Pile foundation coordinate data management method and system

Abstract

The invention provides a pile foundation coordinate data management method and system, and the method comprises the steps: collecting a pile position coordinate table image, carrying out the preprocessing of the image, and extracting the image features; performing text recognition on the image, extracting text information, performing structured processing, and generating preliminary coordinate data; correcting the initial coordinate data; outputting the corrected coordinate data in a structured format to form a data table of stake numbers, X coordinates and Y coordinates; analyzing a pile foundation design drawing, extracting pile numbers and corresponding X and Y coordinates to generate a structured database, and collecting real-time three-dimensional coordinates and attitude data in the pile foundation construction process; establishing an association mapping relation between the pile number and the design coordinate and the real-time coordinate, generating a visual symbol unit through a two-dimensional symbolization examination module, and presenting the design position, the actual position and the inclination state of the pile foundation; and performing manual verification based on the deviation state of the symbol unit, and storing qualified data into a system database after final confirmation.

Description

Technical Field

[0001] This invention relates to the field of digital construction, specifically to a method and system for managing pile foundation coordinate data. Background Technology

[0002] In the field of building construction, pile foundations, as a core component of the foundation structure, are directly related to construction quality and structural safety through the accurate management of their coordinate data. The full-process management of pile location coordinate data covers multiple stages, from extraction from design drawings and data collection during construction to data verification and storage, and is a key support for the digital management and control of engineering projects.

[0003] Current technologies for managing pile location coordinate data have significant limitations. In the data extraction stage, traditional methods rely heavily on manual reading of coordinate information from design drawings and manual input into the system. This is not only inefficient but also prone to errors due to human intervention. While some technologies have attempted to incorporate Optical Character Recognition (OCR) algorithms to assist extraction, the inconsistent quality of the drawings and the lack of targeted preprocessing mean that text recognition accuracy is insufficient to meet engineering precision requirements, especially when distinguishing key information such as pile number, X-coordinate, and Y-coordinate, leading to structural confusion. The data processing stage also suffers from significant shortcomings. After initial extraction, pile location coordinate data often suffers from non-standard formatting and may contain errors due to oversight. Furthermore, data verification relies heavily on comparing purely numerical tables, failing to visually represent the three-dimensional pose of the pile foundation. Manual verification of large amounts of data can easily lead to misjudgments due to visual fatigue, resulting in excessive data entering the system and creating potential problems for subsequent construction. Therefore, an optimized method for managing pile location coordinate data is urgently needed to fill these technological gaps. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems mentioned above and to propose a method for managing pile foundation coordinate data, comprising the following steps:

[0005] S1. Collect images of the pile location coordinate table, preprocess the images, and extract image features;

[0006] S2. Use optical character recognition algorithms to perform text recognition on the preprocessed image, extract text information and perform structured processing to generate preliminary coordinate data;

[0007] S3. Correct the preliminary coordinate data, including removing extra spaces and correction symbols, and perform coding verification and completion of the station number sequence;

[0008] S4. Output the corrected coordinate data in a structured format to form a data table of station number, X coordinate, and Y coordinate;

[0009] S5. Analyze the pile foundation design drawings to extract the pile number and corresponding X and Y coordinates to generate a structured database, and at the same time collect real-time three-dimensional coordinates and attitude data during the pile foundation construction process;

[0010] S6. Establish the association mapping relationship between the station number and the design coordinates and real-time coordinates, and generate visual symbol units through the two-dimensional symbol review module to present the design location, actual location and tilt status of the pile foundation;

[0011] S7. Manually verify the deviation status based on the symbol unit, and store qualified data in the system database after final confirmation.

[0012] In the preferred embodiment, the symbol units generated by the two-dimensional symbolization review module in step S6 include:

[0013] Design reference symbol: represented by a crosshair target, with the center corresponding to the pile foundation design coordinates, and the length of the crosshair matching the allowable deviation threshold;

[0014] Actual location symbol: represented by a dot, with the center corresponding to the actual coordinates of the pile foundation, and the color is coded as green, yellow or red according to the deviation level;

[0015] Inclination symbol: indicated by an arrow, the direction of which corresponds to the horizontal projection direction of the pile foundation inclination, and the length reflects the degree of inclination proportionally.

[0016] In the preferred embodiment, the process of analyzing the pile foundation design drawings in step S5 includes:

[0017] Identify the coordinate table area in the drawing;

[0018] Extract station number, X coordinate, and Y coordinate information;

[0019] Verify data integrity and complete any missing information;

[0020] Output structured data.

[0021] In the preferred embodiment, the verification process in step S7 includes:

[0022] Automatically identify color and size deviations of symbol units;

[0023] For units with green symbols, confirm and enter the information directly;

[0024] Warnings are issued for units marked with yellow or red symbols, prompting manual review and correction.

[0025] A system employing a pile foundation coordinate data management method includes:

[0026] The coordinate analysis unit is used to parse design drawings and generate structured coordinate data;

[0027] The data acquisition unit is used to collect real-time coordinate and attitude data during the construction process;

[0028] The associated mapping unit establishes the relationship between the station number and the design coordinates and real-time coordinates;

[0029] Two-dimensional symbolic review unit, generating a visual unit containing design reference symbols, actual position symbols, and tilt status symbols;

[0030] The data storage unit stores the coordinate data after verification.

[0031] In the preferred embodiment, the two-dimensional symbolization review unit includes:

[0032] The symbol generation sub-unit generates corresponding symbols based on the design and real-time coordinates.

[0033] The color-coded subunit assigns colors to symbols based on the deviation level;

[0034] The interactive display sub-unit provides a visual interface for symbols and manual editing functions.

[0035] In the preferred embodiment, the coordinate analysis unit includes:

[0036] The drawing recognition module locates the coordinate table area;

[0037] The information extraction module extracts station number and coordinate data;

[0038] The data validation module completes missing data and corrects its format.

[0039] In the preferred embodiment, the data acquisition unit includes:

[0040] The three-dimensional positioning module obtains the real-time location of the pile foundation;

[0041] The attitude detection module detects the tilt angle and direction of the pile foundation.

[0042] The data preprocessing module performs noise reduction and integration on the collected data.

[0043] In the preferred embodiment, the interactive display sub-unit includes:

[0044] Symbol scaling and panning functions;

[0045] Deviation value display function;

[0046] Batch filtering of abnormal symbols;

[0047] Coordinate editing and symbol update functions.

[0048] In the preferred embodiment, a construction trajectory analysis unit is also included, which generates a single pile construction path report based on pile foundation coordinate data, calculates construction parameters such as drilling speed and positioning adjustment frequency by combining timestamp data, identifies abnormal trajectories, and marks risk periods to assist in optimizing construction technology.

[0049] In the preferred embodiment, a spatial interference detection unit is also included, which calculates the spatial distance and relative tilt angle of adjacent pile foundations through coordinate data. When the measured distance is less than the minimum design clearance or the relative tilt angle exceeds the threshold, a collision risk warning is automatically triggered and adjustment suggestions are output.

[0050] The beneficial effects of this invention are as follows:

[0051] (1) The pile coordinate information is automatically extracted through image preprocessing and OCR technology, avoiding the inefficiency and errors of traditional manual data entry. The initially extracted data is automatically corrected and the pile number sequence is verified and completed to ensure the integrity and correctness of the data.

[0052] (2) Using the two-dimensional symbolic review module, the abstract design coordinates, actual construction coordinates and pile foundation tilt status are transformed into intuitive graphic symbols. The graphic display makes the position deviation and tilt degree clear at a glance.

[0053] (3) Based on symbol units, the deviation level is automatically identified and distinguished by three colors, which allows for quick and automatic confirmation of qualified data, accurate location of data exceeding the standard and triggering manual review, greatly reducing the workload of manually checking a large number of data tables and the misjudgment caused by visual fatigue.

[0054] (4) Real-time acquisition of three-dimensional coordinates and attitude data during construction, and dynamic correlation and mapping with design coordinates. This enables timely detection of construction deviations and triggers verification and correction processes before data entry or during construction, preventing unqualified pile foundations from entering the next stage.

[0055] (5) Generate construction path reports based on coordinate data, calculate parameters such as drilling speed and positioning adjustment frequency, identify abnormal trajectories and mark risks, assist in process optimization, calculate the spacing and relative inclination of adjacent pile foundations, and automatically trigger collision warnings and provide adjustment suggestions when the spacing is too small or the inclination is too large. Attached Figure Description

[0056] Figure 1 It is a flowchart of the method.

[0057] Figure 2 This is a schematic diagram of a two-dimensional symbol unit.

[0058] Figure 3 This is a preview image of the pile foundation coordinates from Example 3.

[0059] Figure 4This is a screenshot of the pile foundation coordinate data confirmation interface in Example 3. Detailed Implementation

[0060] Example 1

[0061] like Figures 1-2 As shown, a method for managing pile foundation coordinate data includes the following steps:

[0062] S1. Collect the image of the pile location coordinate table, and perform preprocessing operations such as grayscale conversion, binarization, noise removal, smoothing and contrast adjustment on the image to extract image features.

[0063] S2. Use optical character recognition algorithms to perform text recognition on the preprocessed image, extract text information and perform structured processing to generate preliminary coordinate data;

[0064] S3. Correct the preliminary coordinate data, including removing extra spaces and correction symbols, and perform coding verification and completion of the station number sequence to ensure the integrity and correctness of the data;

[0065] S4. Output the corrected coordinate data in a structured format to form a data table of station number, X coordinate, and Y coordinate;

[0066] S5. Analyze the pile foundation design drawings to extract the pile number and corresponding X and Y coordinates to generate a structured database, and at the same time collect real-time three-dimensional coordinates and attitude data during the pile foundation construction process;

[0067] S6. Establish the association mapping relationship between the station number and the design coordinates and real-time coordinates, and generate visual symbol units through the two-dimensional symbol review module to present the design location, actual location and tilt status of the pile foundation;

[0068] S7. Manually verify the deviation status based on the symbol unit, and store qualified data in the system database after final confirmation.

[0069] Preferably, the symbol units generated by the two-dimensional symbolization review module in step S6 include:

[0070] Design reference symbol: represented by a crosshair target, with the center corresponding to the pile foundation design coordinates, and the length of the crosshair matching the allowable deviation threshold;

[0071] Actual location symbol: represented by a dot, with the center corresponding to the actual coordinates of the pile foundation, and the color is coded as green, yellow or red according to the deviation level;

[0072] Inclination symbol: indicated by an arrow, the direction of which corresponds to the horizontal projection direction of the pile foundation inclination, and the length reflects the degree of inclination proportionally.

[0073] Preferably, the process of analyzing pile foundation design drawings in step S5 includes:

[0074] Identify the coordinate table area in the drawing;

[0075] Extract station number, X coordinate, and Y coordinate information;

[0076] Verify data integrity and complete any missing information;

[0077] Output structured data.

[0078] Preferably, the verification process in step S7 includes:

[0079] Automatically identify color and size deviations of symbol units;

[0080] For units with green symbols, confirm and enter the information directly;

[0081] Warnings are issued for units marked with yellow or red symbols, prompting manual review and correction.

[0082] A system employing a pile foundation coordinate data management method includes:

[0083] The image acquisition module is used to acquire images of the pile location coordinate table;

[0084] The algorithm processing module is used to automatically extract coordinate data by performing text recognition on the image;

[0085] The data storage module is used to store the extracted coordinate data;

[0086] The user interface module is used to display pile location coordinate information and receive user operations.

[0087] The coordinate analysis unit is used to parse design drawings and generate structured coordinate data;

[0088] The data acquisition unit is used to collect real-time coordinate and attitude data during the construction process;

[0089] The associated mapping unit establishes the relationship between the station number and the design coordinates and real-time coordinates;

[0090] Two-dimensional symbolic review unit, generating a visual unit containing design reference symbols, actual position symbols, and tilt status symbols;

[0091] The data storage unit stores the coordinate data after verification.

[0092] In the preferred embodiment, the two-dimensional symbolization review unit includes:

[0093] The symbol generation sub-unit generates corresponding symbols based on the design and real-time coordinates.

[0094] The color-coded subunit assigns colors to symbols based on the deviation level;

[0095] The interactive display sub-unit provides a visual interface for symbols and manual editing functions.

[0096] Preferably, the coordinate analysis unit includes:

[0097] The drawing recognition module locates the coordinate table area;

[0098] The information extraction module extracts station number and coordinate data;

[0099] The data validation module completes missing data and corrects its format.

[0100] Preferably, the data acquisition unit includes:

[0101] The three-dimensional positioning module obtains the real-time location of the pile foundation;

[0102] The attitude detection module detects the tilt angle and direction of the pile foundation.

[0103] The data preprocessing module performs noise reduction and integration on the collected data.

[0104] Preferably, the interactive display sub-unit includes:

[0105] Symbol scaling and panning functions;

[0106] Deviation value display function;

[0107] Batch filtering of abnormal symbols;

[0108] Coordinate editing and symbol update functions.

[0109] Preferably, it also includes a construction trajectory analysis unit, which generates a single pile construction path report based on pile foundation coordinate data, calculates construction parameters such as drilling speed and positioning adjustment frequency by combining timestamp data, identifies abnormal trajectories and marks risk periods, and assists in optimizing construction technology.

[0110] The abnormal trajectory includes abrupt shifts and repeated adjustments.

[0111] Preferably, it also includes a spatial interference detection unit, which calculates the spatial distance and relative tilt angle of adjacent pile foundations through coordinate data. When the measured distance is less than the minimum design clearance or the relative tilt angle exceeds the threshold, it automatically triggers a collision risk warning and outputs adjustment suggestions.

[0112] Preferably, the spatial interference detection unit calculates the spatial distance and relative tilt angle between adjacent pile foundations using coordinate data, wherein the calculation method for the distance between adjacent pile foundations includes:

[0113] First, obtain the three-dimensional coordinates of the top and bottom of two adjacent piles (pile A and pile B): the top coordinates of pile A are (Xa1, Ya1, Za1) and the bottom coordinates are (Xa2, Ya2, Za2); the top coordinates of pile B are (Xb1, Yb1, Zb1) and the bottom coordinates are (Xb2, Yb2, Zb2).

[0114] Establish the spatial axis parametric equations for the two pile foundations: the axis of pile A is (Xa1+t·ΔXa, Ya1+t·ΔYa, Za1+t·ΔZa), where ΔXa=Xa2-Xa1, ΔYa=Ya2-Ya1, ΔZa=Za2-Za1, t∈[0,1]; the axis of pile B is (Xb1+s·ΔXb, Yb1+s·ΔYb, Zb1+s·ΔZb), where ΔXb=Xb2-Xb1, ΔYb=Yb2-Yb1, ΔZb=Zb2-Zb1, s∈[0,1];

[0115] Calculate the shortest spatial distance D between two axes: using the vector cross product formula, i.e.:

[0116]

[0117] Where vector P1P2 is the vector from the top of pile A to the top of pile B (Xb1-Xa1, Yb1-Ya1, Zb1-Za1), and direction vector A is the cross product of the direction vectors of the axis of pile A and the axis of pile B.

[0118] Determine the net spacing S between adjacent pile foundations: S = D - (Ra + Rb), where Ra and Rb are the design radii of pile A and pile B, respectively;

[0119] When the net clearance S is less than the minimum design net clearance or the relative tilt angle exceeds 5°, a collision risk warning will be automatically triggered, and adjustment suggestions will be output, including: optimizing the construction sequence and correcting the tilt direction.

[0120] Example 2

[0121] A method and system for managing pile foundation coordinates based on two-dimensional symbolic review, the method comprising the following steps:

[0122] A1. Analysis of pile foundation design drawings:

[0123] Image recognition technology is used to locate the coordinate table in the design drawings, extract the station number, X coordinate, and Y coordinate information, verify the data integrity (such as filling in missing station numbers), and generate a structured database (format example: [{"station number":"Z1","X":100.0,"Y":200.0},...]).

[0124] A2. Collect real-time construction data:

[0125] Using positioning equipment and attitude sensors, the three-dimensional coordinates (X real, Y real, Z real) and tilt parameters (tilt angle, tilt direction) of the pile foundation are collected in real time during the pile foundation construction process, with a data sampling frequency ≥10Hz.

[0126] A3. Establish association mapping:

[0127] By binding the design coordinates (X-design, Y-design) with the real-time coordinates (X-real, Y-real) through the station number, a correspondence of "station number - design - real-time" is formed.

[0128] A4. Two-dimensional symbolic review:

[0129] The two-dimensional symbolization review module is invoked to generate symbolic units for each pile foundation:

[0130] Design reference symbol: crosshair target, center at (X setting, Y setting), crosshair length 50mm (corresponding to allowable deviation 50mm);

[0131] Actual position symbol: dot, centered (X solid, Y solid), green (deviation ≤ 50mm), yellow (50mm < deviation ≤ 80mm), red (deviation > 80mm);

[0132] Inclined state symbol: arrow, starting from a dot, in the direction of the inclined horizontal projection, length = inclination amount (mm) / 10.

[0133] A5. Verification and Storage:

[0134] The symbol units are displayed in a visual interface, and staff can identify them by their color and shape.

[0135] Green symbol: Direct confirmation, data is transferred to the storage unit;

[0136] Yellow / red symbols: Manual review. The symbol status is updated after the coordinates are edited through the interface, and stored after passing the review.

[0137] The system includes:

[0138] 1) Coordinate analysis unit:

[0139] Drawing recognition module: Scans design drawings and locates the coordinate table area;

[0140] Information extraction module: Identifies and extracts station number, X coordinate, and Y coordinate;

[0141] Data validation module: checks data format and completes missing information.

[0142] 2) Data acquisition unit:

[0143] 3D positioning module: Employs high-precision positioning technology to obtain real-time coordinates;

[0144] Attitude detection module: acquires tilt angle and direction through tilt sensor;

[0145] Data preprocessing module: Filters noisy data and integrates it into a standard format.

[0146] 3) Association mapping unit: Establishes a one-to-one correspondence between station number and design coordinates and real-time coordinates, and supports dynamic updates.

[0147] 4) Two-dimensional symbolic review unit:

[0148] 5) Symbol generation sub-unit: Draw the crosshair target, dots, and arrows based on the coordinate data;

[0149] 6) Color coding subunit: Assign colors to dots and arrows according to the deviation threshold;

[0150] 7) Interactive Display Sub-unit: Provides a visual interface that supports symbol zooming, hovering to display detailed data, and coordinate editing.

[0151] 8) Data storage unit: Encrypts and stores verified coordinate data, supporting querying and tracing.

[0152] Preferably, the system operation flow is as follows:

[0153] Users upload pile foundation design drawings, the drawing recognition module of the coordinate analysis unit locates the coordinate table, the information extraction module recognizes data such as "Z1, 100.0, 200.0", the data verification module completes the missing "Z2" pile number data, and generates a structured database.

[0154] The 3D positioning module on the construction equipment collects data (X, Y, and Z values) every cycle, while the attitude detection module simultaneously collects the tilt angle and direction. The data is uploaded after preprocessing.

[0155] The association mapping unit binds the design coordinates (100.0, 200.0) of "Z1" to the real-time coordinates (100.02, 200.01), and the two-dimensional symbolic review unit is generated:

[0156] Crosshair target (center 100.0, 200.0);

[0157] Green dot (center 100.02, 200.01, deviation 1.4mm);

[0158] Green arrow (direction 10° east of north, length 0.3cm, corresponding tilt 3mm).

[0159] The interactive display sub-unit displays symbols. When staff confirm that the green symbol is qualified, the data is transmitted to the data storage unit for encrypted storage, generating the record "Z1_20250710_Qualified".

[0160] Example 3

[0161] like Figures 3-4 As shown, a method and system for intelligent pile location coordinate input are presented. Based on optical character recognition technology, it can automatically extract pile location coordinate information from image files, realizing automated input of pile location coordinate tables. This method not only automatically identifies and processes key information in the coordinate table, but also achieves accurate data input and processing through optimized algorithms, avoiding the inefficiency and error rate of traditional manual input.

[0162] The implementation process of this system is as follows:

[0163] B1. Upload pile foundation coordinate file

[0164] Users upload a PDF file containing pile foundation coordinate tables through the front-end interface. Specifically, the user clicks the "Add" button, selects the file, and the system... <input type＝"file"> The system uses tags to select files, then uses AJAX requests to upload the files to the backend server for processing and storage. After receiving the files, the system returns the upload result, allowing the user to proceed to the next step.

[0165] B2. Coordinate File Format Detection

[0166] After the file is uploaded, the system checks the file extension on the backend to ensure that the file is a valid PDF format. If the file does not meet the requirements, the system will return an error message informing the user that the uploaded file is invalid and requesting that a compliant PDF file be uploaded again.

[0167] B3. Coordinate File Preview

[0168] The system uses the PDF rendering tool PDF.js to convert PDF files into image format and display them on the front-end interface. Each page of the PDF file is converted into an image, and users can view the file content through the interface. The image preview automatically adjusts the display ratio according to the page size and content of the PDF file so that users can clearly view the content of each page. Users can view the complete pile foundation coordinate table file and perform operations such as zooming and page turning on the file.

[0169] B4. Image Extraction from Coordinate Files

[0170] The system provides a cropping tool on the front-end interface, allowing users to select the desired text portion by dragging the mouse or clicking on a specified area. This cropping box supports customizable size and position adjustments, facilitating precise text selection. Users drag the mouse to define the area containing the desired text. The cropping box can be moved across the image, and users can adjust its boundaries for accurate selection. The selected area includes the station number, X and Y coordinates, and the system displays the selected area in real-time to confirm the accuracy of the cropping.

[0171] B5. Coordinate Image Preview and Input

[0172] The system will display the selected area in the right sidebar of the file, allowing users to visually see the selected region and confirm the content by comparing the images. Once the user confirms that the selected content is correct, the system provides a "Confirm" button. After the user clicks "Confirm," the system will save the selected image and process it using the algorithm. If the user finds an error in the selected area, they can click the "Cancel" button, which will clear the current operation and allow the user to re-select the image.

[0173] B6. Grayscale conversion of coordinate image

[0174] After acquiring the user-uploaded image, the system converts the color image to a grayscale image, thereby reducing the data dimensionality of the image while preserving basic structural information such as edges and textures. The grayscale image contains only brightness information and no color information, making subsequent processing more efficient and simple. This process is achieved by converting the pixel data of the red, green, and blue channels of the image into grayscale values. The grayscale value Gray is obtained by weighted summation as shown in formula (1):

[0175] Gray=0.2989·R+0.5870·G+0.1140·B (1)

[0176] In this model, R, G, and B represent the red, green, and blue component values ​​of a pixel, respectively. The weighting coefficients are weighting factors determined based on the human eye's sensitivity to different colors. Green light contributes the most to brightness, followed by red light, while blue light contributes the least. The generated new image contains only a single-channel grayscale matrix for subsequent processing steps. However, when the blue channel mean is detected to be >150, the weights of the red and green channels are automatically increased, and the grayscale values...

[0177] B7. Binarization of Coordinate Images

[0178] Based on grayscale images, this system performs binarization. The binarization process divides the pixels in the grayscale image into two categories: text areas are converted to black, and the background is converted to white, distinguishing the text from the background and providing clearer text features for subsequent recognition and extraction. In this calculation, it is assumed that the image's grayscale levels range from 0 to L-1, and the image's grayscale histogram is p(i), where i is the grayscale value, and p(i) represents the probability of a pixel with grayscale value i among the total number of pixels in the image. Let T be the threshold. The image can be divided into two categories: background and foreground. The background represents all pixels with grayscale values ​​less than or equal to the threshold T, and the foreground represents all pixels with grayscale values ​​greater than the threshold T. By calculating the inter-class variance of the image, the system maximizes the inter-class variance between these two categories, thereby achieving optimal binarization of the image. This automatically determines the optimal threshold for the image and provides clearer text features for subsequent recognition and extraction.

[0179] In the process of binarizing the system image, firstly, the total average gray value u of the image is calculated according to formula (2). t :

[0180]

[0181] Next, the weights of class 1 and class 2 are calculated according to formulas (3) and (4), and denoted as ω1(T) and ω2(T) respectively:

[0182]

[0183] ω2(T)=1-ω1(T) (4)

[0184] Next, calculate the average gray values ​​u1(T) and u2(T) of class 1 and class 2 according to formulas (5) and (6):

[0185]

[0186] Subsequently, the inter-class variance between the background and foreground classes is calculated according to formula (7).

[0187]

[0188] Finally, by iterating through all possible thresholds T, we select the one that minimizes the inter-class variance. Maximize the threshold T * , which serves as the final binarization threshold.

[0189] B8. Noise Removal from Coordinate Images

[0190] To remove noise from images and improve the accuracy of subsequent text extraction, the system performs noise removal processing on the binarized image. The goal of noise removal is to preserve key features such as edges and textures of the image while significantly reducing the impact of noise on subsequent processing. The system uses statistical methods or sample analysis to make a preliminary judgment on the type of noise in the input image.

[0191] The system achieves accurate denoising of pile foundation coordinate images through a collaborative processing flow of detection, identification, and repair. First, multi-scale analysis technology is used, combined with Gabor directional filtering and an improved LOF algorithm, to accurately locate noise such as grid lines and ink spots. Then, based on an engineering knowledge rule base and a convolutional CRF model, semantic understanding is used to distinguish between true coordinates and interference information. Finally, an adaptive repair algorithm is applied to achieve background denoising of the pile position coordinate image while protecting the integrity of the text.

[0192] B9. Coordinate Image Smoothing Processing

[0193] To enhance the smoothness of image edge features, the system performs smoothing processing after denoising. This process reduces high-frequency noise (such as random noise or texture details) in the image by using the bilateral filtering smoothing function according to formula (10) to improve image quality.

[0194]

[0195] Where I'(x,y) represents the pixel value after smoothing, I(x,y) represents the pixel value of the pixel within the filtering window, and W p G represents the normalization factor, ω represents the filter window, and G represents the normalization factor. s The weights G represent the distance between pixel spatial locations. r This represents the weight of the difference in grayscale values.

[0196] B10, Coordinate Image Contrast Adjustment

[0197] The system adjusts the image contrast, changing its brightness and contrast to enhance the difference between the foreground and background, making key features of the image more prominent. It also uses a linear stretching method to adjust the image's grayscale range, expanding the grayscale value range to its maximum possible range through a linear function. This operation is calculated according to formula (11):

[0198]

[0199] Among them, I min with I max Let I'(x,y) represent the minimum and maximum gray values ​​in the original image, respectively. Let I'(x,y) represent the pixel value after contrast adjustment, and I(x,y) represent the pixel value of the pixel. min With G max These represent the minimum and maximum ranges of grayscale values ​​of the target image, respectively.

[0200] B11. Coordinate Image Feature Extraction

[0201] The system extracts low-level features (such as edges, textures, and corners) and high-level semantic features from the input image through convolution operations. Through layer-by-layer convolution, non-linear activation, and pooling operations, the model captures multi-scale features of the image for subsequent object detection or classification.

[0202] First, the system performs a convolution operation on the input image using a convolution kernel to detect features such as edges, textures, and contours in the image, and scans the entire image to generate a feature map. The feature values ​​of the image pixels are calculated according to formula (12):

[0203]

[0204] Where F(x,y) represents the value of the new image feature map at position (x,y), W(i,j) represents the weight of the convolution kernel, I(x+i,y+j) represents the value of the neighboring pixels in the input image, and k represents the radius of the convolution kernel.

[0205] Subsequently, the system uses an activation function to introduce nonlinear factors to learn deep features. According to formula (13), the system compresses the image feature output values ​​to the range of (0,1) using the activation function:

[0206]

[0207] Finally, this system introduces max pooling to reduce the spatial dimension of the feature map and gradually extracts low-level edge information and high-level semantic information. The feature value calculation of the image pixel is shown in formula (14):

[0208] P(x,y)=max{F(x+i,y+j)|(i,j)∈ω} (14)

[0209] B12. Generating Coordinate Text Anchor Boxes

[0210] The system uses the feature maps extracted in the early stages as input to a specific text region detection network for processing. For common text box shapes and sizes found in the training data, the system generates K predefined anchor boxes at each grid point in the feature map. The center of each anchor box is aligned with the grid point, and different aspect ratios and scales are used to ensure adaptability to various text regions.

[0211] B13. Generation of Coordinate Text Candidate Boxes

[0212] First, for each anchor box's position on the feature map, the system adjusts its bounding box coordinates using the regression values ​​output by the network. The regression objective of the anchor box is to calculate the offset of the anchor box relative to the real text region. The regression objective of each anchor box is represented by four offsets: the offset of the anchor box center coordinates (Δx, Δy) and the scaling of the anchor box width and height (Δw, Δh). The regression calculations of the above four indicators are shown in formulas (15)-(18):

[0213]

[0214] Where, x gt y gt w gt with h gt These represent the bounding box coordinates of the actual text region, x and x. anchor x anchor x anchor With x anchor These represent the bounding box coordinates of the current anchor box. Next, the system learns these offsets during training, training the model by minimizing the regression loss. In the application phase, as shown in the figure below, these offsets are applied to the initial position and size of the anchor box to obtain the final predicted bounding box, thereby generating the coordinates and size of the candidate boxes.

[0215] B14. Candidate Box Classification and Scoring

[0216] Classification and scoring are the main objectives of object detection tasks. The goal is to classify each candidate box and assign a confidence score to each candidate box. The confidence score reflects the probability that the box contains the target (text region). The classification task determines the target category corresponding to the candidate box and assigns a category label to each candidate box. This system uses a binary classification layer to determine whether each anchor box contains a text region. For each candidate box, the model determines whether the box contains text. If it contains text, it is marked as a "text box"; otherwise, it is marked as a "non-text box". The calculation formula for this binary classification process is shown in formula (19):

[0217] S class =σ(W class ·features+b class (19)

[0218] Where σ represents the Sigmoid activation function, W class The weight matrix of the classification layer is represented by b, and features represents the feature vectors of the input feature map. class This represents the bias term for the classification layer.

[0219] In addition, the classification network calculates the confidence score S for each candidate box according to formula (20). confThis is used to indicate the likelihood of text being present in the candidate box.

[0220] S conf =σ(W conf ·features+b conf (20)

[0221] After classification and score calculation, multiple candidate boxes may overlap. To remove redundant boxes and retain the best box, the system uses non-maximum suppression to further filter candidate boxes. First, the intersection over union (IoU) ratio between each pair of candidate boxes is calculated. IoU represents the ratio of the overlapping portion of two boxes to their union, and is calculated as shown in formula (21):

[0222]

[0223] In this context, region A and region B represent two candidate boxes, and their intersection and union are calculated using the boundary coordinates of the two candidate boxes. Subsequently, the confidence scores S are... conf Candidate boxes are sorted from highest to lowest score, and the highest-scoring candidate boxes are selected sequentially. Overlapping boxes with an IoU value greater than a threshold are removed, and the final selected candidate boxes are retained. After classification, scoring, and non-maximum suppression processing, the system outputs the final text region candidate boxes and their corresponding confidence scores. These outputs will be passed to subsequent character segmentation and text recognition stages.

[0224] B15, Coordinate Text Segmentation

[0225] The system traverses the image pixel by pixel, identifying and labeling connected components by analyzing the connectivity between pixels and their neighbors. For each unvisited pixel, the system checks the values ​​of its surrounding pixels; if these pixel values ​​are the same and have not yet been labeled, they are labeled as belonging to the same connected component. Each time a new connected component is discovered, the system assigns a unique label and assigns the same label to all pixels within that component, ensuring that all connected pixels belong to the same character region. Finally, the system extracts the character region by calculating the bounding box of each connected component and filters out invalid regions, such as noise or non-character connected components, based on rules such as area and shape, thus ensuring that each labeled region corresponds to a valid character.

[0226] B16, Coordinate Sequence Prediction

[0227] After character segmentation, the system performs high-precision prediction for each independent character region or text sequence region. First, the system uses a convolutional neural network to extract high-dimensional features from the character region images, converting each input character image into a feature map with semantic information. For continuous text regions, the system further utilizes temporal modeling techniques, employing a Transformer model based on a self-attention mechanism to process the extracted feature maps and capture the contextual dependencies between characters. Subsequently, the system uses a classification network to classify the features of each character, generating a predicted value for the character category. If the characters are a continuous sequence, the system uses a connection-based temporal classification decoding technique to decode the predicted temporal features into the final text sequence output.

[0228] B17, Coordinate Sequence Correction

[0229] The system extracts key data from the pile location coordinate table from the recognition results, including the X and Y coordinates and the pile number. First, the system parses the text recognition results to identify fields containing coordinate information. To ensure data accuracy, the system performs post-processing on the extracted data. For common errors occurring during recognition, if the coordinate values ​​contain spaces, the system automatically detects and removes extra spaces, concatenating characters into a valid numerical representation. Subsequently, the system corrects symbols in the text, converting characters recognized as commas into decimal points to conform to standard number format requirements.

[0230] B18, Coordinate Row Structure Detection Group

[0231] Considering the possibility of missed or incorrect text detections during the system recognition process, the system uses a y-coordinate discrimination method to traverse all text boxes one by one. First, the y-coordinate of the center point of the first text box is selected as the baseline, and it is then checked whether it falls within the y-coordinate range (y1, y2) of subsequent text boxes. If it does, the process continues until the center point's y-coordinate no longer falls within the range of any text box. At this point, all text boxes determined to be within the same y-coordinate range are merged into one row, and the center point of the next unmatched text box is selected as the new baseline. This traversal process is repeated until all text boxes are divided into multiple rows of data.

[0232] B19, Coordinate Column Structure Detection Group

[0233] The system processes the text data, which has been divided into multiple lines, line by line. First, the first three characters of the second text data in each line are defined as the X-coordinate identifier, and the first three characters of the third text data are defined as the Y-coordinate identifier. Next, the system uses regular expressions to match the first three characters of each text data in each line. If an X-coordinate identifier is matched, the text data is placed in the second position of the line; if a Y-coordinate identifier is matched, it is placed in the third position, thus achieving automatic classification and positioning of X-coordinate and Y-coordinate data.

[0234] B20, Station Number Data Generation

[0235] For unclassified text data, the system calculates the proportion of letter elements. If this proportion is below a preset threshold, the station number sequence is determined to be a pure numeric sequence. The system then determines the starting value of the station number based on the first or second line of data and sequentially generates the complete station number sequence. If the proportion of letters in the unclassified data exceeds the preset threshold, the system extracts the letter parts of the text data prefixes and calculates their frequency. The most frequent letter prefix is ​​selected as the letter part of the station number sequence, and the starting value of the numeric part is determined based on the first or second line of data, sequentially generating the complete station number sequence. If there is no unclassified text data, the system automatically assigns station numbers starting from 1.

[0236] B21. Station Number Sequence Verification and Completion

[0237] Check if the station numbers in each row conform to the expected numbering pattern. Establish station number parsing rules, extract the basic numbering (such as the numeric part after the letter), and verify whether it forms a continuous sequence. When a missing number is found, intelligently complete it based on the context. For example, when ZK-1 and ZK-3 exist but ZK-2 is missing, insert the missing station number at the appropriate row position. The completed station numbers are marked with special markers (such as different colors or fonts) for easy identification, and an auto-completion log is recorded.

[0238] B22, Coordinate Structured Output

[0239] After the above steps, the system finally generates structured data output with station number, X coordinate, and Y coordinate as columns. An example of the data format is shown below:

[0240]

[0241] B23. Data Verification and System Storage

[0242] The system outputs results to the user, who can view and confirm the accuracy of the recognized content on the front-end interface. Users can directly click on the text in the recognition results to make modifications; the system provides real-time editing functionality. When a character or text area is clicked, the cursor automatically jumps to the corresponding text box, allowing for quick and easy editing. After the user confirms the recognition results, the system performs final data processing and saves the data to ensure accuracy.

[0243] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for managing pile foundation coordinate data, characterized in that: Includes the following steps: S1. Collect images of the pile location coordinate table, preprocess the images, and extract image features; S2. Use optical character recognition algorithms to perform text recognition on the preprocessed image, extract text information and perform structured processing to generate preliminary coordinate data; S3. Correct the preliminary coordinate data, including removing extra spaces and correction symbols, and perform coding verification and completion of the station number sequence; S4. Output the corrected coordinate data in a structured format to form a data table of station number, X coordinate, and Y coordinate; S5. Analyze the pile foundation design drawings to extract the pile number and corresponding X and Y coordinates to generate a structured database, and at the same time collect real-time three-dimensional coordinates and attitude data during the pile foundation construction process; S6. Establish the association mapping relationship between the station number and the design coordinates and real-time coordinates, and generate visual symbol units through the two-dimensional symbol review module to present the design location, actual location and tilt status of the pile foundation; S7. Manually verify the deviation status based on the symbol unit, and store qualified data in the system database after final confirmation.

2. The pile foundation coordinate data management method according to claim 1, characterized in that: The symbol units generated by the two-dimensional symbolization review module in step S6 include: Design reference symbol: represented by a crosshair target, with the center corresponding to the pile foundation design coordinates, and the length of the crosshair matching the allowable deviation threshold; Actual location symbol: represented by a dot, with the center corresponding to the actual coordinates of the pile foundation, and the color is coded as green, yellow or red according to the deviation level; Inclination symbol: indicated by an arrow, the direction of which corresponds to the horizontal projection direction of the pile foundation inclination, and the length reflects the degree of inclination proportionally.

3. The pile foundation coordinate data management method according to claim 1, characterized in that: The process of analyzing pile foundation design drawings in step S5 includes: Identify the coordinate table area in the drawing; Extract station number, X coordinate, and Y coordinate information; Verify data integrity and complete any missing information; Output structured data.

4. The pile foundation coordinate data management method according to claim 1, characterized in that: The verification process in step S7 includes: Automatically identify color and size deviations of symbol units; For units with green symbols, confirm and enter the information directly; Warnings are issued for units marked with yellow or red symbols, prompting manual review and correction.

5. A system employing the pile foundation coordinate data management method according to any one of claims 1 to 4, characterized in that: include: The coordinate analysis unit is used to parse design drawings and generate structured coordinate data; The data acquisition unit is used to collect real-time coordinate and attitude data during the construction process; The associated mapping unit establishes the relationship between the station number and the design coordinates and real-time coordinates; Two-dimensional symbolic review unit, generating a visual unit containing design reference symbols, actual position symbols, and tilt status symbols; The data storage unit stores the coordinate data after verification.

6. The pile foundation coordinate data management method according to claim 5, characterized in that: The two-dimensional symbolization review unit includes: The symbol generation sub-unit generates corresponding symbols based on the design and real-time coordinates. The color-coded subunit assigns colors to symbols based on the deviation level; The interactive display sub-unit provides a visual interface for symbols and manual editing functions.

7. The system for pile foundation coordinate data management according to claim 5, characterized in that: The data acquisition unit includes: The three-dimensional positioning module obtains the real-time location of the pile foundation; The attitude detection module detects the tilt angle and direction of the pile foundation. The data preprocessing module performs noise reduction and integration on the collected data.

8. The system for pile foundation coordinate data management according to claim 6, characterized in that: The interactive display sub-units include: Symbol scaling and panning functions; Deviation value display function; Batch filtering of abnormal symbols; Coordinate editing and symbol update functions.

9. The system for pile foundation coordinate data management according to claim 6, characterized in that: It also includes a construction trajectory analysis unit, which generates a single pile construction path report based on pile foundation coordinate data, calculates drilling speed and positioning adjustment frequency parameters by combining timestamp data, identifies abnormal trajectories and marks risk periods.

10. The system for pile foundation coordinate data management according to claim 6, characterized in that: It also includes a spatial interference detection unit, which calculates the spatial distance and relative tilt angle between adjacent pile foundations using coordinate data. When the measured distance is less than the minimum design clearance or the relative tilt angle exceeds the threshold, it automatically triggers a collision risk warning and outputs adjustment suggestions.

Images