CN114707206A - Method, system, equipment and medium for identifying highway square column pier stirrup information - Google Patents
Method, system, equipment and medium for identifying highway square column pier stirrup information Download PDFInfo
- Publication number
- CN114707206A CN114707206A CN202210231682.1A CN202210231682A CN114707206A CN 114707206 A CN114707206 A CN 114707206A CN 202210231682 A CN202210231682 A CN 202210231682A CN 114707206 A CN114707206 A CN 114707206A
- Authority
- CN
- China
- Prior art keywords
- frame
- points
- stirrup
- line segments
- line
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/20—Configuration CAD, e.g. designing by assembling or positioning modules selected from libraries of predesigned modules
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Theoretical Computer Science (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Architecture (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method, a system, equipment and a medium for identifying highway square column pier stirrup information. The method comprises the following steps: automatic structured data output is achieved through model recognition and computer vision algorithms, and the structured data is used for subsequent three-dimensional model creation. The scheme provided by the invention can quickly identify the square pillar pier stirrups from the picture and accurately extract the structural information in the square pillar pier stirrups. The extracted structured information can help subsequent projects to quickly create related models. The method is suitable for the square pillar pier, but the outline of the frame and the point is not limited to the square pillar pier, the method has good practicability in a vase pier or a hollow thin-wall pier, the robustness is strong, and the method is not easily influenced by dirty pictures.
Description
Technical Field
The invention belongs to the field of highways, and particularly relates to a method, a system, equipment and a medium for identifying highway square pillar pier stirrup information.
Background
With the development of society and the progress of scientific technology, digital technology has been increasingly applied in various directions. In the current rapid development stage of China, digitization has become the core driving force of future economic development in China. However, the construction industry in China at present, particularly the digital words of the capital construction industry, still belong to a quite laggard stage, so the digitalization aiming at the capital construction field is indispensable. In the industries such as digital reference and house building in the field of capital construction, firstly, a design drawing is quickly converted into a digital model (bim model), and then project progress is transparently displayed in real time based on the digital model and project progress information implemented in combination, so as to prevent, control and early warn project risks and the like. The information of the square column pier stirrup in the highway field is an essential step in quickly converting a drawing into a digital model.
The square pier is a lower load bearing member for bearing an upper structure in a bridge structure, is arranged at a bridge span, and is called a square pier or a rectangular pier due to the rectangular cross section. The square column pier is an important component in the projects of highway bridges, railway bridges, overpasses, ramp bridges, overpasses and the like.
The stirrups are used for meeting the shear strength of the oblique section, connecting the stressed main reinforcement and the reinforcement of the concrete framework in the stressed area, hooping the main reinforcement on the cross section, and sequentially and longitudinally arranging the stirrups on the pier column (the stirrups on each cross section are the same in arrangement). The stirrups are divided into according to the style: single limb stirrup (drag hook stirrup), open rectangular stirrup, closed rectangular stirrup, rhombic stirrup, polygonal stirrup and the like.
At present, the construction mode of the capital construction industry mainly depends on issuing printing or scanning drawing paper, and most drawings are created by adopting a design institute through CAD. If the constructor wants to create a model, it has to do a restore operation manually by a person through CAD through printed or scanned drawings, or to structure the data manually, as shown in fig. 2.
Disadvantages of the prior art
The labor cost and the professional cost for manually inputting the structured information or manually recovering the CAD are very high, and the operation is difficult in an actual scene. In particular, in a drawing, there may be more than one figure of stirrup, so that the existing method is basically infeasible in practice, so that the process cannot be realized if the user is required to create the model by himself to the digitization degree. Without the use of automated identification techniques, the labor and professional costs can severely impede the digital roadways of the infrastructure direction.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method, a system, equipment and a medium for identifying highway square pillar stirrup information, and aims to solve the technical problems.
The invention discloses a method for identifying highway square column pier stirrup information in a first aspect, which comprises the following steps:
s1, acquiring training data of different forms of square pier, and training by using the training data to acquire an optimized target detection model;
step S2, inputting the picture into the target detection model to obtain detection results of the frames and the grooves, dividing the frames and the grooves in the corresponding directions into a group according to the detection results, and determining the inner and outer frame areas of the stirrups according to specific grouping;
step S3, finding out oblique line stirrups, namely oblique line segments, in the stirrup area according to the inner and outer frame areas of the stirrups;
step S4, acquiring a binary image after interference is cleared according to the acquired oblique line segment;
step S5, based on the interference-cleared binary image, finding main reinforcement combining points and the number of the main reinforcement combining points on the frame;
step S6, 3 parts of to-be-hooped line segments are totally arranged on the basis of the diagonal line segments of the hooped bars, the line segments of the transverse hooped bars and the line segments of the vertical hooped bars;
step S7, grouping the line segments of the stirrups and acquiring codes;
step S8, based on the main reinforcement combining points, combining the obtained grouping of the line segments of the stirrups, and matching the positions of the main reinforcement combining at the beginning and the end of the stirrups;
step S9, if a groove is found in step S2, combining the frame found in step S2 and the main merging point found in step S5 to obtain groove position information;
step S10, outputting the structured data according to the results obtained in step S5, step S7, step S8 and step S9.
According to the method of the first aspect of the present invention, in the step S1, the different forms of training data include: the device comprises a left frame, a left groove, an upper frame, an upper groove, a right frame, a right groove, a lower frame and a lower groove.
According to the method of the first aspect of the present invention, in step S2, the specific method for grouping the frames and the grooves in the corresponding directions includes:
the upper frame and the upper groove on the uppermost part are divided into a group, the left groove on the left frame on the leftmost part is divided into a group, the right frame and the right groove on the rightmost part are divided into a group, and the lower frame and the lower groove on the lowermost part are divided into a group;
the specific method for determining the inner and outer frame areas of the stirrups according to the specific grouping comprises the following steps:
s2.1, if the grooves are formed in the stirrup area, searching the single transverse line which is most inside the grooves by using a computer vision method;
s2.2, if the frame is the frame, finding a transverse line in the stirrup area range based on a computer vision method;
s2.3, if two transverse lines exist, setting the transverse lines as an inner frame and an outer frame respectively;
s2.4, if only one transverse line exists, splitting the line segment of the problem that the inner frame and the outer frame are too thick or the definition of the printed paper is not clear;
s2.5, if three transverse lines exist, taking the point with the most points on the two innermost line segments as an inner frame, and taking the adjacent outer line segments as outer frames;
s2.6, if the number of the transverse lines exceeds three, selecting three lines with the longest length, and then taking the point with the most points on the two line segments at the innermost side as an inner frame and taking the adjacent outer line segments as outer frames according to the step S2.5;
and S2.7, determining the inner and outer frame areas of the stirrups through the determined inner and outer frames.
According to the method of the first aspect of the present invention, in step S3, the specific method for finding the diagonal stirrups, i.e. the diagonal segments, within the stirrup area according to the inside and outside frame areas of the stirrups includes:
s3.1, ensuring that the included angle of the inclined line of the inner frame area and the outer frame area of the stirrup is kept more than 15 degrees in the horizontal or vertical direction;
s3.2, intersecting the oblique line with the line segment of the adjacent frame;
s3.3, filtering all the oblique lines meeting the conditions through the filling rate of the oblique line segments to generate line segments of the oblique lines;
and S3.4, positioning the start point and the end point of the line segment of the oblique line at the intersection point of the frame, and performing deduplication operation based on the distance between the center points of the line segment of the oblique line which needs to be more than 10 pixels.
According to the method of the first aspect of the present invention, in step S4, the specific method for acquiring the interference-cleaned binary image according to the obtained diagonal segments includes:
s4.1, removing horizontal line segments and vertical line segments which exceed 50% of the height of the outline in the internal binary image;
s4.2, removing internal contour line segments from the binary image with the horizontal line segments and the vertical line segments removed;
s4.3, removing the oblique line segment in the binary image with the outline segment removed;
and S4.4, point supplementing is carried out on the end point part of the oblique line segment.
According to the method of the first aspect of the present invention, in step S5, the specific method for finding the main parallel bar points and the number of the main parallel bar points on the frame based on the binary map after the interference is cleared includes:
step S5.1, determining the area of the search point: dynamically acquiring the region of the internal outline of the binary image after the interference is cleared;
s5.2, aiming at the area, if the oblique line is inclined, carrying out orientation correction based on affine change;
s5.3, if the small interfering oblique line exists in the area, ending the finding of the small oblique line segment, and removing the small oblique line segment;
s5.4, in the area where the small oblique line segments are removed, removing points of which the length exceeds 5 and the number of pixels in the height direction is less than 2;
step S5.5, in the area of removing the points with the length exceeding 5 and the pixel number less than 2 in the height direction, finding the points with the width larger than 2, wherein the starting positions of the points must start from the contour line, and taking the points as candidate main rib combining points;
s5.6, according to the gaps among the candidate main reinforcement combining points, solving the median distances among the candidate main reinforcement combining points, searching for candidate main reinforcement combining points with more than 2 median distances, and performing point supplementing operation on the middles of the candidate main reinforcement combining points;
s5.7, if the distance between the two candidate main reinforcement combining points is smaller than 1.5 times of the width of the diagonal line segment or smaller than 1.2 times of the width of the line segment in the removed horizontal line segment and the removed vertical line segment, combining the two candidate main reinforcement combining points;
step S5.8, the repaired points are transformed back to the original positions based on the radiation, at which point new contour points are obtained.
According to the method of the first aspect of the present invention, in step S7, the specific method for grouping the line segments of the stirrups and acquiring the codes includes:
s7.1, finding out the marking points of the circles, if the marking lines are the diagonal line segments, finding out 4 matched diagonal line segments based on the diagonal line segments to form an octagon, and taking pictures in the circles to perform character recognition to serve as the coding of the stirrups;
s7.2, based on the external contour, erasing the middle content, finding out the text with the N beginning through character recognition for the external content, finding out the closest horizontal line in the corresponding direction as the marking line, then finding out the corresponding stirrup line based on the end point of the marking line, wherein the two stirrup lines are used as a group, and the text with the N beginning is the code of the group.
The second aspect of the present invention discloses a system for identifying information of a square pillar pier stud of a road, the system comprising:
the first processing module is configured to acquire training data of different forms of the square pillar pier, and train the training data to acquire an optimized target detection model;
the second processing module is configured to input the picture into the target detection model to obtain detection results of the frames and the grooves, divide the frames and the grooves in the corresponding directions into a group according to the detection results, and determine the inner and outer frame areas of the stirrups according to specific grouping;
the third processing module is configured to find a diagonal stirrup, namely a diagonal segment, in the range of the stirrup area according to the inner and outer frame areas of the stirrup;
and the fourth processing module is configured to acquire the interference-cleaned binary image according to the acquired oblique line segment.
The fifth processing module is configured to find main reinforcement combining points and the number of the main reinforcement combining points on the frame based on the interference-cleared binary image;
the sixth processing module is configured to have 3 parts of line segments to be hooped in total based on the oblique line segments of the hooped reinforcement, the line segments of the transverse hooped reinforcement and the line segments of the vertical hooped reinforcement;
a seventh processing module configured to group the segments of the stirrup and obtain a code;
an eighth processing module configured to match, based on the main merge point, the obtained grouping of the line segments of the stirrup to the positions of the main merge at the beginning and the end of the stirrup.
The ninth processing module is configured to, if a groove is found in the second processing module, obtain groove position information by combining the frame found in the second processing module and the main reinforcement combining point found in the fifth processing module;
and the tenth processing module is configured to output the structured data according to the results obtained by the fifth processing module, the seventh processing module, the eighth processing module and the ninth processing module.
According to the system of the second aspect of the present invention, the first processing module is configured to, the different forms of training data include: the device comprises a left frame, a left groove, an upper frame, an upper groove, a right frame, a right groove, a lower frame and a lower groove.
According to the system of the second aspect of the present invention, the second processing module is configured such that the uppermost upper frame and the upper groove are grouped into one group, the leftmost left frame left groove is grouped into one group, the rightmost right frame and the rightmost right groove are grouped into one group, and the lowermost lower frame and the lower groove are grouped into one group;
the inner and outer frame areas of the stirrups are determined according to the specific grouping, and the specific steps are as follows:
if the stirrup region is a groove, searching the single transverse line which is closest to the inner part of the groove based on a computer vision method;
if the frame is the frame, searching a transverse line in the stirrup area range based on a computer vision method;
if two transverse lines are arranged, the transverse lines are respectively arranged as an inner frame and an outer frame;
if only one transverse line exists, splitting the line segments of the problems of too thick inner and outer frames or the definition of printing paper;
if three transverse lines exist, the point with the most points on the two line segments at the innermost side is used as an inner frame, and the adjacent outer line segments are used as outer frames;
if the number of the cross lines exceeds three, taking three lines with the longest length, and then taking the point with the most points on the two line segments at the innermost side as an inner frame and taking the adjacent external line segments as an outer frame according to the step S2.5;
and determining the inner and outer frame areas of the stirrups through the determined inner and outer frames.
According to the system of the second aspect of the present invention, the third processing module is configured to ensure that the angle of the oblique line of the inner and outer frame areas of the stirrup keeps an included angle greater than 15 degrees with the horizontal or vertical direction; the oblique line needs to be intersected with the line segment of the adjacent frame; filtering all the oblique lines meeting the conditions through the filling rate of the oblique line segments to generate oblique line segments; the starting point and the ending point of the line segment of the oblique line are positioned at the intersection point of the frame, and the distance between the center points of the line segments of the oblique line is required to be more than 10 pixels to perform the duplicate removal operation.
According to the system of the second aspect of the present invention, the fourth processing module is configured to remove horizontal line segments and vertical line segments exceeding 50% of the contour height in the internal binary image; removing the inner contour line segment from the binary image with the horizontal line segment and the vertical line segment removed; in the binary image with the outline line segment removed, removing the oblique line segment; and performing point complementing on the end point part of the oblique line segment.
According to the system of the second aspect of the present invention, the fifth processing module is configured to determine the area of the search point: dynamically acquiring the region of the internal outline of the binary image after the interference is cleared; for the region, if the oblique line is inclined, performing orientation correction based on affine change; if the small interfering oblique line exists in the area, ending the finding of the small oblique line segment, and removing the small oblique line segment; removing points in which the length exceeds 5 but the number of pixels in the height direction is less than 2 in the area where the small diagonal line segment is removed; in the region where the points with the length exceeding 5 but the number of pixels less than 2 in the height direction are removed, points with the width larger than 2 are searched, the starting positions of the points are required to start from the contour line, and the points are used as candidate main rib merging points; according to the gaps among the candidate main reinforcement combining points, the median distances among the candidate main reinforcement combining points are obtained, the candidate main reinforcement combining points with the median distances exceeding 2 median distances are searched, and point supplementing operation is needed among the candidate main reinforcement combining points; if the distance between the two candidate main reinforcement combining points is smaller than 1.5 times of the width of the diagonal line segment or smaller than 1.2 times of the width of the removed line segment in the horizontal line segment and the vertical line segment, combining the two candidate main reinforcement combining points; the repaired points are transformed back to the original position based on the radiation, at which point new contour points are obtained.
According to the system of the second aspect of the present invention, the seventh processing module is configured to find out the marked point of the circle, if the marked line is the diagonal line segment, find out 4 diagonal line segments matched based on the diagonal line segment to form an octagon, and take the picture in the circle therein to perform character recognition to be used as the coding of the stirrup; based on the external contour, erasing the middle content, finding out a text with the beginning of N through character recognition for the external content, finding out a transverse line with the closest corresponding direction as a marking line of the text, then finding out corresponding stirrup lines based on the end points of the marking line, taking the two stirrup lines as a group at the moment, and obtaining the text with the beginning of N as the code of the group.
A third aspect of the invention discloses an electronic device. The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor realizes the steps of the method for identifying the highway square pillar stirrup information in any one of the first aspect of the disclosure when executing the computer program.
A fourth aspect of the invention discloses a computer-readable storage medium. The computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of a method of identifying highway square pier stirrup information according to any one of the first aspect of the disclosure.
The scheme provided by the invention has the following beneficial effects: the square pillar pier stirrups can be quickly identified from the picture, and the structural information in the square pillar pier stirrups can be accurately extracted; the extracted structured information can help the subsequent engineering to quickly create a relevant model; the method is suitable for the square pillar pier, but the outline of the frame and the point is not limited to the square pillar pier, the method has good practicability in a vase pier or a hollow thin-wall pier, the robustness is strong, and the method is not easily influenced by dirty pictures.
Drawings
In order to more clearly illustrate the detailed description of the present invention or the technical solutions in the prior art, the detailed description or the technical solutions in the prior art will be briefly described below by using the attached drawings.
Fig. 1 is a flowchart of a method for identifying information of a square pillar pier stud of a road according to an embodiment of the invention;
FIG. 2 is a diagram of a stirrup according to the background art;
FIG. 3 is a schematic diagram of 8 types of data according to an embodiment of the invention;
FIG. 4 is a diagram illustrating the detection results of the frame and the groove according to the embodiment of the invention;
FIG. 5 is a flowchart of a specific method for determining the inside and outside frame areas of a stirrup according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of determining the inside and outside frame areas of a stirrup according to an embodiment of the invention;
FIG. 7 is a binary diagram after clearing of interference according to an embodiment of the present invention;
FIGS. 8a-8b are diagrams illustrating the effect of the main merge bar points on the left and upper sides according to the embodiment of the present invention;
FIG. 9 is a schematic diagram of finding diagonal stirrups within the stirrup zone, i.e. diagonal segments, according to an embodiment of the present invention;
FIG. 10 is a diagram illustrating the determination of the area of a search point according to an embodiment of the present invention;
FIG. 11 is a schematic diagram of an external annotation according to an embodiment of the present invention;
FIG. 12 is a schematic illustration of a callout line according to an embodiment of the present invention;
FIG. 13 is a schematic view of a stirrup line according to an embodiment of the invention;
fig. 14 is a diagram of an information system for identifying road square pillar stirrup according to an embodiment of the present invention;
fig. 15 is a block diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention.
The invention discloses a method for identifying highway square column pier stirrup information in a first aspect. Fig. 1 is a flowchart of a method of identifying highway square pillar pier stirrup information according to an embodiment of the present invention, as shown in fig. 1, the method comprising:
s1, acquiring training data of different forms of the square pier, and training by using the training data to acquire an optimized target detection model;
step S2, inputting the picture into the target detection model to obtain detection results of the frames and the grooves, dividing the frames and the grooves in the corresponding directions into a group according to the detection results, and determining the inner and outer frame areas of the stirrups according to specific grouping;
step S3, finding out oblique line stirrups, namely oblique line segments, in the stirrup area according to the inner and outer frame areas of the stirrups;
step S4, acquiring a binary image after interference is cleared according to the acquired oblique line segments;
step S5, based on the interference-cleared binary image, finding main reinforcement combining points and the number of the main reinforcement combining points on the frame;
step S6, 3 parts of to-be-hooped line segments are totally arranged on the basis of the diagonal line segments of the hooped bars, the line segments of the transverse hooped bars and the line segments of the vertical hooped bars;
step S7, grouping the line segments of the stirrups and acquiring codes;
step S8, based on the main reinforcement combining points, combining the obtained grouping of the line segments of the stirrups, and matching the positions of the main reinforcement combining at the beginning and the end of the stirrups;
step S9, if a groove is found in step S2, combining the frame found in step S2 and the main merging point found in step S5 to obtain groove position information;
step S10, outputting the structured data according to the results obtained in step S5, step S7, step S8 and step S9.
In step S1, training data of different forms of the square pier are obtained, and the training data are used to perform training to obtain the optimized target detection model.
In some embodiments, in the step S1, the different forms of training data include: the device comprises a left frame, a left groove, an upper frame, an upper groove, a right frame, a right groove, a lower frame and a lower groove.
Specifically, training data of different forms of a square pillar pier are acquired, 8 types of data are designed for the training data, and the 8 types are respectively a left frame (corrup _ ll), a left groove (corrup _ ls), an upper frame (corrup _ tl), an upper groove (corrup _ ts), a right frame (corrup _ rl), a right groove (corrup _ rs), a lower frame (corrup _ bl), and a lower groove (bs), wherein t represents top, s represents slot, l represents line, and corrup represents a stirrup, and the labels are shown in fig. 3. Training by using training data to obtain an optimized target detection model: the target detection model adopts a pre-training model of PP-YOLOv2, in which we adopt a backbone network of 50 layers of residual error networks (Resnet), and introduces a small amount of deformable convolution to improve the extraction capability of the features of the target detection.
In step S2, the picture is input into the target detection model to obtain detection results of the frames and the grooves, the frames and the grooves in the corresponding directions are grouped into one group according to the detection results, and the inner and outer frame regions of the stirrup are determined according to the specific grouping.
In some embodiments, in the step S2, the specific method for grouping the frames and the grooves in the corresponding directions includes:
the upper frame and the upper groove on the uppermost part are divided into a group, the left groove on the left frame on the leftmost part is divided into a group, the right frame and the right groove on the rightmost part are divided into a group, and the lower frame and the lower groove on the lowermost part are divided into a group;
the specific method for determining the inner and outer frame areas of the stirrups according to the specific grouping comprises the following steps:
s2.1, if the grooves are formed in the stirrup area, searching the single transverse line which is most inside the grooves by using a computer vision method;
s2.2, if the frame is the frame, finding a transverse line in the stirrup area range based on a computer vision method;
s2.3, if two transverse lines exist, the transverse lines are respectively set as an inner frame and an outer frame;
s2.4, if only one transverse line exists, splitting the line segment by considering the problems of too thick inner and outer frames or the definition of printed paper;
s2.5, if three transverse lines exist, considering the points with the most points on the two innermost line segments as an inner frame, and taking the adjacent outer line segments as an outer frame;
s2.6, if the number of the transverse lines exceeds three, selecting three lines with the longest length, and then taking the point with the most points on the two line segments at the innermost side as an inner frame and taking the adjacent outer line segments as outer frames according to the step S2.5;
and S2.7, determining the inner and outer frame areas of the stirrups through the determined inner and outer frames.
Specifically, the input picture is subjected to contour prediction. In practical applications, a picture that is as clear as possible and has a resolution of 216DPI is used as an input picture. The picture can be broadly understood as a picture containing a stirrup figure, but the picture containing the stirrup figure only can contain one stirrup figure.
At this time, the input picture is input into the model trained in step S1, and the result of detecting the frame and the groove can be obtained, as shown in fig. 4. By way of example in FIG. 4, we can find 1 corrup _ ts, one corrup _ bl, two corrup _ ll, one corrup _ ts, two corrup _ rl, and one corrup _ rs in the graph. The contour can now be obtained based on the algorithm flow for the upper and lower regions, as shown in fig. 5. The left and right frames are first grouped into two frames because of the groove information, and then the inner and outer frames can be obtained based on the flow 4 for each frame and groove, as shown in fig. 6. At this time, the upper line of the inner frame is a line segment, and the left line segment and the right line segment are composed of three line segments.
At this time, one part of information may be lost, and the prediction accuracy is error due to picture definition with high probability, and at this time, a target can be found through finding under the current threshold, and then, the target is compensated under the condition of low accuracy if the target is found to be lacking. This approach may greatly increase the flexibility of step S2.
In step S3, a diagonal stirrup, i.e., a diagonal segment, within the stirrup area is found according to the inner and outer frame areas of the stirrup.
In some embodiments, in the step S3, the specific method for finding the diagonal stirrup, i.e. the diagonal segment, in the range of the stirrup area according to the inner and outer frame areas of the stirrup includes:
s3.1, ensuring that the included angle of the inclined line of the inner frame area and the outer frame area of the stirrup is more than 15 degrees in the horizontal or vertical direction;
s3.2, intersecting the oblique line with the line segment of the adjacent frame;
s3.3, filtering all the oblique lines meeting the conditions through the filling rate of the oblique line segments to generate line segments of the oblique lines;
and S3.4, positioning the start point and the end point of the line segment of the oblique line at the intersection point of the frame, and performing deduplication operation based on the distance between the center points of the line segment of the oblique line which needs to be more than 10 pixels.
Specifically, finding slash segments identifies 8 slash segments in fig. 2. The diagonal segments then intersect the adjacent variable contours. At this time, the houghline method is adopted to find out the possible oblique lines, the obtained oblique lines exceed 1000, and the start and stop positions of the line segments are unknown, at this time, based on the frame found in step S2, the oblique lines are found to intersect with the adjacent frame, at this time, the candidate line segments can be obtained, and at this time, the number of the candidate line segments is reduced to 85 based on the filtering of 15 degrees. However, since each houghline search angle is 0.5 degrees, a plurality of candidate diagonal segments may appear in each diagonal line. And some oblique line segments are enough in point, but the line segments are long, but the threshold value is low, so that discontinuous line segments can be displayed. The filtering is performed according to the filling rate of the segment part (the diagonal segment has a point in a region of 3x 3) being more than 50%, and only 52 segments are remained. When the distance of the line segment is less than 10 pixels based on the center point, only 8 line segments remain after the filtering is finished, and the result is referred to fig. 10 (the diagonal line segments are continuous when the horizontal and vertical lines are removed because the correct diagonal line segments are found).
In step S4, a binary map after the interference is cleared is acquired based on the acquired diagonal segments.
In some embodiments, in the step S4, the specific method for acquiring the interference-cleaned binary image according to the acquired diagonal segments includes:
s4.1, removing horizontal line segments and vertical line segments which exceed 50% of the height of the outline in the internal binary image;
s4.2, removing internal contour line segments from the binary image from which the horizontal line segments and the vertical line segments are removed;
s4.3, removing the oblique line segment in the binary image with the outline segment removed;
and S4.4, point supplementing is carried out on the end point part of the oblique line segment, and the main rib combining point is prevented from being interrupted by points.
Specifically, the long horizontal and vertical lines and the diagonal line segments are removed to obtain a clean binary image after removal, as shown in fig. 7, at this time, the diagonal line is not seen in the image, but two points of the diagonal line are supplemented, and it can be found that the points passed by the 8 diagonal line segments are not divided into two parts although the diagonal line is removed.
In step S5, based on the interference-cleared binary map, main merging points and the number of main merging points on the frame are found.
In some embodiments, in the step S5, the specific method for finding the main merging bars and the number of the main merging bars on the frame based on the interference-cleared binary map includes:
s5.1, determining the area of the search point: dynamically acquiring the region of the internal outline of the binary image after the interference is cleared;
s5.2, aiming at the area, if the oblique line is inclined, carrying out orientation correction based on affine change;
step S5.3, if the area has the small interfering oblique lines, finishing finding the small oblique line segments by using computer vision of houghline _ p (the small oblique line segments can only be used for noise reduction and are not suitable for the step S3 because the finding mode is not accurate), and removing the small oblique line segments, so that the influence of the oblique line segments of the oblique hoop reinforcement line in the step S3 on the calculation of the main reinforcement point is solved;
s5.4, in the area where the small oblique line segments are removed, removing points which are longer than 5 and have the number of pixels less than 2 in the height direction, and ensuring that the noise points are not adhered to the main reinforcement points;
step S5.5, in the area of removing the points with the length exceeding 5 and the pixel number less than 2 in the height direction, finding the points with the width larger than 2 based on the computer vision findContour method, wherein the starting positions of the points are required to start from the contour line and the points are used as candidate main combining points;
s5.6, considering the gaps among the candidate main reinforcement combining points, solving the median distances among the candidate main reinforcement combining points, searching for candidate main reinforcement combining points with the median distances exceeding 2, performing point supplementing operation among the candidate main reinforcement combining points, and mainly considering that main reinforcement combining points of a horizontal line and a vertical line dry erasing part can be taken out;
s5.7, if the distance between the two candidate main reinforcement combining points is smaller than 1.5 times of the width of the diagonal line segment, or smaller than 1.2 times of the width of the removed line segments in the horizontal line segment and the vertical line segment, combining the two candidate main reinforcement combining points;
step S5.8, transform the repaired point back to the original position based on radiation, at this time, obtain a new contour point, and fig. 8a-8b show the effect of finding the main merging points on the left and upper sides.
Specifically, in step S5.1, the area shown in fig. 10 can be obtained with the frame as the distance. It is now possible after each flow process based on steps S5.1-S5.8 that the upper points total 11 and their positions are shown in fig. 8 b. Similarly, we take the left point, and since the frame has 3 points, each line from top to bottom has 4 points for each frame, and the total number of the points is 12, and their positions are shown in fig. 8 a.
In step S6, there are 3 sections of the wire to be hooped in total based on the diagonal sections of the hooped wire and the sections of the horizontal and vertical hooped wires.
In step S7, the segments of the stirrups are grouped and the code is acquired.
In some embodiments, in the step S7, the grouping the line segments of the stirrups, and the specific method for obtaining the codes includes:
s7.1, finding out the marking points of the circles, if the marking lines are the diagonal line segments, finding out 4 matched diagonal line segments based on the diagonal line segments to form an octagon, and taking pictures in the circles to perform character recognition to serve as the coding of the stirrups;
s7.2, based on the external contour, erasing the middle content, finding out the text with the N beginning through character recognition for the external content, finding out the closest horizontal line in the corresponding direction as the marking line, then finding out the corresponding stirrup line based on the end point of the marking line, wherein the two stirrup lines are used as a group, and the text with the N beginning is the code of the group.
In particular, 8 diagonal stirrup wires have been found, as well as 8 transverse stirrup wires and 4 longitudinal stirrup wires, as shown in fig. 13. At this time, two circles with the numbers 5 and 6 are found according to the circle in step S7.1, the marked lines of the two circles respectively point to the outer oblique line segment and the inner oblique line segment, and 4 candidate oblique lines of the two circles are found according to the flow, so that two groups of stirrups with the numbers 5 and 6 respectively are obtained. For S7.2, at this time, N2, N2, N4, N4, N3, N3, 6 external labels are found as shown in fig. 11, the identification characters are found while corresponding label lines are found, the found label lines are shown in fig. 12, and at this time, all end points of the label lines, corresponding 8 transverse hoop lines and 4 longitudinal hoop lines are found. All the stirrup wires found in the steps S6-S7 have been grouped so far.
In step S8, the positions of the main merge bars at the start and end of the stirrup are matched based on the main merge bar points in combination with the obtained grouping of the line segments of the stirrup.
In step S9, if a groove is found in step S2, groove position information is obtained in combination with the frame found in step S2 and the main merge point found in step S5.
In step S10, structured data is output according to the results obtained in step S5, step S7, step S8, and step S9.
The content of the structured data output includes:
1. the number of main parallel bars on the outline in the square pillar pier stirrup;
2. the stirrup form of the square pillar pier stirrup is as follows: coding including starting and ending rebar points from different sides and stirrups;
3. position information of the groove on the contour;
4. the type of main merge rib in the edge profile.
Specifically, the process of step S8 mainly includes matching the stirrup points found in step S5, grouping in step S7, and calculating to determine the position of each stirrup group.
Steps S9-S10 are to determine the position information of the stirrups and grooves, and to gather the information of S8, the json structure has been outputted by the main reinforcement combining point information of step S5, with the result.
The invention discloses a system for identifying information of square column pier studs of a highway in a second aspect. Fig. 14 is a structural diagram of a system for recognizing road square pillar stirrup information according to an embodiment of the present invention; as shown in fig. 14, the system 100 includes:
the first processing module 101 is configured to obtain training data of different forms of the square pier, and train with the training data to obtain an optimized target detection model;
the second processing module 102 is configured to input the picture into the target detection model, obtain detection results of the frames and the grooves, combine the detection results, divide the frames and the grooves in the corresponding directions into a group, and determine inner and outer frame areas of the stirrups according to specific grouping;
a third processing module 103, configured to find a diagonal stirrup, i.e. a diagonal segment, within the stirrup area according to the inner and outer frame areas of the stirrup;
and the fourth processing module 104 is configured to acquire the interference-cleaned binary image according to the acquired diagonal segments.
A fifth processing module 105, configured to find main parallel rib points and the number of main parallel rib points on a frame based on the interference-cleaned binary image;
the sixth processing module 106 is configured to add line segments of the vertical and horizontal stirrups based on the diagonal line segments of the stirrups, so that there are 3 sections of the to-be-hooped line segments in total;
a seventh processing module 107 configured to group the segments of the stirrups and obtain the codes;
an eighth processing module 108 configured to match, based on the master merge point, the positions of the master merge to the beginning and the end of the stirrup in combination with the resulting grouping of the segments of the stirrup.
A ninth processing module 109 configured to, if a groove is found in the second processing module, obtain groove position information by combining the frame found in the second processing module and the main reinforcement point found in the fifth processing module;
a tenth processing module 110 configured to output the structured data according to the results obtained by the fifth, seventh, eighth and ninth processing modules.
According to the system of the second aspect of the present invention, the first processing module 101 is specifically configured to, the different forms of training data include: the device comprises a left frame, a left groove, an upper frame, an upper groove, a right frame, a right groove, a lower frame and a lower groove.
Specifically, training data of different forms of the square pier are obtained, 8 types of data are designed for the training data and are marked, and the 8 types are respectively a left frame (corrup _ ll), a left groove (corrup _ ls), an upper frame (corrup _ tl), an upper groove (corrup _ ts), a right frame (corrup _ rl), a right groove (corrup _ rs), a lower frame (corrup _ bl) and a lower groove (bs), and the marks are shown in fig. 3. Training by using training data to obtain an optimized target detection model: the target detection model adopts a pre-training model of PP-YOLOv2, wherein a backbone network of 50 layers of residual error networks (Resnet) is adopted, and a small amount of deformable convolution is introduced to improve the extraction capability of the characteristics of target detection.
According to the system of the second aspect of the present invention, the second processing module 102 is specifically configured such that the specific method for grouping the frames and the grooves in the corresponding directions includes:
the upper frame and the upper groove on the uppermost part are divided into a group, the left groove on the left frame on the leftmost part is divided into a group, the right frame and the right groove on the rightmost part are divided into a group, and the lower frame and the lower groove on the lowermost part are divided into a group;
the specific method for determining the inner and outer frame areas of the stirrups according to the specific grouping comprises the following steps:
if the stirrup area is a groove, searching the innermost single transverse line of the groove based on a computer vision method;
if the frame is the frame, searching a transverse line for the stirrup area range based on a computer vision method;
if two transverse lines are arranged, the transverse lines are respectively arranged as an inner frame and an outer frame;
if only one transverse line exists, the line segment is split by considering the problems of too thick inner and outer frames or the definition of printing paper;
if three transverse lines exist, the point with the most points on the two line segments at the innermost side is taken as an inner frame, and the adjacent outer line segments are taken as outer frames;
if the number of the transverse lines exceeds three, selecting three lines with the longest length, and then carrying out inner frame according to the step S2.5;
and determining the inner and outer frame areas of the stirrups through the determined inner and outer frames.
Specifically, the input picture is subjected to contour prediction. In practical applications, a picture that is as clear as possible and has a resolution of 216DPI is used as an input picture. The picture can be broadly understood as a picture containing a stirrup figure, but the picture containing the stirrup figure only can contain one stirrup figure.
At this time, the input image is input into the model trained in the first processing module 101, and the result of detecting the frame and the groove can be obtained, as shown in fig. 4. By way of example in FIG. 4, we can find 1 corrup _ ts, one corrup _ bl, two corrup _ ll, one corrup _ ts, two corrup _ rl, and one corrup _ rs in the graph. The contour can now be obtained based on the algorithm flow for the upper and lower regions, as shown in fig. 5. The left and right frames are first grouped into two frames because of the groove information, and then the inner and outer frames can be obtained based on the flow 4 for each frame and groove, as shown in fig. 6. At this time, the upper line of the inner frame is a line segment, and the left line segment and the right line segment are composed of three line segments.
According to the system of the second aspect of the present invention, the third processing module 103 is specifically configured to, the specific method for finding the diagonal stirrups, i.e. diagonal segments, within the range of the stirrup area according to the inner and outer frame areas of the stirrup includes:
the angle of the oblique line of the inner frame area and the outer frame area of the stirrup needs to be ensured to be kept at an included angle of more than 15 degrees with the horizontal direction or the vertical direction; the oblique line needs to be intersected with the line segment of the adjacent frame; filtering all oblique lines meeting the conditions through the filling rate of oblique line segments to generate line segments of the oblique lines;
the starting point and the ending point of the line segment of the oblique line are positioned at the intersection point of the frame, and the distance between the center points of the line segments of the oblique line is required to be more than 10 pixels to perform the duplicate removal operation.
Specifically, finding slash segments identifies 8 slash segments in fig. 2. The diagonal segments then intersect the adjacent variable contours. At this time, the houghline method is adopted to find out possible oblique lines, the number of the obtained oblique lines exceeds 1000, and the start-stop positions of the line segments are unknown, at this time, based on the frame found in the second processing module 102, the oblique lines are found to intersect with the adjacent frame, at this time, candidate line segments can be obtained, and at this time, the number of the candidate line segments is reduced to 85 based on the filtering of 15 degrees. However, since each houghline search angle is 0.5 degrees, a plurality of candidate diagonal segments may appear in each diagonal line. And some oblique line segments are enough in point, but the line segments are long, but the threshold value is low, so that discontinuous line segments can be displayed. The filtering is performed according to the filling rate of the segment part (the diagonal segment has a point in a region of 3x 3) being more than 50%, and only 52 segments are remained. When the distance of the line segment is less than 10 pixels based on the center point, only 8 line segments remain after the filtering is finished, and the result is referred to fig. 10 (the diagonal line segments are continuous when the horizontal and vertical lines are removed because the correct diagonal line segments are found).
According to the system of the second aspect of the present invention, the fourth processing module 104 is specifically configured to, the specific method for acquiring the interference-cleaned binary image according to the acquired diagonal segments includes:
removing horizontal line segments and vertical line segments which exceed 50% of the height of the outline in the internal binary image;
removing the inner contour line segment from the binary image with the horizontal line segment and the vertical line segment removed;
in the binary image with the outline line segment removed, removing the oblique line segment;
and (4) point supplementing is carried out on the end point part of the inclined line segment, so that the point is prevented from breaking the main rib combining point.
Specifically, the long horizontal and vertical lines and the diagonal line segments are removed to obtain a clean binary image after removal, as shown in fig. 7, at this time, the diagonal line is not seen in the image, but two points of the diagonal line are supplemented, and it can be found that the points passed by the 8 diagonal line segments are not divided into two parts although the diagonal line is removed.
According to the system of the second aspect of the present invention, the fifth processing module 105 is specifically configured to, based on the interference-cleared binary map, find main merging points and the number of main merging points on a frame by a specific method including:
determining the area of the search point: dynamically acquiring the region of the internal outline of the binary image after the interference is cleared;
for the region, if the oblique line is inclined, performing orientation correction based on affine change;
if the area has the small interfering oblique lines, finishing finding the small oblique line segments by using computer vision of houghline _ p (the small oblique line segments can only be used for noise reduction and are not suitable for the step S3 due to inaccurate finding mode), and removing the small oblique line segments, so that the influence of the oblique line segments of the oblique hoop reinforcement line in the step S3 on the calculation of the main reinforcement point is solved;
in the area where the small oblique line segment is removed, removing points which are longer than 5 and have the number of pixels less than 2 in the height direction, and ensuring that the noise points can not be adhered to the main rib combining points;
in a region where points having a length exceeding 5 but having a pixel number less than 2 in the height direction are removed, points having a width greater than 2 are found based on the computer vision findContour method, the starting positions of the points must start from the contour line, and the points are taken as candidate principal merging points;
considering the gap between the candidate main reinforcement combining points, solving the median distance between the candidate main reinforcement combining points, searching the candidate main reinforcement combining points with the median distance of more than 2, wherein point supplementing operation is needed in the middle of the candidate main reinforcement combining points, and mainly considering that the main reinforcement combining points of the horizontal and vertical line dry erase part can be taken out;
if the distance between the two candidate main reinforcement combining points is smaller than 1.5 times of the width of the diagonal line segment or smaller than 1.2 times of the width of the removed line segments in the horizontal line segment and the vertical line segment, combining the two candidate main reinforcement combining points;
the repaired points are transformed back to the original position based on the radiation, now resulting in new contour points, fig. 8a-8b show the effect of finding the main merging points on the left and upper side.
Specifically, the area of fig. 10 can be obtained by using the above frame as the distance. It is found that the upper dots have a total of 11 dots and their positions are shown in fig. 8b based on the above flow processing. Similarly, we take the left point, and since the frame has 3 points, each line from top to bottom has 4 points for each frame, and the total number of the points is 12, and their positions are shown in fig. 8 a.
According to the system of the second aspect of the present invention, the seventh processing module 107 is specifically configured to group the line segments of the stirrups, and a specific method for acquiring the code includes:
finding out the marking points of the circles, if the marking lines are the diagonal line segments, finding out 4 matched diagonal line segments based on the diagonal line segments to form an octagon, and taking pictures in the circles to perform character recognition to serve as the coding of the stirrups;
based on the external contour, erasing the middle content, finding out a text with the beginning of N through character recognition for the external content, finding out a transverse line with the closest corresponding direction as a marking line of the text, then finding out corresponding stirrup lines based on the end points of the marking line, taking the two stirrup lines as a group at the moment, and obtaining the text with the beginning of N as the code of the group.
In particular, 8 diagonal stirrup wires have been found, as well as 8 transverse stirrup wires and 4 longitudinal stirrup wires, as shown in fig. 13. At this time, two circles with the numbers of 5 and 6 are found according to the circles in the method, the marking lines of the two circles respectively point to the outer oblique line segment and the inner oblique line segment, and 4 candidate oblique lines of the two circles are found according to the process, so that two groups of stirrups with the numbers of 5 and 6 are obtained. For the above method, N2, N2, N4, N4, N3, N3, 6 external labels are found at this time, as shown in fig. 11, the corresponding label lines are found while the identification characters are found, the found label lines are shown in fig. 12, and at this time, all the end points of the label lines, the corresponding 8 transverse hoop lines and the corresponding 4 longitudinal hoop lines are found.
A third aspect of the invention discloses an electronic device. The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor realizes the steps of the method for identifying the highway square pillar pier stirrup information in any one of the first aspect of the disclosure when executing the computer program.
Fig. 15 is a block diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 15, the electronic device includes a processor, a memory, a communication interface, a display screen, and an input device, which are connected by a system bus. Wherein the processor of the electronic device is configured to provide computing and control capabilities.
A fourth aspect of the invention discloses a computer-readable storage medium. The computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of the method of identifying highway square pier stirrup information of any one of the first aspect of the disclosure.
Claims (10)
1. A method of identifying highway square pillar pier stirrup information, the method comprising:
s1, acquiring training data of different forms of square pier, and training by using the training data to acquire an optimized target detection model;
step S2, inputting the picture into the target detection model to obtain detection results of the frames and the grooves, dividing the frames and the grooves in the corresponding directions into a group according to the detection results, and determining the inner and outer frame areas of the stirrups according to specific grouping;
step S3, finding out oblique line stirrups, namely oblique line segments, in the stirrup area according to the inner and outer frame areas of the stirrups;
step S4, acquiring a binary image after interference is cleared according to the acquired oblique line segment;
step S5, based on the interference-cleared binary image, finding main reinforcement combining points and the number of the main reinforcement combining points on the frame;
step S6, 3 parts of to-be-hooped line segments are totally arranged on the basis of the diagonal line segments of the hooped bars, the line segments of the transverse hooped bars and the line segments of the vertical hooped bars;
step S7, grouping the line segments of the stirrups and acquiring codes;
step S8, based on the main reinforcement combining points, combining the obtained grouping of the line segments of the stirrups, and matching the positions of the main reinforcement combining at the beginning and the end of the stirrups;
step S9, if a groove is found in step S2, combining the frame found in step S2 and the main merging point found in step S5 to obtain groove position information;
step S10, outputting the structured data according to the results obtained in step S5, step S7, step S8 and step S9.
2. The method of claim 1, wherein in the step S1, the different forms of training data include:
the device comprises a left frame, a left groove, an upper frame, an upper groove, a right frame, a right groove, a lower frame and a lower groove.
3. The method of claim 2, wherein in the step S2, the specific method for grouping the frames and the grooves in the corresponding directions comprises:
the upper frame and the upper groove on the uppermost part are divided into a group, the left groove on the left frame on the leftmost part is divided into a group, the right frame and the right groove on the rightmost part are divided into a group, and the lower frame and the lower groove on the lowermost part are divided into a group;
the specific method for determining the inner and outer frame areas of the stirrups according to the specific grouping comprises the following steps:
s2.1, if the grooves are formed in the stirrup area, searching the single transverse line which is closest to the inner part of the grooves by a computer vision method;
s2.2, if the frame is the frame, searching a transverse line for the stirrup area range based on a computer vision method;
s2.3, if two transverse lines exist, the transverse lines are respectively set as an inner frame and an outer frame;
s2.4, if only one transverse line exists, splitting the line segment of the problem that the inner frame and the outer frame are too thick or the definition of the printed paper is not clear;
s2.5, if three transverse lines exist, taking the point with the most points on the two innermost line segments as an inner frame, and taking the adjacent outer line segments as outer frames;
s2.6, if the number of the transverse lines exceeds three, selecting three lines with the longest length, and then taking the point with the most points on the two line segments at the innermost side as an inner frame and taking the adjacent outer line segments as outer frames according to the step S2.5;
and S2.7, determining the inner and outer frame areas of the stirrups through the determined inner and outer frames.
4. The method of claim 1, wherein in step S3, the specific method for finding the diagonal stirrups (diagonal segments) in the stirrup area according to the inside and outside frame areas of the stirrups comprises:
s3.1, ensuring that the included angle of the inclined line of the inner frame area and the outer frame area of the stirrup is more than 15 degrees in the horizontal or vertical direction;
s3.2, intersecting the oblique line with the line segment of the adjacent frame;
s3.3, filtering all oblique lines meeting the conditions through the filling rate of oblique line segments to generate line segments of the oblique lines;
and S3.4, positioning the start point and the end point of the line segment of the oblique line at the intersection point of the frame, and performing deduplication operation based on the distance between the center points of the line segment of the oblique line which needs to be more than 10 pixels.
5. The method for identifying road pier stirrup information according to claim 1, wherein in the step S4, the specific method for acquiring the interference-cleaned binary image according to the acquired diagonal segments comprises:
s4.1, removing horizontal line segments and vertical line segments which exceed 50% of the height of the outline in the internal binary image;
s4.2, removing internal contour line segments from the binary image with the horizontal line segments and the vertical line segments removed;
s4.3, removing the oblique line segment in the binary image with the outline segment removed;
and S4.4, performing point supplementing on the end point part of the oblique line segment.
6. The method of claim 5, wherein in the step S5, the specific method for finding the main reinforcement combining points and the number of the main reinforcement combining points on the frame based on the interference-cleared binary image comprises:
step S5.1, determining the area of the search point: dynamically acquiring the region of the internal outline of the binary image after the interference is cleared;
s5.2, aiming at the area, if the oblique line is inclined, carrying out orientation correction based on affine change;
s5.3, if the small interfering oblique line exists in the area, ending the finding of the small oblique line segment, and removing the small oblique line segment;
s5.4, in the area where the small oblique line segments are removed, removing points of which the length exceeds 5 and the number of pixels in the height direction is less than 2;
step S5.5, in the area which eliminates the points with the length exceeding 5 but the pixel number less than 2 in the height direction, finding the points with the width more than 2, wherein the starting positions of the points must start from the contour line, and taking the points as candidate main rib combining points;
s5.6, according to the gaps among the candidate main reinforcement combining points, solving the median distances among the candidate main reinforcement combining points, searching for candidate main reinforcement combining points with more than 2 median distances, and performing point supplementing operation on the middles of the candidate main reinforcement combining points;
s5.7, if the distance between the two candidate main reinforcement combining points is smaller than 1.5 times of the width of the diagonal line segment or smaller than 1.2 times of the width of the line segment in the removed horizontal line segment and the removed vertical line segment, combining the two candidate main reinforcement combining points;
step S5.8, the repaired points are transformed back to the original positions based on the radiation, at which point new contour points are obtained.
7. The method of claim 1, wherein in step S7, the grouping of the segments of the stirrup and the specific method of obtaining the code comprises:
s7.1, finding out the marking points of the circles, if the marking lines are the diagonal line segments, finding out 4 matched diagonal line segments based on the diagonal line segments to form an octagon, and taking pictures in the circles to perform character recognition to be used as the coding of the stirrups;
s7.2, based on the external contour, erasing the middle content, finding out the text with the N beginning through character recognition for the external content, finding out the closest horizontal line in the corresponding direction as the marking line, then finding out the corresponding stirrup line based on the end point of the marking line, wherein the two stirrup lines are used as a group, and the text with the N beginning is the code of the group.
8. A system for identifying highway square column pier stud information, the system comprising:
the first processing module is configured to acquire training data of different forms of the square pillar pier, and train the training data to acquire an optimized target detection model;
the second processing module is configured to input the picture into the target detection model to obtain detection results of the frames and the grooves, divide the frames and the grooves in the corresponding directions into a group according to the detection results, and determine the inner and outer frame areas of the stirrups according to specific grouping;
the third processing module is configured to find a diagonal stirrup, namely a diagonal segment, in the range of the stirrup area according to the inner and outer frame areas of the stirrup;
the fourth processing module is configured to acquire a binary image after interference is cleared according to the acquired oblique line segment;
the fifth processing module is configured to find main reinforcement combining points and the number of the main reinforcement combining points on the frame based on the interference-cleared binary image;
the sixth processing module is configured to have 3 parts of line segments to be hooped in total based on the oblique line segments of the hooped reinforcement, the line segments of the transverse hooped reinforcement and the line segments of the vertical hooped reinforcement;
a seventh processing module configured to group the segments of the stirrup and obtain a code;
an eighth processing module configured to match, based on the main reinforcement merging point, the positions of the main reinforcement merging at the start and the end of the stirrup in combination with the obtained grouping of the line segments of the stirrup;
the ninth processing module is configured to, if a groove is found in the second processing module, obtain groove position information by combining the frame found in the second processing module and the main reinforcement combining point found in the fifth processing module;
and the tenth processing module is configured to output the structured data according to the results obtained by the fifth processing module, the seventh processing module, the eighth processing module and the ninth processing module.
9. An electronic device, characterized in that the electronic device comprises a memory storing a computer program and a processor implementing the steps of a method of identifying highway square pier stirrup information according to any of claims 1 to 7 when the computer program is executed by the processor.
10. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, carries out the steps of a method of identifying highway square pier stirrup information according to any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210231682.1A CN114707206B (en) | 2022-03-10 | 2022-03-10 | Method, system, equipment and medium for identifying highway square column pier stirrup information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210231682.1A CN114707206B (en) | 2022-03-10 | 2022-03-10 | Method, system, equipment and medium for identifying highway square column pier stirrup information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114707206A true CN114707206A (en) | 2022-07-05 |
CN114707206B CN114707206B (en) | 2022-11-29 |
Family
ID=82169406
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210231682.1A Active CN114707206B (en) | 2022-03-10 | 2022-03-10 | Method, system, equipment and medium for identifying highway square column pier stirrup information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114707206B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN207846229U (en) * | 2017-12-29 | 2018-09-11 | 中交二公局第一工程有限公司 | The integral prefabricated moulding bed of rectangle pier shaft steel reinforcement cage |
CN111179232A (en) * | 2019-12-20 | 2020-05-19 | 山东大学 | Steel bar size detection system and method based on image processing |
CN112069562A (en) * | 2020-09-19 | 2020-12-11 | 南昌大学 | Zero-collision rapid arrangement method for three-way hoop reinforcement cage structure in rectangular component |
CN113051639A (en) * | 2021-03-05 | 2021-06-29 | 福建省晨曦信息科技股份有限公司 | Stirrup and lacing wire identification method, computer equipment and readable storage medium |
CN113051640A (en) * | 2021-03-05 | 2021-06-29 | 福建省晨曦信息科技股份有限公司 | Column proof data reproduction method, computer device and readable storage medium |
CN113188529A (en) * | 2021-05-12 | 2021-07-30 | 中国五冶集团有限公司 | Bridge rectangular pier stud template positioning detection method and detection system |
WO2021218114A1 (en) * | 2020-04-28 | 2021-11-04 | 昆山市建设工程质量检测中心 | Method for measuring surface roughness of precast composite concrete slab |
-
2022
- 2022-03-10 CN CN202210231682.1A patent/CN114707206B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN207846229U (en) * | 2017-12-29 | 2018-09-11 | 中交二公局第一工程有限公司 | The integral prefabricated moulding bed of rectangle pier shaft steel reinforcement cage |
CN111179232A (en) * | 2019-12-20 | 2020-05-19 | 山东大学 | Steel bar size detection system and method based on image processing |
WO2021218114A1 (en) * | 2020-04-28 | 2021-11-04 | 昆山市建设工程质量检测中心 | Method for measuring surface roughness of precast composite concrete slab |
CN112069562A (en) * | 2020-09-19 | 2020-12-11 | 南昌大学 | Zero-collision rapid arrangement method for three-way hoop reinforcement cage structure in rectangular component |
CN113051639A (en) * | 2021-03-05 | 2021-06-29 | 福建省晨曦信息科技股份有限公司 | Stirrup and lacing wire identification method, computer equipment and readable storage medium |
CN113051640A (en) * | 2021-03-05 | 2021-06-29 | 福建省晨曦信息科技股份有限公司 | Column proof data reproduction method, computer device and readable storage medium |
CN113188529A (en) * | 2021-05-12 | 2021-07-30 | 中国五冶集团有限公司 | Bridge rectangular pier stud template positioning detection method and detection system |
Non-Patent Citations (5)
Title |
---|
CHITRAKALA MUTHUVEERAPPAN 等: "A KNOWLEDGE BASED SYSTEM OF ELEMENT REGOGNITION IN 2D CONSTRUCTION DRAWINGS FOR AN AUTOMATED BILLS OF QUANTITIES", 《INTERNATIONAL JOURNAL OF REAL ESTATE STUDIES》 * |
曹阳 等: "建筑图钢筋自动识别与统计***VHDetails的设计与实现", 《计算机辅助工程》 * |
李伟青: "建筑构件智能识别方法研究", 《浙江大学学报(理学版)》 * |
罗志伟 等: "截面表示法柱平面图的自动识别方法", 《计算机应用研究》 * |
韩虎: "矩形桥墩中箍筋长度的计算方法", 《安徽建筑》 * |
Also Published As
Publication number | Publication date |
---|---|
CN114707206B (en) | 2022-11-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110738207B (en) | Character detection method for fusing character area edge information in character image | |
US20200302189A1 (en) | Lane line data processing method and apparatus, computer device, and storage medium | |
CN107633250B (en) | Character recognition error correction method, error correction system and computer device | |
CN105184292B (en) | The structural analysis of handwritten form mathematical formulae and recognition methods in natural scene image | |
CN109190481B (en) | Method and system for extracting road material of remote sensing image | |
US20180292328A1 (en) | Information processing device and information processing method | |
Sánchez-Rodríguez et al. | Automated detection and decomposition of railway tunnels from Mobile Laser Scanning Datasets | |
WO2018084715A1 (en) | Method and system for transforming handwritten text to digital ink | |
CN110838105B (en) | Business process model image recognition and reconstruction method | |
CN110689012A (en) | End-to-end natural scene text recognition method and system | |
CN110390666A (en) | Road damage detecting method, device, computer equipment and storage medium | |
US20160019419A1 (en) | Method and apparatus for generating map data based on construction designs | |
CN116311310A (en) | Universal form identification method and device combining semantic segmentation and sequence prediction | |
US11587299B2 (en) | Systems and methods for detection of anomalies in civil infrastructure using context aware semantic computer vision techniques | |
CN115953797A (en) | Form recognition method, document acquisition method, and storage medium | |
Wang et al. | Improving facade parsing with vision transformers and line integration | |
CN114707206B (en) | Method, system, equipment and medium for identifying highway square column pier stirrup information | |
CN113435370A (en) | Method and device for obtaining vehicle queuing length based on image feature fusion | |
CN115346206B (en) | License plate detection method based on improved super-resolution deep convolution feature recognition | |
CN115761782A (en) | Road engineering drawing title bar information extraction method | |
JPH04141783A (en) | Device and method for preparing electron beam graphic data | |
CN116778497A (en) | Method and device for identifying hand well number, computer equipment and storage medium | |
JPH06208625A (en) | Method and device for processing image | |
CN114283364A (en) | Detection method and detection device for pet tether and electronic equipment | |
KR100954137B1 (en) | Edge-based text localization and segmentation algorithms for automatic slab information recognition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |