Deep point cloud based high-substitution-rate recycled concrete shape state detection method

Abstract

The present application relates to the technical field of high-substitution-rate recycled concrete, and discloses a high-substitution-rate recycled concrete shape state detection method based on a deep point cloud, which comprises the following steps: step S1, data acquisition; step S2, point cloud preprocessing, regional cutting and noise filtering are performed on each frame of point cloud, and effective point cloud data representing the concrete entity is extracted; step S3, key morphological parameter analysis; step S4, result output, T500 time, final expansion degree and slump estimation value are outputted, and a detection report is formed. Through automatic triggering and collection of a depth camera and automatic processing and analysis of an industrial computer, manual observation and measurement are replaced, the algorithm objectively calculates T500 time, expansion degree and slump based on point cloud data, individual differences and subjective deviations caused by human experience, operation methods and visual errors are eliminated, the consistency and repeatability of the detection results are ensured, and the time point and frequency of detection can be strictly guaranteed.

Description

Technical Field

[0001] This invention relates to the field of high-replacement-rate recycled concrete technology, specifically a method for detecting the external shape of high-replacement-rate recycled concrete based on depth point clouds. Background Technology

[0002] Due to the complex sources of high-replacement-rate recycled concrete aggregates, the concrete strength exhibits large dispersion, making slump control significantly more difficult than that of ordinary concrete. When using recycled concrete in construction, dynamic monitoring of slump is required, with a high testing frequency. Generally, slump changes are tested every 30 minutes, but this is shortened to 15 minutes per test when the temperature is above 30℃.

[0003] However, the sampling and testing process in the current recycled concrete industry still heavily relies on manual operation, making it difficult to guarantee the accuracy of sampling frequency and testing timing. The judgment of concrete quality upon leaving the factory also mainly depends on the subjective experience of technicians, and the differences in individual technical skills have a significant impact on the control results. At the same time, the degree of digitization of quality data is low, the feedback and response process is lengthy, response measures lack intelligent support, and management models rely heavily on manual experience and manual governance. Overall, the level of automation, digitization, and intelligence is lagging behind. Based on this, this invention designs a high-replacement-rate recycled concrete shape condition detection method based on deep point cloud to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to provide a method for detecting the shape and condition of high-replacement-rate recycled concrete based on deep point clouds, so as to solve the problem mentioned above that the quality depends on the subjective experience of technicians.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for detecting the external appearance of high-replacement-rate recycled concrete based on deep point clouds includes the following steps: Step S1, Data Acquisition: Using a depth camera, at the start of the concrete slump test, continuously acquire depth point cloud sequences of its flow and diffusion process at a set frequency. Step S2, point cloud preprocessing, performing region cropping and noise filtering on each frame of point cloud to extract effective point cloud data representing the concrete entity; Step S3, Key morphological parameter analysis; Step S4: Output the results, outputting the estimated values ​​of T500 time, final expansion and slump, and forming a test report.

[0006] Preferably, step S3 includes the following steps: Step S3.1, T500 time detection: Calculate the fitted circle diameter of the concrete diffusion edge frame by frame. When the diameter reaches 500 mm for the first time, record the corresponding time as T500 time. Step S3.2, final spread calculation: determine the moment when the concrete stops spreading, and calculate the maximum diameter of the projection of the concrete point cloud onto the horizontal plane at that moment as the final spread; Step S3.3, slump estimation: Based on the height distribution of the concrete point cloud after diffusion stops, calculate the average height difference between the top and bottom regions, and then convert it into an estimated slump value through a pre-calibrated linear relationship.

[0007] Preferably, in step S3.2, the criterion for determining the moment when concrete stops spreading is: the change in the diameter of the fitted circle over multiple consecutive frames is less than a set threshold. The calculation of the expansion degree includes the following steps: first, calculate the convex hull of the projection point set on the horizontal plane, then calculate the Euclidean distance between all vertices of the convex hull, and take the maximum value as the final expansion degree.

[0008] Preferably, in step S3.3, the pre-calibrated linear relationship is obtained by fitting the height difference measured by the point cloud with the actual slump measured by the standard test, and its formula is: Slump estimate = Height difference × Calibration coefficient + Calibration constant.

[0009] Preferably, in step S1, the set frequency of the depth camera is 60Hz; the data acquisition is initiated by an industrial control computer connected to the depth camera and triggered the instant the concrete begins to pour from the slump cone.

[0010] Preferably, in step S2, the region clipping specifically uses the center of the slump cone as a reference, sets a square region with a side length of 800mm as the region of interest, and sets a height threshold to filter out background point clouds that are more than 80mm above the top surface of the slump cone. The noise filtering uses a density-based spatial clustering algorithm to remove outliers, with the neighborhood search radius set to 10 mm and the minimum number of points required to form a cluster set to 20.

[0011] Preferably, in step S3.1, the frame-by-frame calculation of the fitted circle diameter of the concrete diffusion edge is implemented using a random sampling consistency circle fitting algorithm, wherein the number of iterations is set to 500 and the inner point distance threshold is 3mm.

[0012] Preferably, in step S4, the generated test report also includes the morphological evolution curve of the concrete flow process, key frame point cloud map, and longitudinal section morphological diagram.

[0013] Preferably, the calibration coefficients and calibration constants are obtained by simultaneously conducting multiple standard slump tests and point cloud measurements on high-replacement recycled concrete with the same mix proportion. The height difference obtained from the point cloud measurement and the actual slump measured by the standard test are used as a dataset, and linear regression fitting is performed using the least squares method to determine the calibration coefficients and calibration constants.

[0014] Preferably, the method further includes step S5, performance determination: comparing the estimated T500 time, final spread and slump obtained from step S3 with the qualified range of the corresponding preset indicators, and outputting the determination result of whether the workability of the batch of recycled concrete is qualified.

[0015] Compared with the prior art, the beneficial effects achieved by the present invention are: 1. This invention replaces manual observation and measurement by automatically triggering data acquisition with a depth camera and automatically processing and analyzing the data with an industrial control computer. The algorithm objectively calculates the T500 time, expansion and collapse based on point cloud data, eliminating individual differences and subjective biases caused by human experience, operation methods and visual errors, ensuring the consistency and repeatability of the detection results, and strictly guaranteeing the detection time and frequency.

[0016] 2. This invention transforms the flow state of concrete into a three-dimensional point cloud sequence, and further analyzes it into precise digital parameters and visualization charts. This process generates a structured digital inspection report, enabling quality information to be recorded and presented in digital form in a timely, accurate, and comprehensive manner, laying the foundation for rapid data transmission, storage, and analysis.

[0017] 3. This invention, by automatically comparing the key parameters obtained from the analysis with the preset qualified range and outputting the judgment result, can directly drive automatic alarm, mix ratio adjustment or factory control decision-making, thereby transforming the post-event management that relies on human experience into intelligent pre-event and in-event control based on real-time data, significantly improving the response speed and intelligence level of quality control. Attached Figure Description

[0018] Figure 1 This is a flowchart of the overall detection method of the present invention; Figure 2 This is a detailed diagram of the data acquisition and preprocessing process of the present invention; Figure 3 This is a calculation diagram of the T500 time and final expansion degree of the present invention; Figure 4 This is a diagram showing the slump estimation, calibration, and output results of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1; Please see Figures 1-4 In this embodiment of the invention, a method for detecting the shape and state of high-replacement-rate recycled concrete based on depth point clouds is characterized by the following steps: Step S1, data acquisition, using a depth camera to continuously acquire a sequence of depth point clouds of the flow and diffusion process of concrete at a set frequency at the start of the concrete slump test; Step S2, point cloud preprocessing, performing region cropping and noise filtering on each frame of point cloud to extract effective point cloud data characterizing the concrete entity; Step S3, key morphological parameter analysis; Step S4, result output, outputting the T500 time, final expansion, and slump estimate to form a test report.

[0021] Step S3 includes the following steps: Step S3.1, T500 time detection: Calculate the fitted circle diameter of the concrete diffusion edge frame by frame. When the diameter reaches 500 mm for the first time, record the corresponding time as T500 time. Step S3.2, final spread calculation: determine the moment when the concrete stops spreading, and calculate the maximum diameter of the projection of the concrete point cloud onto the horizontal plane at that moment as the final spread; Step S3.3, slump estimation: Based on the height distribution of the concrete point cloud after diffusion stops, calculate the average height difference between the top and bottom regions, and then convert it into an estimated slump value through a pre-calibrated linear relationship.

[0022] In step S3.2, the criterion for determining when the concrete stops spreading is: the change in the diameter of the fitted circle over multiple consecutive frames is less than a set threshold. The calculation of the expansion degree includes the following steps: first, calculate the convex hull of the projection point set on the horizontal plane, then calculate the Euclidean distance between all vertices of the convex hull, and take the maximum value as the final expansion degree.

[0023] In step S3.3, the pre-calibrated linear relationship is obtained by fitting the height difference measured by the point cloud with the actual slump measured by the standard test. Its formula is: Slump estimate = Height difference × Calibration coefficient + Calibration constant.

[0024] The working principle of this invention is as follows: the moment the concrete begins to pour from the slump cone, the depth camera controlled by the industrial control computer is immediately activated and acquires a three-dimensional point cloud sequence of the concrete flow and diffusion process at a high frequency of 60Hz. This high-frequency acquisition mechanism ensures that all the subtle deformations of the concrete from stillness to flow and then to cessation can be captured, providing a high temporal resolution data foundation for subsequent analysis of time and space parameters.

[0025] The acquired raw point cloud sequence first enters the preprocessing pipeline. The system uses the center of the slump cone as the origin, delineates a square region of interest (ROI) with sides of 800 mm on a horizontal plane, and sets a height threshold of 80 mm above the cone top. Each frame of the point cloud is then automatically cropped within this three-dimensional spatial range. This step intelligently eliminates irrelevant point clouds such as those from the test bench and background environment, focusing on the concrete entity. Subsequently, a density-based spatial clustering algorithm uses a neighborhood radius of 10 mm and a minimum cluster size of 20 points to filter out outliers from the cropped point cloud. This effectively eliminates discrete points caused by material splashing and sensor noise, thereby extracting a high-quality, high-density effective point cloud characterizing the concrete mass.

[0026] The preprocessed clean point cloud sequence is then fed into the key morphological parameter analysis engine. The system employs a random sampling consistency circle fitting algorithm to fit the projection edge of each frame's point cloud onto the horizontal plane. The iteration count is set to 500 times, and the in-point distance threshold is 3 mm to ensure robustness and accuracy of the fitting. The algorithm calculates the diameter of the fitted circle frame by frame, and the system monitors this value in real time. When the diameter first reaches 500 mm, this moment is immediately recorded as time T500. This process quantifies the concrete flow velocity. The determination of flow cessation relies on monitoring the change in the fitted circle diameter over multiple consecutive frames. When the change consistently falls below a preset threshold, the system determines that diffusion has terminated and locks the state of the concrete point cloud at that moment. For the point cloud at this moment, the system first calculates the convex hull of its horizontal projection point set, then solves for the maximum value of the Euclidean distance between all vertices of the convex hull, using this value as the final expansion degree. This method more accurately reflects the irregular boundary contour of the spread concrete than simply measuring the diameter. The slump estimation is based on a pre-calibrated linear model. The system calculates the average height difference between the top and bottom regions of the concrete point cloud after diffusion stops, and then substitutes this height difference into the formula "slump estimate equals height difference multiplied by calibration coefficient plus calibration constant" to directly calculate the slump value.

[0027] Finally, all the parameters obtained from the analysis, including T500 time, final spread, slump estimate, morphological evolution curve of concrete flow process, key frame point cloud map, longitudinal section morphology diagram, etc., are automatically summarized to generate a structured digital test report, completing the complete transformation from physical phenomena to quantitative data and visualization results.

[0028] Example 2; Please see Figures 1-4 In this embodiment of the invention, in step S1, the set frequency of the depth camera is 60Hz; data acquisition is initiated by an industrial control computer connected to the depth camera and triggered the instant the concrete begins to pour from the slump cone. In step S2, region clipping is performed, specifically using the center of the slump cone as a reference, a square region with a side length of 800mm is set as the region of interest, and a height threshold is set to filter out background point clouds that are more than 80mm above the top surface of the slump cone. Noise filtering employs a density-based spatial clustering algorithm to remove outliers, with the neighborhood search radius set to 10 mm and the minimum number of points required to form a cluster set to 20.

[0029] In step S3.1, the diameter of the fitted circle of the concrete diffusion edge is calculated frame by frame using the random sampling consistency circle fitting algorithm, where the number of iterations is set to 500 and the threshold for the distance between inner points is 3mm.

[0030] The working principle of this invention is as follows: At the data acquisition end, the depth camera operates at a fixed frequency of 60 Hz. This is not arbitrary, but rather designed to capture a sufficient number of frames in the early stages of rapid concrete flow to accurately calculate the T500 time, while avoiding excessively large amounts of data. The linkage between the industrial control computer and the camera is crucial. It not only provides control and storage support in hardware, but also logically automates the acquisition process, starting acquisition the instant the pouring begins, eliminating timing errors introduced by manual operation delays.

[0031] In the point cloud preprocessing stage, parameter settings directly serve the purification objective. An 800mm square trimming area, based on the cylinder center, ensures complete coverage of the entire dynamic range from the bottom diameter of the slump cylinder to the maximum possible expansion, while eliminating distant interference. Setting an 80mm height threshold cleverly removes background points that might be located on the cylinder wall's upper edge or an operator's arm, focusing attention entirely on the falling concrete entity. In the noise filtering stage, a density-based spatial clustering algorithm is employed, with a neighborhood radius of 10mm and a minimum number of points of 20. This configuration is based on prior knowledge of the spacing and density of typical concrete point clouds.

[0032] In the core step of morphological analysis—circle fitting—a random sampling consensus algorithm is employed, with 500 iterations and an inlier threshold of 3 mm. Its working principle is to mitigate the influence of potential local irregularities or residual outliers in the point cloud through multiple random samplings. The high number of iterations ensures the finding of the optimal circle model with high probability, while the 3 mm inlier threshold serves as a precision control parameter, allowing the fitted circle to encompass the natural, minor undulations of the point cloud edges within a certain tolerance. This results in a stable and reliable diameter sequence, providing a solid data foundation for accurate T500 time determination and reliable flow cessation assessment.

[0033] Example 3; Please see Figures 1-4 In this embodiment of the invention, the test report generated in step S4 also includes the morphological evolution curve of the concrete flow process, key frame point cloud map, and longitudinal section morphology map. The calibration coefficients and calibration constants are obtained as follows: For high-replacement-rate recycled concrete with the same mix proportion, multiple standard slump tests and point cloud measurements are performed simultaneously. The height difference obtained from the point cloud measurement and the actual slump measured in the standard test are used as the dataset. Linear regression fitting is performed using the least squares method to determine the calibration coefficients and calibration constants.

[0034] The method also includes step S5, performance determination: the estimated values ​​of T500 time, final spread and slump obtained from step S3 are compared with the qualified range of the corresponding preset indicators, and the result of determining whether the performance of this batch of recycled concrete is qualified is output.

[0035] The working principle of this invention is as follows: In the calibration stage, the working principle is based on establishing a high-precision measurement model. For each specific high-replacement-rate recycled concrete mix proportion, a parallel calibration process needs to be executed. Under the same environmental conditions, multiple traditional standard slump tests and point cloud measurements of this system are performed simultaneously. The paired height differences and the actual slump data are combined to form a dataset. Then, the least squares method is applied for linear regression. This process determines the calibration coefficients and calibration constants in the aforementioned linear formula, so that the point cloud system has a reliable conversion basis for estimating the slump of subsequent concrete with the same mix proportion. Its accuracy is directly comparable to the standard test method.

[0036] In terms of output, the system goes beyond simple numerical listing. It automatically plots the diameter of the fitted circle for the time series into a morphological evolution curve, visually displaying the flow rate and trend of concrete. Simultaneously, the system extracts point cloud maps of key time points such as the start of diffusion, T500 time, and the cessation of diffusion, as well as longitudinal section morphology maps after diffusion stops, integrating these along with numerical results such as T500 time, final spread, and slump estimate into the test report.

[0037] Furthermore, through the optional step S5, the system's working principle achieves a leap from detection to intelligent judgment. This step automatically compares the three core parameters obtained from the analysis—T500 time, final spread, and slump estimate—with the pre-input standards for the acceptable range of the project or concrete mix for that specific project. Based on the comparison results, the system directly outputs a clear judgment conclusion, such as whether the workability is qualified or unqualified, and indicates that a specific indicator exceeds the range. This allows the method to not only be used for quality inspection but also directly serve real-time quality control and factory acceptance during the production process, forming a complete intelligent chain of perception, analysis, and decision-making.

[0038] Working principle: During the standard slump test, the depth camera is automatically triggered the moment the concrete begins to pour from the slump cone, and continuously acquires a three-dimensional point cloud sequence of the concrete flowing, spreading and coming to rest on the plane at a high frequency of 60 frames per second. This high-frequency acquisition ensures that the rapid dynamic changes in the concrete shape can be fully captured, providing a high temporal resolution data foundation for subsequent accurate timing and dimensional measurement.

[0039] After acquiring the raw point cloud sequence, the system immediately enters the automated preprocessing process. First, intelligent region clipping is performed. Using the center of the slump cone as a reference, the system defines a square region of interest with sides of 800 mm on the horizontal plane and sets a height threshold to filter out background interference above 80 mm of the cone top, thus focusing computational resources on the concrete entity itself. Next, a density-based spatial clustering algorithm is used to denoise the clipped point cloud. The parameters are set to a neighborhood search radius of 10 mm and a minimum number of points required to form a cluster of 20. This step effectively filters out discrete points caused by slurry splashing or sensor noise, retaining high-quality, high-density point cloud clusters that characterize the overall shape of the concrete.

[0040] After preprocessing, the system analyzes key morphological parameters of each frame of clean point cloud. First, the system performs T500 time detection, which involves fitting the projection edge of each frame's point cloud onto the horizontal plane using a random sampling consensus circle fitting algorithm. The iteration count is set to 500 times, and the in-point distance threshold is 3 mm to robustly calculate the diameter of the fitted circle for concrete spread at that moment. The system monitors this diameter in real time, accurately recording the moment it first reaches 500 mm as the T500 time, used to characterize the concrete flow velocity. Then, the final spread is calculated. The system continuously monitors the rate of change of the fitted circle diameter; when the change in diameter falls below a set threshold for multiple consecutive frames, the concrete flow is considered to have stopped. Next, the system calculates the convex hull of the point cloud projection onto the horizontal plane at this final moment and finds the maximum Euclidean distance between all vertices of the convex hull, using this as the final spread. This method more accurately reflects the possible irregular contours of the spread concrete. Finally, slump estimation is performed. After the flow stops, the system analyzes the height distribution of the static concrete point cloud, calculates the average height difference between the top and bottom regions, and converts this height difference into a slump estimate using a pre-calibrated linear model. This calibration model is established for concrete with a specific mix proportion. It involves collecting data on the corresponding height difference and actual slump by conducting multiple simultaneous standard slump tests and measurements using the system, and then using the least squares method for linear regression to determine a unique conversion coefficient and constant.

[0041] Finally, the system combines the automatically obtained T500 time, final spread, and slump estimates, along with the morphological evolution curves, keyframe point cloud maps, and longitudinal section morphology diagrams of the concrete flow process, to generate a structured digital inspection report. Furthermore, the system can automatically compare the analysis results with preset acceptable ranges, directly outputting a judgment on whether the concrete's workability is qualified or unqualified, thus achieving full automation and digitization of the entire process from data acquisition and intelligent analysis to quality assessment.

[0042] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting the external shape of high-replacement-rate recycled concrete based on deep point clouds, characterized in that, Includes the following steps: Step S1, Data Acquisition: Using a depth camera, at the start of the concrete slump test, continuously acquire depth point cloud sequences of its flow and diffusion process at a set frequency. Step S2, point cloud preprocessing, performing region cropping and noise filtering on each frame of point cloud to extract effective point cloud data representing the concrete entity; Step S3, Key morphological parameter analysis; Step S4: Output the results, outputting the estimated values ​​of T500 time, final expansion and slump, and forming a test report.

2. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 1, characterized in that, Step S3 includes the following steps: Step S3.1, T500 time detection: Calculate the fitted circle diameter of the concrete diffusion edge frame by frame. When the diameter reaches 500 mm for the first time, record the corresponding time as T500 time. Step S3.2, final spread calculation: determine the moment when the concrete stops spreading, and calculate the maximum diameter of the projection of the concrete point cloud onto the horizontal plane at that moment as the final spread; Step S3.3, slump estimation: Based on the height distribution of the concrete point cloud after diffusion stops, calculate the average height difference between the top and bottom regions, and then convert it into an estimated slump value through a pre-calibrated linear relationship.

3. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 2, characterized in that, In step S3.2, the criterion for determining the moment when concrete stops spreading is: the change in the diameter of the fitted circle over multiple consecutive frames is less than a set threshold. The calculation of the expansion degree includes the following steps: first, calculate the convex hull of the projection point set on the horizontal plane, then calculate the Euclidean distance between all vertices of the convex hull, and take the maximum value as the final expansion degree.

4. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 2, characterized in that: In step S3.3, the pre-calibrated linear relationship is obtained by fitting the height difference measured by the point cloud with the actual slump measured by the standard test. Its formula is: Slump estimate = Height difference × Calibration coefficient + Calibration constant.

5. The method for detecting the external shape of high-replacement-rate recycled concrete based on deep point cloud as described in claim 1, characterized in that: In step S1, the set frequency of the depth camera is 60Hz; the data acquisition is initiated by the industrial control computer connected to the depth camera and is triggered the instant the concrete begins to pour from the slump cone.

6. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 1, characterized in that: In step S2, the region clipping specifically uses the center of the slump cone as a reference, sets a square region with a side length of 800mm as the region of interest, and sets a height threshold to filter out background point clouds that are more than 80mm above the top surface of the slump cone. The noise filtering uses a density-based spatial clustering algorithm to remove outliers, with the neighborhood search radius set to 10 mm and the minimum number of points required to form a cluster set to 20.

7. The method for detecting the external shape of high-replacement-rate recycled concrete based on deep point cloud as described in claim 2, characterized in that: In step S3.1, the frame-by-frame calculation of the fitted circle diameter of the concrete diffusion edge is implemented using a random sampling consistency circle fitting algorithm, wherein the number of iterations is set to 500 and the inner point distance threshold is 3mm.

8. The method for detecting the external shape of high-replacement-rate recycled concrete based on deep point cloud as described in claim 1, characterized in that: In step S4, the generated test report also includes the morphological evolution curve of the concrete flow process, key frame point cloud map, and longitudinal section morphological diagram.

9. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 4, characterized in that, The calibration coefficients and calibration constants are obtained as follows: For high-replacement-rate recycled concrete with the same mix proportion, multiple standard slump tests and point cloud measurements are carried out simultaneously. The height difference obtained from the point cloud measurement and the actual slump measured by the standard test are used as the dataset. Linear regression fitting is performed using the least squares method to determine the calibration coefficients and calibration constants.

10. The method for detecting the external shape of high-replacement-rate recycled concrete based on depth point cloud as described in claim 1, characterized in that, The method further includes step S5, performance determination: comparing the estimated T500 time, final spread and slump obtained from step S3 with the qualified range of the corresponding preset indicators, and outputting the determination result of whether the workability of this batch of recycled concrete is qualified.

Images