A method for identifying a leakage binding steel bar joint based on node space analysis and a binding quality evaluation method

Abstract

The application discloses a kind of based on node space analysis's identification method of leakage binding steel bar node, binding quality evaluation method, using computer vision identification method, extracts steel mesh node binding state and geometric feature;Then design computational geometry algorithm, analyze the spatial relationship of steel bar node, determine the adjacent position relationship of each steel bar node;Again through plane geometry star topological structure model and optimization algorithm, identify leakage binding steel bar node, calculate qualified binding rate.The method of the application only needs to identify and calculate the information of steel bar node in steel mesh, saves the processing of steel bar section, target detection recognition speed is faster;The computational geometry algorithm and star topological structure model used have higher accuracy and running speed when analyzing the spatial position relationship of steel bar node and identifying leakage binding node.The application greatly improves the construction acceptance efficiency of reinforced concrete warehouse, standardizes the acceptance work flow, improves the informatization and intelligentization in civil construction engineering field.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of quality detection, and particularly relates to a method for identifying missed binding of a steel bar node and a method for evaluating the binding quality based on node spatial analysis. BACKGROUND

[0002] Reinforced concrete structure is the most widely used building structure in water conservancy and hydropower engineering, and concrete pouring preparation is the preparation work of concrete construction. Steel bar binding is the main construction activity of the preparation. In order to ensure the safety and stability of the steel skeleton during the construction process of erecting, pouring and vibrating, and to make the formed reinforced concrete structure meet the design shape and stress requirements, it is necessary to ensure the binding quality of the steel bar net. According to the structure design, the steel bar net pattern and binding requirements in the preparation construction are various, and the acceptance standards of the steel bar net binding quality are not unique.

[0003] The current acceptance of the steel bar net binding quality in engineering application is mainly based on the on-site visual inspection by quality inspection personnel. The acceptance result is highly dependent on the experience level of the quality inspection personnel, and it is time-consuming and laborious. The acceptance data is mainly recorded in paper, which is not convenient for subsequent data storage and retrieval. In order to overcome the defects of manual detection, some existing technologies provide a method based on machine vision recognition, which first identifies and analyzes the geometric features of the steel bar segments and steel bar nodes between the steel bar nets, and then calculates the qualified rate index by comparing the standard binding scheme of the steel bar net. However, this method needs to process the steel bar segments, so there is still a problem of low efficiency. SUMMARY

[0004] The present application provides a method for identifying missed binding of a steel bar node and a method for evaluating the binding quality based on node spatial analysis, to solve or at least partially solve the technical problem of low calculation efficiency in the prior art.

[0005] In order to solve the above technical problems, the first aspect of the present application provides a method for identifying missed binding of a steel bar node based on node spatial analysis, comprising:

[0006] S1: using a pre-trained target detection model to identify the steel bar net binding effect picture of the engineering construction site to be identified, identifying the contained steel bar nodes, and obtaining the coordinate data and node binding state of the steel bar nodes;

[0007] S2: according to the binding requirements, using a geometric calculation method to aggregate and calculate the coordinate data of the steel bar nodes obtained in step S1, to obtain the spatial position relationship of the steel bar nodes;

[0008] S3: according to the spatial adjacent position relationship of the steel bar nodes, using a planar geometric multi-layer star topology structure model and an optimization algorithm to identify the missed binding of the steel bar nodes.

[0009] In an embodiment, step S2 comprises:

[0010] S2.1: Calculate the distance of each steel bar node to the center point of the steel bar mesh image, and select the steel bar node with the minimum distance to the center point of the steel bar mesh image as the initial node;

[0011] S2.2: Find several steel bar nodes with the closest distance to the initial node among all steel bar nodes, connect the found nodes and the initial node respectively to obtain several straight lines, find the node corresponding to the straight line with the minimum angle with the horizontal direction as the first-row node, and store the first-row node set, find the node corresponding to the straight line with the maximum angle with the horizontal direction as the first-column node, wherein the angle between the straight line and the horizontal direction takes the acute angle, the found first-row node is the node on the same horizontal steel bar as the initial node, and the horizontal steel bar is the first-row steel bar; the found first-column node is the node on the same vertical steel bar as the initial node, and the vertical steel bar is the first-column steel bar;

[0012] S2.3: According to the angle relationship between the connecting lines of adjacent nodes, determine other unknown nodes on the first-row steel bar, and integrate into the first-row node set to complete the aggregation of the first-row steel bar nodes;

[0013] S2.4: According to the angle relationship between the connecting lines of adjacent nodes, determine other unknown nodes on the first-column steel bar, and integrate into the first-column node set to complete the aggregation of the first-column steel bar nodes;

[0014] S2.5: Take all nodes on the first-row steel bar except the initial node as new initial nodes, repeat steps S2.2 and S2.4, and sequentially aggregate each column of steel bar nodes to obtain each column of steel bar node sets, and sort the aggregated column steel bars from left to right and label the column numbers;

[0015] S2.6: Take the nodes on the first-column steel bar except the initial node as new initial nodes, repeat steps S2.2 and S2.3, and sequentially aggregate each row of steel bar nodes, each row of steel bar node set, sort the aggregated row steel bars from top to bottom, and label the row numbers;

[0016] S2.7: Screen out the nodes only marked with row numbers, find the arrangement positions of these nodes in the same row of steel bar node sets, supplement the column numbers of the nodes by calculating the difference between the arrangement positions of the nodes and other known column numbers in the same row of steel bar node sets; screen out the nodes only marked with column numbers, find the arrangement positions of these nodes in the same column of steel bar node sets, and supplement the row numbers of the nodes by calculating the difference between the arrangement positions of the nodes and other known row numbers in the same column of steel bar node sets.

[0017] In an embodiment, step S3 comprises:

[0018] S3.1: Firstly, a first layer star topology model is determined, the initial node in step S2.1 is taken as a central node of the first layer star topology model, and the layer identifier of the central node in the multi-layer star topology model is set as 1; the adjacent nodes of the central node are found as star nodes of the first layer star topology model, and the layer identifier of the star nodes in the multi-layer star topology model is set as 2, wherein if the sum of the absolute value of the row number difference and the absolute value of the column number difference of two nodes is 1, the two nodes are adjacent to each other;

[0019] S3.2: All star nodes in the first layer star topology model are taken as central nodes of a second layer topology model, and the adjacent nodes of the central nodes are found as star nodes of the second layer star topology model, and the layer identifier of the star nodes in the multi-layer star topology model is set as 3;

[0020] S3.3: Step S3.2 is repeated until the number of star nodes of the newly added star topology model is zero, and a planar geometric multi-layer star topology model is obtained;

[0021] S3.4: If the steel bar node with the layer identifier of 1 is in an unbonding state, it is verified whether the steel bar node with the layer identifier of 2n is in a bonding state, and if the steel bar node with the layer identifier of 2n is in an unbonding state, the steel bar node is marked as a current unbonding node; if the steel bar node with the layer identifier of 1 is in a bonding state, it is verified whether the steel bar node with the layer identifier of 2n+1 is in a bonding state, and the steel bar node with the layer identifier of 2n+1 in an unbonding state is marked as a current unbonding node, n = 1, 2, …;

[0022] S3.5: Adjacent nodes of the steel bar nodes in the current unbonding set are found in turn, and the steel bar nodes with all adjacent nodes in a bonding state are removed from the current unbonding marking to obtain an updated current unbonding set;

[0023] S3.6: All nodes in the updated current unbonding set are checked, if there are two nodes with one common adjacent node in an unbonding state, and the two nodes each have and only have one unbonding adjacent node, the two steel bar nodes are removed from the current unbonding marking, and the common adjacent node in an unbonding state is marked as a current unbonding state to obtain final unbonding steel bar nodes as an optimal supplemental unbonding scheme.

[0024] Based on the same inventive concept, the second aspect of the present application provides a binding quality evaluation method, which is realized based on the method of the first aspect, and the method comprises: calculating a steel bar binding rate according to the identified unbonding steel bar nodes as a quality evaluation result.

[0025] Compared with the prior art, the application has the following advantages and beneficial technical effects:

[0026] The method of the application adopts computer vision recognition technology, uses a target detection model to identify the binding state and spatial geometric features of the reinforcement mesh nodes, then designs a computational geometry algorithm to analyze the spatial relationship of the reinforcement nodes and determine the spatial position relationship of the reinforcement nodes, and then uses a star-type topological structure design optimization algorithm to identify the missed binding reinforcement nodes. The method of the application only needs to identify and calculate the information of the reinforcement nodes in the reinforcement mesh, and eliminates the processing of the reinforcement segments, so the target detection and recognition speed is faster. Moreover, the computational geometry algorithm and the star-type topological structure optimization algorithm of the method of the application have higher accuracy and running speed when analyzing the spatial relationship of the reinforcement nodes and identifying the missed binding reinforcement nodes. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0028] Figure 1 is a schematic diagram of the first key technical point in the specification part of the application;

[0029] Figure 2 is a schematic diagram of the second key technical point in the specification part of the application, a multi-layer star-type topological structure;

[0030] Figure 3 is an original image of the reinforcement mesh to be stored in the warehouse for acceptance evaluation in the embodiment of the application;

[0031] Figure 4 is an effect picture of reinforcement node identification obtained in step S1 of the embodiment of the application, a dark color identification frame indicates that the node at this position is in a binding state, and a light color identification frame indicates that the node at this position is in a non-binding state;

[0032] Figure 5 is a first row node set and a first column node set obtained in steps S2.2 to S2.4 of the embodiment of the application, wherein the node marked with a “★” symbol is the initial node in step S2.1, and the direction of the black straight line segment in the figure indicates the trend of the reinforcement segment obtained by aggregation;

[0033] Figure 6 is the horizontal reinforcement trend and the vertical reinforcement trend obtained by aggregation in steps S2.5 and S2.6 of the embodiment of the application;

[0034] Figure 7is the supplementary complete transverse steel bar polymerization effect and vertical steel bar polymerization effect obtained by step S2.7 of the embodiment of the present application;

[0035] Figure 8 is the current missed binding node obtained by step S3.4 of the embodiment of the present application, and the current missed binding node position is represented by a symbol "x";

[0036] Figure 9 is the corrected current missed binding steel bar node obtained by step S3.5 of the embodiment of the present application, wherein the node on which the current missed binding mark can be removed is represented by a symbol "●", and the current missed binding steel bar node is represented by a symbol "x";

[0037] Figure 10 is the optimized current missed binding state node obtained by step S3.6 of the embodiment of the present application, wherein the current missed binding steel bar node is represented by a symbol "x", the node on which the missed binding mark needs to be added through optimization is represented by a symbol "▲", and the node on which the missed binding mark can be removed through optimization is represented by a symbol "√";

[0038] Figure 11 is the final evaluation result obtained in the embodiment of the present application, and the light color bounding box represents the missed binding node position, and the dark color bounding box represents the adjacent unbound node of each missed binding node. DETAILED DESCRIPTION

[0039] The present application discloses a missed binding steel bar node identification method and a binding quality evaluation method based on node space analysis, and aims to solve the problems of low efficiency, poor accuracy and complete dependence on manual operation in the existing quality acceptance mode during the pouring preparation stage of the reinforced concrete structure.

[0040] The method of the present application adopts a computer vision identification method combined with a computational geometry algorithm to perform target identification and space analysis on the steel bar net image. First, a target detection model is trained by using a computer vision identification algorithm, the steel bar nodes in the image are identified by using the target detection model, and the coordinate data of the steel bar nodes in the image coordinate system and the binding state of the steel bar nodes are returned. According to the construction requirements, when the binding requirement is interval binding, the nodes on the same steel bar are analyzed by using a computational geometry algorithm to obtain the spatial position relationship of all the nodes in the steel bar net. Then, the missed binding nodes of the optimal scheme are identified by using a star-type topological structure model and an optimization algorithm, and the steel bar net binding rate is calculated. When the construction requirement is full binding, the steel bar net binding rate can be directly calculated after target detection to realize intelligent evaluation of the steel bar net binding.

[0041] The method has two key technical points in realizing intelligent evaluation of the binding quality of the reinforcement mesh. One is a method for obtaining the spatial position relationship of the reinforcement node based on the reinforcement node feature information obtained through computer vision recognition and designing a computational geometry algorithm. The other is designing a planar geometry multi-layer star topology model and an optimization algorithm to identify the missed binding nodes and calculate the optimal binding rate, so that the number of nodes that need to be supplemented is the least under the condition of meeting the acceptance requirements.

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0043] Embodiment one

[0044] The embodiments of the present application provide a missed binding reinforcement node identification method based on node spatial analysis, comprising:

[0045] S1: using a pre-trained target detection model to identify the reinforcement mesh binding effect picture of the engineering construction site to be identified, identifying the contained reinforcement nodes, and obtaining the coordinate data and node binding state of the reinforcement nodes;

[0046] S2: according to the binding requirements, using a geometric calculation method to aggregate and calculate the coordinate data of the reinforcement nodes obtained in step S1, to obtain the spatial position relationship of the reinforcement nodes;

[0047] S3: according to the spatial position relationship of the reinforcement nodes, using a planar geometry multi-layer star topology model and an optimization algorithm to identify the missed binding reinforcement nodes.

[0048] Specifically, in order to obtain the trained target detection model, a large number of engineering construction site reinforcement mesh binding effect pictures need to be collected, the effect pictures are labeled and extracted for target features, and a YOLO open source structure model is used to train the target detection model of oneself, to obtain a trained model with a target recognition rate better than 0.9, taking the binding reinforcement nodes and the unbonding reinforcement nodes as the characteristic targets. Different target detection models have different focuses in model accuracy, training time, and recognition speed, and other appropriate target detection models can be flexibly selected in other implementation manners.

[0049] In engineering construction, steel mesh binding requires full binding and interval binding. Full binding means that all steel mesh nodes are bound, and the unbound nodes in the image are all missed binding nodes. Interval binding, also known as skip binding, means that the nodes on the same steel bar can be bound at intervals, and when two or more nodes on the same steel bar are not bound, it is not satisfied. Therefore, when the construction requirement is interval binding, the spatial adjacent position relationship of the nodes in the steel mesh needs to be obtained first, and then it is judged which nodes need to be supplemented to make the binding quality meet the requirements. When the number of steel nodes that need to be supplemented is the least, it is the optimal scheme, and the binding rate at this time is the optimal binding rate.

[0050] See Figure 3 The original image of the steel mesh in the warehouse to be accepted and evaluated in the embodiment of the application is the steel mesh binding effect image to be identified in the engineering construction site.

[0051] The geometric calculation algorithm and the planar geometric multi-layer star topology model used in the application are mainly for the case where the binding requirement is interval binding. When the construction requirement is full binding, the steel mesh binding rate can be calculated directly after target detection.

[0052] In step S2, the steel node aggregation means that the nodes on the same steel bar (the same row or column) are identified, sorted and marked according to the steel direction. Step S3 is to construct a multi-layer star topology model on the basis of the spatial position relationship of the steel nodes obtained in step S2, so as to identify the optimal missed binding node.

[0053] In one embodiment, step S2 includes:

[0054] S2.1: Calculate the distance of each steel node to the center point of the steel mesh image, and select the steel node with the smallest distance to the center point of the steel mesh image as the initial node;

[0055] S2.2: Find several steel nodes closest to the initial node among all steel nodes, connect these nodes with the initial node respectively to obtain several straight lines, find the node corresponding to the straight line with the smallest angle with the horizontal direction as the first row node, and store it in the first row node set, find the node corresponding to the straight line with the largest angle with the horizontal direction as the first column node, wherein the angle between the straight line and the horizontal direction is an acute angle, the found first row node is the node on the same horizontal steel bar as the initial node, and the horizontal steel bar is the first row steel bar; the found first column node is the node on the same vertical steel bar as the initial node, and the vertical steel bar is the first column steel bar;

[0056] S2.3: According to the angle relationship between the adjacent node connection lines, other unknown nodes on the first row of steel bars are determined and are classified into the first row node set, and the first row of steel bar nodes are completely aggregated;

[0057] S2.4: According to the angle relationship between the adjacent node connection lines, other unknown nodes on the first column of steel bars are determined and are classified into the first column node set, and the first column of steel bar nodes are completely aggregated;

[0058] S2.5: Taking all the nodes on the first row of steel bars except the initial node as new initial nodes, steps S2.2 and S2.4 are repeated, and each column of steel bar nodes is aggregated in turn to obtain each column of steel bar node set, and the aggregated all column steel bars are sorted from left to right, and the column number is labeled;

[0059] S2.6: Taking the nodes on the first column of steel bars except the initial node as new initial nodes, steps S2.2 and S2.3 are repeated, and each row of steel bar nodes is aggregated in turn, and each row of steel bar node set is obtained, and the aggregated all row steel bars are sorted from top to bottom, and the row number is labeled;

[0060] S2.7: Screening out the nodes only marked with the row number, finding the arrangement position of these nodes in the same row of steel bar node set, and supplementing the column number of the node by calculating the difference value of the arrangement position of the node with other known column number nodes in the same row of steel bar node set; screening out the nodes only marked with the column number, finding the arrangement position of these nodes in the same column of steel bar node set, and supplementing the row number of the node by calculating the difference value of the arrangement position of the node with other known row number nodes in the same column of steel bar node set.

[0061] The key technical points of the method of the application are as follows: in step S2, the relevant knowledge of the computational geometry part is mainly used, including the included angle formula of two vectors in plane geometry, The specific algorithm is described by taking Figure 1 as an example. It is assumed that Figure 1 The center coordinates of each node in the steel bar mesh image are O(x0, y0), A(x1, y1), B(x2, y2), and the node vectors are Let The value range of is [0°, 180°]. If the value of is in the interval range of [0°, 20°] or [160°, 180°], it is considered that the three steel bar nodes are on the same steel bar segment, such as the positions of AOB three points or A′OB′ three points in the figure. If the value of is in the interval range of [70°, 110°], it is considered that the three nodes are located on two intersecting steel bar segments, such as the positions of A′OB three points or AOB′ three points in the figure. Figure 1 Figure 1

[0062] ​​​In the implementation process, S2.2 finds several steel bar nodes closest to the initial node in all steel bar nodes, and the number of the several nodes can be selected according to actual conditions, for example, 6, 8, etc. The first row and the first column of steel bar nodes can be found through step S2.2.

[0063] According to geometric knowledge, if three adjacent nodes are known, a straight line can be drawn from the middle node to the other two nodes, and if the included angle of the straight line is within a preset angle range, it is considered that the three nodes are on the same steel bar. Thus, if two adjacent nodes on a steel bar are known, all nodes on the steel bar can be aggregated.

[0064] Since the aggregation of nodes on the same steel bar is sorted according to the trend of the steel bar, for the same horizontal steel bar, the column mark of a node thereon is known, and the column marks of other nodes can be obtained according to the bit sequence difference with other nodes. Similarly, for the same vertical steel bar, the row mark of a node thereon is known, and the row marks of other nodes can be obtained according to the bit sequence difference. The nodes only marked with row numbers or column numbers are filtered out, and the row and column numbers of the nodes are completed by the above method.

[0065] In an embodiment, step S3 includes:

[0066] S3.1: First, determine the first layer star topology model, take the initial node in step S2.1 as the central node of the first layer star topology model, and let its layer identification δ in the multi-layer star topology model be 1. Find the adjacent nodes of the central node as the star nodes of the first layer star topology model, and let their layer identification δ in the multi-layer star topology model be 2. If the sum of the absolute value of the row number difference and the absolute value of the column number difference of two nodes is 1, the two nodes are adjacent to each other;

[0067] S3.2: Take all star nodes in the first layer star topology model as the central nodes of the second layer topology model, and find the adjacent nodes of the central nodes as the star nodes of the second layer star topology model, and let their layer identification δ in the multi-layer star topology model be 3;

[0068] S3.3: Loop step S3.2 until the number of star nodes of the newly added star topology model is zero to end, and obtain the planar geometric multi-layer star topology model;

[0069] S3.4: If the steel bar node with layer identification δ = 1 is in unbonded state, verify whether the steel bar node with layer identification δ = 2n is in bonded state, if in unbonded state, mark it as the current unbonded node; if the steel bar node with layer identification δ = 1 is in bonded state, verify whether the steel bar node with layer identification δ = 2n + 1 is in bonded state, mark the steel bar node in unbonded state as the current unbonded node, all the current unbonded nodes constitute the current unbonded set, n = 1, 2, …;

[0070] S3.5: Find the adjacent nodes of the steel bar nodes in the current unbonded set in turn, remove the steel bar nodes with all adjacent nodes in bonded state from the current unbonded mark to obtain the updated current unbonded set;

[0071] S3.6: Check all the nodes in the updated current unbonded set, if there are two nodes with one common adjacent node in unbonded state, and both of the two nodes have and only have one unbonded adjacent node, remove the two steel bar nodes from the current unbonded mark, and mark the common unbonded adjacent node as the current unbonded state to obtain the final unbonded steel bar node as the optimal supplemental unbonding scheme.

[0072] The schematic diagram of the star topology structure is shown in Figure 2 The layer identification δ = 1 of the central node of the first layer topology structure in the figure, in the steel bar mesh image, each star topology structure unit is a "cross" type, as shown by the dashed line frame in the figure. From the first layer topology structure unit, the layer identification layer increases layer by layer, the number in each node frame in the figure represents the layer identification of the node in the entire star topology structure.

[0073] The plane geometric multi-layer star topology structure model can be constructed through steps S3.1-S3.3, and steps S3.4-S3.6 are the identification process of the unbonded node using the model, that is, the identification process of the unbonded node using the optimization algorithm.

[0074] In which, the unbonding state of each steel bar node is known, which is obtained by the target identification model in step S1. The current unbonded node can be preliminarily screened out through step S3.4.

[0075] For an unbonded steel bar node, if all its adjacent steel bar nodes are in bonded state, no matter where the node is located in the topology structure, the node does not need to be bonded. Therefore, for the current unbonded node in step S3.4, the steel bar nodes with all adjacent nodes in bonded state do not need to be bonded, and the current unbonded mark of these nodes is removed.

[0076] For three adjacent steel bar nodes in a continuous unbonded cross shape, it is considered that the scheme of binding one steel bar node at the intersection is better than the scheme of binding the other two steel bar nodes. Replacing the binding of the other two steel bar nodes with the binding of one steel bar node at the intersection is called one-group optimized binding. If there is a situation that can be optimized binding in the current unbonded node, the replacement is carried out, and the two steel bar nodes that are replaced are removed from the current unbonded node, and the one steel bar node that is replaced is marked as the current unbonded node. At this point, all the nodes marked as the current unbonded node are all the unbonded steel bar nodes in the optimal scheme.

[0077] Embodiment two

[0078] Based on the same inventive concept, the embodiment provides a binding quality evaluation method, which is realized based on the method of embodiment one, and the method comprises: calculating a steel bar binding rate according to the identified unbonded steel bar nodes, as a quality evaluation result.

[0079] In the specific implementation process, the steel bar mesh binding rate K=(S-L) / S*100% (S is the total number of steel bar nodes, and L is the number of unbonded steel bar nodes). When the evaluation requirement is full binding, the value of L is the number of unbonded nodes in the identification result in step S1 in embodiment one; when the evaluation requirement is interval binding, the value of L is the number of unbonded steel bar nodes in step S3 in embodiment one.

[0080] In the specific implementation process, if the visualization output of the evaluation result and the evaluation process image is to be realized, a GUI file suitable for the output content needs to be designed in advance, and the pyside2 software selected in this embodiment can also be flexibly selected as other suitable graphical user interface making tools.

[0081] The quality evaluation method proposed in the embodiment will be described in detail below through specific embodiments.

[0082] Step S101: The target detection model trained in advance is used to identify the steel bar nodes contained in the picture, and the coordinate data and node binding state of the steel bar nodes in the image coordinate system are returned. The node identification effect is as shown in Figure 4 .

[0083] Step S201: The nodes on the same steel bar are aggregated by using a computational geometry algorithm to obtain the spatial position relationship of the steel bar mesh nodes. The detailed decomposition steps of this step are as follows T1 to T7:

[0084] Step T1: Determine the initial node position, such as the node marked with “★” in Figure 5 .

[0085] Step T2: Find two nodes adjacent to the initial node on the first row of steel bars and the first column of steel bars, respectively.

[0086] Step T3: aggregate the first row of reinforcement nodes, as Figure 5 .

[0087] Step T4: aggregate the first column of reinforcement nodes, as Figure 5 .

[0088] Step T5: aggregate each vertical reinforcement node, as Figure 6 .

[0089] Step T6: aggregate each horizontal reinforcement node, as Figure 6 .

[0090] Step T7: supplement the row and column markers of all nodes, as Figure 7 .

[0091] Step S301: identify the optimal solution of the missing binding reinforcement nodes using a star topology optimization algorithm. The detailed steps of this step are as follows P1 to P7:

[0092] Step P1: first determine the first layer star topology model, take the initial node in step T1 as the central node of the first layer star topology model, as Figure 8 the "★" mark in the middle, let its layer identification δ = 1 in the multi-layer star topology model, find the adjacent nodes of the central node (if the sum of the absolute value of the row number difference and the absolute value of the column number difference of two nodes is 1, then the two nodes are adjacent to each other), as the star nodes of the first layer star topology model, let its layer identification δ = 2 in the multi-layer star topology model.

[0093] Step P2: take all star nodes in the first layer star topology model as the central nodes of the second layer topology model, find the adjacent nodes of the central nodes, as the star nodes of the second layer star topology model, let its layer identification δ = 3 in the multi-layer star topology model.

[0094] Step P3: repeat step P2 until the number of star nodes of the new layer star topology model is zero.

[0095] Step P4: if the reinforcement node with layer identification δ = 1 is in the unbound state (i.e. in the unbound state), verify whether the reinforcement node with layer identification δ = 2n (n = 1, 2…) is in the bound state, and select the reinforcement node in the unbound state as the current missing binding state; if the reinforcement node with layer identification δ = 1 is in the bound state, verify whether the reinforcement node with layer identification δ = 2n + 1 (n = 1, 2…) is in the bound state, and select the reinforcement node in the unbound state as the current missing binding state, as Figure 8 the "×" mark in the middle.

[0096] Step P5: for the current missed binding node identified in step P4, the steel bar node with all adjacent nodes in the binding state does not need to be bound, and the missed binding identifier is removed. As shown in Figure 9 , 'X' represents the current missed binding node, and '●' represents the node that can remove the missed binding mark.

[0097] Step P6: optimize the current missed binding node, as shown in Figure 10 , 'X' represents the current missed binding steel bar node, '▲' represents the node that needs to add a missed binding mark through optimization, and '√' represents the node that can remove the missed binding mark through optimization.

[0098] Step S401: calculate the binding rate. The steel mesh binding rate = 193 / 204 x 100% = 94.6%, the optimal number of missed binding nodes is 11, and the missed binding node position is as shown in Figure 11 .

[0099] The method of the present application adopts a computer vision recognition method to extract the binding state and geometric features of the steel mesh nodes; then a computational geometry algorithm is designed to analyze the spatial relationship of the steel nodes and determine the adjacent position relationship of each steel node; then through a planar geometric star topological structure model and an optimization algorithm, the missed binding steel nodes are identified, and the qualified binding rate is calculated. The method of the present application only needs to identify and calculate the information of the steel nodes in the steel mesh, and saves the processing of the steel segments, and the target detection and recognition speed is faster; the computational geometry algorithm and the star topological structure model of the present application have higher accuracy and running speed in analyzing the spatial position relationship of the steel nodes and identifying the missed binding nodes. The method of the present application greatly improves the construction acceptance efficiency of the reinforced concrete warehouse, standardizes the work flow of the acceptance, and improves the informatization and intelligentization in the field of civil construction engineering.

[0100] Although the preferred embodiments of the present application have been described, those skilled in the art can make further changes and modifications to the embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including all changes and modifications falling within the scope of the present application.

[0101] Obviously, those skilled in the art can make various modifications and variations to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method for identifying a lap splice reinforcement joint based on nodal space analysis, characterized by, Comprise: S1: using a pre-trained target detection model to identify the steel bar binding effect picture of the construction site to be identified, identifying the contained steel bar nodes, and obtaining the coordinate data and node binding state of the steel bar nodes; S2: according to the binding requirements, the coordinate data of the steel bar nodes obtained in step S1 is aggregated and calculated by using the geometric calculation method, and the spatial position relationship of the steel bar nodes is obtained; S3: according to the spatial adjacent position relationship of the steel bar nodes, using a planar geometric multi-layer star topology structure model and an optimization algorithm, the missed binding steel bar nodes are identified; Wherein, step S3 includes: S3.1: first determine the first layer star topology structure model, take the initial node as the central node of the first layer star topology structure model, let its layer identification δ=1 in the multi-layer star topology structure model, find out the adjacent nodes of the central node as the star nodes of the first layer star topology structure model, let its layer identification δ=2 in the multi-layer star topology structure model, wherein, if the sum of the absolute value of the row number difference and the absolute value of the column number difference of two nodes is 1, then the two nodes are adjacent nodes, wherein, the initial node is the steel bar node with the minimum distance to the center point of the steel bar mesh image calculated in S2; S3.2: take all star nodes in the first layer star topology structure model as the central nodes of the second layer topology structure model, find out the adjacent nodes of the central nodes as the star nodes of the second layer star topology structure model, let its layer identification δ=3 in the multi-layer star topology structure model; S3.3: loop step S3.2 until the number of star nodes of the newly added star topology structure model is zero to end, and obtain the planar geometric multi-layer star topology structure model; S3.4: if the steel bar node with layer identification δ=1 is in the unbundling state, verify whether the steel bar node with layer identification δ=2n is in the binding state, if it is in the unbundling state, mark it as the current missed binding node; if the steel bar node with layer identification δ=1 is in the binding state, verify whether the steel bar node with layer identification δ=2n+1 is in the binding state, mark the steel bar node with the unbundling state as the current missed binding node, all current missed binding nodes constitute the current missed binding set, n=1, 2…; S3.5: find out the adjacent nodes of the steel bar nodes in the current missed binding set in turn, remove the steel bar nodes with all adjacent nodes in the binding state from the current missed binding mark to obtain the updated current missed binding set; S3.6: check all nodes in the updated current missed binding set, if there are two nodes with a common adjacent node in the unbundling state, and both of them have and only have one unbundling adjacent node, then remove the two steel bar nodes from the current missed binding mark, and mark the common unbundling adjacent node as the current missed binding state to obtain the final missed binding steel bar node as the optimal missed binding scheme.

2. The method for identifying a lap splice reinforcement joint based on nodal space analysis according to claim 1, wherein, Step S2 includes: S2.1: calculate the distance from each steel bar node to the center point of the steel bar mesh image, select the steel bar node with the minimum distance to the center point of the steel bar mesh image as the initial node; S2.2: Find several steel bar nodes closest to the initial node in all steel bar nodes, connect the found nodes with the initial node respectively, get several straight lines, find the node corresponding to the straight line with the smallest angle with the horizontal direction as the first row node from the several straight lines, store the first row node set, find the node corresponding to the straight line with the largest angle with the horizontal direction as the first column node, wherein the angle between the straight line and the horizontal direction takes the acute angle, the found first row node is the node on the same horizontal steel bar as the initial node, and the horizontal steel bar is the first row steel bar; the found first column node is the node on the same vertical steel bar as the initial node, and the vertical steel bar is the first column steel bar; S2.3: According to the angle relationship between the adjacent node connecting lines, determine other unknown nodes on the first row steel bar, and integrate into the first row node set to complete the aggregation of the first row steel bar node; S2.4: According to the angle relationship between the adjacent node connecting lines, determine other unknown nodes on the first column steel bar, and integrate into the first column node set to complete the aggregation of the first column steel bar node; S2.5: Take all nodes on the first row steel bar except the initial node as a new initial node, repeat steps S2.2 and S2.4, and aggregate each column steel bar node in turn to obtain each column steel bar node set. After aggregation, all column steel bars are sorted from left to right, and the column number is labeled; S2.6: Take the nodes on the first column steel bar except the initial node as a new initial node, repeat steps S2.2 and S2.3, and aggregate each row steel bar node in turn, each row steel bar node set, and after aggregation, all row steel bars are sorted from top to bottom, and the row number is labeled; S2.7: Screen out the nodes only marked with the row number, find the arrangement position of these nodes in the same row steel bar node set, supplement the column number of the node by calculating the difference value of the arrangement position of the node with other known column numbers in the same row steel bar node set; screen out the nodes only marked with the column number, find the arrangement position of these nodes in the same column steel bar node set, and supplement the row number of the node by calculating the difference value of the arrangement position of the node with other known row numbers in the same column steel bar node set.

3. A method for evaluating the quality of a binding, implemented on the basis of the method according to any one of claims 1 to 2, characterized in that, The method comprises: calculating the steel bar binding rate according to the identified missed binding steel bar nodes, as a quality evaluation result.

Images