

[0039] The method for identifying rebar laying structures through construction photos described in this invention can use machine vision edge recognition to determine the direction of the rebars near the intersection, or it can directly determine the direction information of the rebars by restricting the angle of the captured photos, which is convenient for determining how to connect adjacent intersections when connecting rebars.

[0040] The method for identifying rebar laying structures through construction photos described in this invention does not require identifying all intersections of a single rebar and other rebars when connecting intersections. It only requires identifying two or more intersections to determine the position of the rebar. This can improve the confidence threshold of the model recognition, thereby increasing the accuracy, while also taking into account the detection rate of the entire rebar.

[0041] The method for identifying rebar laying structures through construction photos described in this invention determines the rebar position by using the principle that the intersections are evenly distributed along the connecting line and at both ends when connecting all the intersections of a single rebar.

[0042] The method described in this invention for identifying rebar laying structures through construction photographs aims to identify and distinguish the intersection points in the middle and at the edges of the rebar area, and to determine the rebar laying area by the rebar intersection points at the edges.

[0043] The method for identifying rebar laying structures through construction photos described in this invention can analyze uniformity when there are no standard length markers, and can provide specific estimated values ​​of the variables that need to be measured when there are standard length markers.

[0044] This invention mainly uses photos of steel reinforcement laying during construction to analyze the steel reinforcement laying structure and determine whether it meets the construction technical requirements.

[0045] This invention is based on the fact that rebar is always laid in a straight line (or approximately a straight line). By marking the intersections of the rebars, theoretically, only two intersections need to be identified for each rebar to draw its entire laying position. Since the visual characteristics of rebar intersections differ significantly at the edges, corners, and middle positions, these three types of intersections can be marked separately. Thus, edge and corner intersections can be used to determine the rebar laying area, while internal intersections can be used to reconstruct the internal rebar laying.

[0046] Visual object detection algorithms (such as YOLO V8) are used to identify the intersections of reinforcing bars. The type of reinforcing bar intersection is labeled as follows: Figure 2 As shown, Figure 2 The diagram is divided into three categories: ① representing intersections at the corners of the reinforcing bars, ② representing intersections on the edges of the reinforcing bars, and ③ representing intersections inside the reinforcing bars. After extensive annotation, a recognition model is trained and used to identify intersections in construction photos after the reinforcing bars have been laid.

[0047] After the model is trained, using the recognition model to identify construction photos will output three types of recognition results, such as... Figure 3 As shown, Figure 3 In the diagram, ① represents the corner intersection point and its location; ③ represents the edge intersection point and its location; and ④ represents the internal intersection point and its location. The algorithm receives the recognition results and converts ① and ③ into points (or small-scale targets) of the same type ②, while converting the internal intersection point ④ into an independent type ⑤.

[0048] The outer target ② is used to confirm the rebar laying range. Within this range, an edge detection algorithm combined with a line transformation algorithm (such as the Canny algorithm plus the Hough Transform algorithm) is used to confirm the rebar direction and assist in connecting intersection points. The rebar laying range represents a closed figure obtained by all corner intersection points and edge intersection points on the outer perimeter.

[0049] The rebar laying structure reconstructed using the above method is as follows: Figure 4 As shown, the number of rebars can be calculated. Through approximate graphics, the rebar spacing can be analyzed. If there are standard-length objects, the spacing can be calculated proportionally; otherwise, the uniformity of the rebar spacing can be analyzed. Due to the depth of field in the photograph, distant rebars are compressed, causing the spacing to appear smaller and difficult to distinguish. Even small variations in the spacing between distant rebars can introduce significant errors. Therefore, a threshold is set during analysis; rebars with spacing below this threshold are not included in the calculation and analysis.

[0050] This invention has the following advantages over existing remote signaling acquisition methods:

[0051] 1) The target recognition method is used to solve the identification of the rebar laying structure. The model is simple, the calculation is convenient, and it is easy to develop and implement.

[0052] 2) The characteristics of edge and internal nodes are distinguished by their shape, which makes it easy to determine the laying range without the need to train other models.

[0053] 3) Two points determine a straight line. Theoretically, only two intersection points need to be identified for a single rebar. Therefore, increasing the confidence threshold for identification may reduce the detection rate while improving the accuracy, but it does not affect the overall detection rate of the rebar, thus balancing both the accuracy and detection rate.

[0054] 4) Machine vision edge detection is only used as an auxiliary method to determine the direction of the reinforcing bars, thus reducing the algorithm complexity.

[0055] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for identifying rebar laying structures through construction photographs, characterized in that, The method includes the following steps: Set up data for labeling rebar intersection types. The data includes three types of labels on the rebar images: intersections at rebar corners, intersections on rebar edges, and intersections inside the rebar. A visual target detection model is trained based on the rebar intersection type annotation data to obtain the rebar intersection detection model; Obtain actual images of reinforcing bars, input them into the reinforcing bar intersection detection model, and obtain the bounding boxes of intersections at the corners of the reinforcing bars, intersections on the edges of the reinforcing bars, and intersections inside the reinforcing bars; The boundary frames of the intersections at the corners and edges of the reinforcing bars are reduced to the middle area. Adjacent middle areas are connected to each other. The boundary frames of the intersections inside the reinforcing bars are reduced to independent type areas to form the reinforcing bar laying range. The inclination angle of the reinforcing bars is used as an auxiliary to connect the middle area and independent type areas within the reinforcing bar laying range to obtain the reinforcing bar laying structure.

2. The method for identifying rebar laying structures through construction photographs according to claim 1, characterized in that, The specific steps for setting the labeling data for rebar intersection types are as follows: For intersections on a rebar image, let the set of intersections be X = {xij}, = 1, ..., m, j = 1, ..., n. For xij, if i = 1 or i = m and j = 1 or j = n, then xij is marked as an intersection on a rebar corner. In the set of intersections excluding those on rebar corners, if i = 1 or i = m, or j = 1 or j = n, then xij is marked as an intersection on a rebar edge. Intersections in the set of intersections excluding those on rebar corners and edges are marked as intersections inside the rebar.

3. The method for identifying rebar laying structures through construction photographs according to claim 2, characterized in that, The central region is a concentric square or center point of the bounding box, wherein the side length of the concentric square is less than the side length of the bounding box.

4. The method for identifying rebar laying structures through construction photographs according to claim 2, characterized in that, Independent type regions are concentric squares, concentric circles, or the center point of the bounding box. The side length of the concentric square is less than the side length of the bounding box, and the concentric circle is located inside the bounding box.

5. The method for identifying rebar laying structures through construction photographs according to claim 2, characterized in that, Using the rebar inclination angle identified by the straight-line algorithm as an aid, the specific steps for connecting the middle area and independent type areas within the rebar laying range to obtain the rebar laying structure are as follows: For an independent type area, select other independent type areas or intermediate areas that are not connected to it in ascending order of distance from the independent type area. If the difference between the inclination angle of the connected line segment and the inclination angle of the corresponding reinforcement is less than a threshold, the connection is retained; otherwise, the connection is discarded. When the number of connections between the independent type area and other independent type areas or intermediate areas reaches the threshold, select the next independent type area and repeat the above steps until all independent type areas are traversed. Integrate all line segments, independent type areas, and intermediate areas to obtain the reinforcement laying structure.

6. The method for identifying rebar laying structures through construction photographs according to claim 5, characterized in that, The specific distance from this independent type region is: The distance from the center point of an independent type region to the center point of that independent type region, or the distance from the center point of an intermediate region to the center point of that independent type region.

7. The method for identifying rebar laying structures through construction photographs according to claim 1, characterized in that, The method also includes: after obtaining the actual steel bar image, identifying the smallest spacing between the two steel bars in the actual steel bar image; if the spacing is less than the spacing threshold, then deleting these two steel bars.

8. The method for identifying rebar laying structures through construction photographs according to claim 1, characterized in that, The visual target detection model is the YOLO V8 model.

9. The method for identifying rebar laying structures through construction photographs according to claim 1, characterized in that, The inclination angle of the reinforcing bars is identified based on a straight line algorithm or by the shooting angle of the image.

10. The method for identifying rebar laying structures through construction photographs according to claim 1, characterized in that, The method also includes: after obtaining the rebar laying structure, analyzing the quantity and uniformity of the rebars in the rebar laying structure.

Images