Building solid waste recycled aggregate compaction degree detection method and device based on image processing

Abstract

The invention discloses a method and a device for detecting the compactness of building solid waste recycled aggregates based on image processing, and the method comprises the following steps: carrying out a vibration compaction test on the building solid waste recycled aggregates with different gradations and dosages, and collecting surface images by using a camera to respectively obtain the compactness corresponding to the aggregates with different gradations and dosages; segmenting the aggregate image by using an improved YOLOv8-seg model, and extracting an image index related to the compactness; training the image index parameters, the gradation, the usage and the corresponding compactness by using a machine learning model to obtain a compactness prediction model; and verifying the compaction degree prediction model by using the to-be-tested building solid waste regenerated filler with known gradation and dosage. Compared with a traditional method, rapid, simple and non-destructive detection of the roadbed compactness of the building solid waste recycled aggregate can be carried out through image processing, machine learning and other methods, the investment of manpower and material resources is reduced, and the industry benefits are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of computer vision, machine learning, and the like, and in particular to a method and device for detecting compaction degree of recycled aggregate from construction solid waste based on image processing. Background Art

[0002] With the acceleration of urbanization, the output of construction solid waste is also increasing, bringing tremendous pressure to society and the environment. However, promoting the resource utilization of construction solid waste can alleviate the environmental and social problems caused by construction solid waste. Therefore, it is of great practical significance to study the use of construction solid waste recycled aggregate as roadbed filler. As an important indicator of roadbed construction, the compaction degree of the roadbed will lead to serious diseases such as roadbed settlement and pavement fracture if the compaction degree does not meet the requirements. The traditional roadbed compaction detection methods mainly include water irrigation method, settlement difference method, nuclear density meter, etc., but these methods not only require a lot of manpower and material resources, but also cause damage to the roadbed. With the rapid development of the field of machine vision, new ideas have been provided for the rapid and non-destructive detection of the compaction degree of roadbed fillers. Therefore, it is necessary to establish a method and device for compaction degree detection of construction solid waste recycled aggregate based on image processing. Summary of the invention

[0003] In order to solve the defects existing in the prior art, the present invention discloses a method for compaction detection of recycled aggregate from construction solid waste based on image processing, and its technical solution is as follows:

[0004] S1, using different gradations and amounts of recycled aggregates from construction solid waste to conduct vibration compaction tests and using a camera to collect surface images to obtain the compaction degrees corresponding to the aggregates with different gradations and amounts;

[0005] S2, using the improved YOLOv8-seg model to segment the aggregate image and extract image indicators related to compaction;

[0006] S3, training the image index related parameters and dosage, gradation and corresponding compaction degree using a machine learning model to obtain a compaction degree prediction model;

[0007] S4, using the construction solid waste recycled filler with known gradation and dosage to be tested to verify the compaction prediction model.

[0008] The present invention also discloses a device for detecting compaction degree of recycled aggregate from construction solid waste based on image processing, comprising:

[0009] The vibration compaction module is used to conduct vibration compaction tests on multiple groups of construction solid waste recycled aggregates with different gradations and dosages, and to match the gradations, dosages and aggregate compaction degrees respectively;

[0010] An image processing module is used to collect the surface image of the recycled aggregate from construction solid waste, segment the image of the construction solid waste using the improved YOLOv8-seg model, and extract image indicators: void ratio, area of ​​each void, void perimeter, maximum void width and void roundness, and obtain fitting curve parameters of image indicator distribution;

[0011] Roadbed compaction detection model: used to convert the gradation, dosage and image corresponding parameters into vectors, and process them using neural networks to obtain the compaction detection model of the construction solid waste recycled aggregate roadbed.

[0012] The present invention also discloses a non-volatile storage medium, characterized in that the non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above-mentioned method for compaction detection of recycled aggregates of construction solid waste based on image processing.

[0013] The present invention also discloses an electronic device for detecting the compaction degree of recycled aggregates from construction solid waste based on image processing, which is characterized by comprising a processor and a memory; the memory stores computer-readable instructions, and the processor is used to run the computer-readable instructions, wherein the computer-readable instructions execute the above-mentioned method for detecting the compaction degree of recycled aggregates from construction solid waste based on image processing when running.

[0014] Beneficial Effects

[0015] (1) Based on the improved YOLOv8-seg model, the image of recycled aggregate from construction solid waste can be segmented accurately to segment the void contours and extract image indicators related to compaction;

[0016] (2) Based on the machine learning method, the aggregate grading, dosage, image parameters and compaction degree are correlated to obtain a compaction prediction model for recycled aggregate from construction solid waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the implementation of the method for detecting compaction degree of recycled aggregate from construction solid waste of the present invention;

[0018] Figure 2 It is a schematic diagram of a vibration table of the present invention;

[0019] Figure 3 It is the SPPF structure diagram in YOLOv8-seg of the present invention;

[0020] Figure 4 This is the structure diagram of SKAPPF in the improved YOLOv8-seg of the present invention;

[0021] Figure 5 It is the convergence curve of the neural network training in the present invention. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. These embodiments are provided to make the present invention more thoroughly and completely conveyed to those skilled in the art.

[0023] The first aspect of the present invention provides a method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing. The schematic diagram of the embodiment of the method is shown in FIG. Figure 1 As shown, the following steps are included:

[0024] S1, using different gradations and amounts of recycled aggregates from construction solid waste to conduct vibration compaction tests and using a camera to collect surface images to obtain the compaction degrees corresponding to the aggregates with different gradations and amounts;

[0025] Specifically, vibration compaction tests were carried out on recycled aggregates from construction solid waste with different gradations and dosages. The coarse aggregate void ratio VCA could be calculated based on the volume density of the aggregate mixture, and then the compaction degree of the recycled aggregate roadbed from construction solid waste could be obtained, and the compaction degree could be matched with the aggregate gradation and dosage.

[0026] Specifically, measure the inner height and inner diameter of the compacting cylinder. After loading, place the compacting cylinder on a vibration table for vibration compaction. Figure 2 As shown, and measuring the distance from the edge of the compaction cylinder to the aggregate, the volume density of the construction solid waste recycled aggregate mixture can be obtained, as shown in formula (1) (2):

[0027]

[0028] Where V represents the compaction degree of aggregate, r represents the radius of the inner wall of the compaction cylinder (cm), and h represents the height of the inner wall of the compaction cylinder (cm).

[0029]

[0030] Where ρ represents the natural bulk density of recycled aggregate from construction solid waste measured by the vibration compaction method (g / cm 3 ), m represents the total mass of aggregate (kg), and h1 represents the height difference between the top of the aggregate and the compacting cylinder after compaction (cm).

[0031] Furthermore, the formula for calculating the aggregate void ratio VCA based on the bulk density of the aggregate mixture is as follows:

[0032]

[0033] Where n represents the void ratio of aggregate, ρ represents the natural packing density of the aggregate mixture measured by the vibration compaction test, and ρ bRepresents the composite gross volume density of the aggregate.

[0034] Furthermore, the calculation formula for the roadbed of recycled aggregate from construction solid waste is:

[0035]

[0036] Where K represents the compaction degree of the roadbed, n represents the void ratio VCA of the aggregate, n min It represents the void ratio VCA when the construction solid waste aggregate reaches the standard maximum dry density under laboratory conditions.

[0037] S2, using the improved YOLOv8-seg model to segment the aggregate image and extract image indicators related to compaction;

[0038] Specifically, the improved YOLOv8-seg model is used to segment the aggregate image, distinguish the aggregate particles from the background information, segment each void on the aggregate surface image, and obtain image indicators related to compaction, namely, void ratio, void area, void perimeter, maximum void width, and void roundness.

[0039] Specifically, the collected construction solid waste images are made into a data set, and the data set is enhanced and divided into a training set, a test set, and a validation set. At the same time, the labeling tool labelme software is used to annotate the voids and aggregate targets in the image. The label of the training set is the category and coordinate position of each target in the image.

[0040] Specifically, the YOLOv8-seg model consists of three main parts: spine, neck, and head. The spine part is used as a feature extraction network and integrates the concept of CSPDarkNet. The backbone network of YOLOv8 replaces the C3 module of YOLOV5 with the C2f module and enhances the gradient flow through additional branches. SPPF (Spatial Pyramid Pooling-Fast) module, such as Figure 3 As shown in the figure, MaxPool2d represents the pooling layer, and CBS stands for Cross-Stage Partial Network, which mainly introduces cross-stage partial connections in the network structure of YOLOv8. As can be seen from the figure, the SPPF module combines three consecutive pooling layers, combines the output of each layer to ensure multi-scale fusion, and dynamically converts feature maps of different sizes to fixed sizes, so as to facilitate the simultaneous fusion of local and global feature information.

[0041] Specifically, the SKAPP (Separable Kernel Attention Pyramid) module is used to replace the SPPF module in the original YOLOv8-seg model, which solves the shortcomings of the SPPF module in detail feature extraction and improves the original model. SKAPPF integrates various aspects of SPPF and LSKA, such as Figure 4 As shown, k is the maximum receptive field, d is the expansion rate, Represents Hadamard product, where DW-D-Conv represents residual network and Conv represents convolutional block. As can be seen from the figure, the LSKA block adopts a residual design with four deep convolutional blocks, which is enhanced by a point-by-point convolutional block equipped with a large and small convolutional kernel, which is designed to linearly combine the output channels from the deep convolution. Compared with the original SPPF module, the SKAPPF module can enhance the feature fusion capability by fully utilizing and reusing multi-scale features, thereby enhancing the processing of fine-grained feature information.

[0042] Furthermore, by using the transfer learning technology to train the data set using the improved YOLOv8-seg model, a weight file will be obtained to achieve the segmentation of the image of recycled aggregate from construction solid waste.

[0043] Furthermore, according to the segmentation results of the instance segmentation model, the image indexes of the collected construction solid waste aggregate images are extracted, and the calculation of void ratio, void area, void perimeter and void roundness is specifically as follows:

[0044] Specifically, the calculation formula for each gap area is shown in formula (5):

[0045] A=α 2 ·Σn x,y ,x=1,2,3,…,w; y=1,2,3,…,h (5)

[0046] Here A is the calculated area of ​​the gap contour, w is the image width, h is the image height, n xy Indicates the mapping value of the pixel. When the pixel value is a specific value, n xy is 1, α is the scale conversion factor, representing the real length represented by a single pixel, unit: mm per pixel;

[0047] Specifically, the surface porosity calculation formula of the image is shown in formula (6):

[0048]

[0049] Where n s Represents the surface void ratio of the aggregate image, s a represents the total area of ​​the gaps in the image, and s represents the total area of ​​the image.

[0050] Specifically, the calculation formula for the perimeter of each gap is shown in formula (7):

[0051]

[0052] Here C vrepresents the perimeter of a certain gap, C represents the number of contour pixels of a certain gap in the image, w is the image width, and h is the image height.

[0053] Specifically, roundness can also reflect the shape angle and problems of recycled aggregates from construction solid waste. Its calculation formula is shown in formula (8):

[0054]

[0055] Where R represents the roundness of each gap, and S represents the number of pixels in each gap.

[0056] Specifically, the OpenCV built-in functions findContours and contourArea can be used to calculate the area size of the segmented material.

[0057] S3, training the image index related parameters and dosage, gradation and corresponding compaction degree using a machine learning model to obtain a compaction degree prediction model;

[0058] Specifically, the distribution of the image indicators obtained above is plotted on a statistical graph, and the curve is used for fitting, and the relevant parameters of the fitting curve are obtained. The relevant parameters of the gradation, dosage, and image indicators (void ratio, void area, void perimeter, and void roundness distribution fitting curve) of the construction solid waste recycled aggregate are converted into vectors as input variables of the machine learning model, and the prediction indicator is the compaction of the aggregate. The prediction of the compaction of the construction solid waste recycled aggregate is converted into a problem of predicting one numerical variable (dependent variable) based on six numerical variables (independent variables).

[0059] Furthermore, we train the neural network based on the data set, select the activation function as ReLU, and the loss function as Meansquare error. We set the number of neural network layers to 3 and the number of hidden layer nodes to 6432. We set the number of training times to 2000, the batch size to 64, the learning rate to 0.001, and the weight decay to 0.0008, and obtain the loss function and training times curve after 2000 training times, as shown in the figure below: Figure 5 As shown in the figure, it can be seen that the loss function value tends to be flat and reaches the lowest.

[0060] S4, using the construction solid waste recycled filler with known gradation and dosage to be tested to verify the compaction prediction model.

[0061] Furthermore, the surface images of the roadbed made of recycled aggregate from construction waste with known grading and dosage were collected on site, and the compaction prediction model was used to detect the compaction of the recycled filler from construction waste on site. At the same time, the compaction of the roadbed was tested using the sand injection method, and compared with the compaction obtained by the compaction prediction model of recycled aggregate from construction waste to analyze the error.

[0062] The second aspect of the present invention provides a device for detecting compaction degree of recycled aggregate from construction solid waste based on image processing, comprising:

[0063] The vibration compaction module is used to conduct vibration compaction tests on multiple groups of construction solid waste recycled aggregates with different gradations and dosages, and to match the gradations, dosages and aggregate compaction degrees respectively;

[0064] An image processing module is used to collect the surface image of the recycled aggregate from construction solid waste, segment the image of the construction solid waste using the improved YOLOv8-seg model, and extract image indicators: void ratio, area of ​​each void, void perimeter, maximum void width and void roundness, and obtain fitting curve parameters of image indicator distribution;

[0065] Roadbed compaction detection model: used to convert the gradation, dosage and image corresponding parameters into vectors, and process them using neural networks to obtain the compaction detection model of the construction solid waste recycled aggregate roadbed.

[0066] In summary, the present invention provides a method and device for compaction detection of recycled aggregates from construction solid waste based on image processing, including: using recycled aggregates from construction solid waste with different gradations and dosages to conduct vibration compaction tests and using a camera to collect surface images to obtain the compaction corresponding to the aggregates with different gradations and dosages; using an improved YOLOv8-seg model to segment the aggregate image, obtain the mask of each gap and extract image indicators related to compaction; using a machine learning model to train the image indicator parameters and the gradation, dosage and corresponding compaction to obtain a compaction prediction model; using a construction solid waste recycled filler with known gradation and dosage to be tested to verify the compaction prediction model. The present invention can conduct fast, simple and non-destructive detection of roadbed compaction of recycled aggregates from construction solid waste through image processing, machine learning and other methods, reduce the input of manpower and material resources, and improve industry benefits.

[0067] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for compaction detection of recycled aggregate from construction solid waste based on image processing. Its characteristics are: S1, using recycled aggregates of construction solid waste with different gradations and dosages to conduct multiple indoor vibration compaction tests and use cameras to collect surface images to obtain the compaction degrees corresponding to aggregates with different gradations and dosages; S2, using the improved YOLOv8-seg model to segment the aggregate image and extract image indicators related to compaction; S3, training the image index, dosage, gradation and corresponding compaction degree using a machine learning model to obtain a compaction degree prediction model; S4, using the construction solid waste recycled filler with known gradation and dosage to be tested to verify the compaction prediction model.

2. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 1 is characterized by: Multiple indoor vibration compaction tests include the following contents: vibration compaction tests are carried out using recycled aggregates from construction solid waste with different gradations and dosages. The coarse aggregate void ratio VCA can be calculated based on the volume density of the aggregate mixture, and then the compaction degree of the recycled aggregate roadbed from construction solid waste can be obtained, and the compaction degree is matched with the aggregate gradation and dosage.

3. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 2 is characterized in that: the formula for calculating the aggregate void ratio VCA according to the volume density of the aggregate mixture is as follows: Where n represents the void ratio of aggregate, ρ represents the natural packing density of the aggregate mixture measured by the vibration compaction test, and ρ b Represents the composite gross volume density of the aggregate.

4. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 2 is characterized by: The calculation formula for the roadbed of recycled aggregate from construction solid waste is: Where K represents the compaction degree of the roadbed, n represents the void ratio VCA of the aggregate, n min It represents the void ratio VCA when the construction solid waste aggregate reaches the standard maximum dry density under laboratory conditions.

5. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 1 is characterized by: The improved YOLOv8-seg model is used to segment the aggregate image, distinguish the aggregate particles from the background information, segment each void on the aggregate surface image, and obtain image indicators related to compaction, namely, void ratio, void area, void perimeter, maximum void width and void roundness.

6. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 5 is characterized by: The collected surface images are made into a data set, and the data set is enhanced and divided into a training set, a test set, and a validation set. At the same time, the labeling tool Labelme software is used to annotate the gaps and aggregate targets in the image. The label of the training set is the category and coordinate position of each target in the image.

7. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 5 is characterized by: The YOLOv8-seg model is used to train the dataset to obtain a weight file to achieve the segmentation of construction solid waste recycled aggregate images.

8. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 5 is characterized by: The separable kernel attention pyramid SKAPP module is used to replace the SPPF module in the original YOLOv8-seg model to improve the original model.

9. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 5 is characterized by: According to the segmentation results of the instance segmentation model, the image indexes of the collected construction solid waste aggregate images are extracted. The calculation of void ratio, void area, void perimeter and void roundness is as follows: The calculation formula for each gap area is shown in formula (3): A=α 2 ·∑n x,y ,x=1,2,3,···,w;y=1,2,3,···,h (3) Where: A is the calculated area of ​​the gap contour, w is the image width, h is the image height, n xy Indicates the mapping value of the pixel. When the pixel value is a specific value, n xy is 1, α is the scale conversion factor, representing the real length represented by a single pixel, unit: mm per pixel; The calculation formula for the perimeter of each gap is shown in formula (4): Here C v represents the perimeter of a certain gap, C represents the number of contour pixels of a certain gap in the image, w is the image width, and h is the image height.

10. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 9, characterized in that: The OpenCV built-in functions findContours and contourArea are used to calculate the size of the gap area of ​​the segmented material.

11. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 6, characterized in that: The grading, dosage, and image index related parameters of recycled aggregates from construction solid waste are converted into vectors as input variables of the machine learning model, and the prediction indicator is the compaction degree of aggregates. The prediction of the compaction degree of recycled aggregates from construction solid waste is converted into a problem of predicting one numerical variable based on six numerical variables.

12. The method for detecting compaction degree of recycled aggregate from construction solid waste based on image processing according to claim 11, characterized in that: Train the neural network based on the dataset, select ReLU as the activation function and Mean square error as the loss function. Set the number of neural network layers to 3 and the number of hidden layer nodes to 6432.

13. A compaction detection device for recycled aggregate from construction solid waste based on image processing, characterized in that: include: The vibration compaction module is used to conduct vibration compaction tests on multiple groups of construction solid waste recycled aggregates with different gradations and dosages, and to match the gradations, dosages and aggregate compaction degrees respectively; An image processing module is used to collect the surface image of the recycled aggregate from construction solid waste, segment the image of the construction solid waste using the improved YOLOv8-seg model, and extract image indicators: void ratio, area of ​​each void, void perimeter, maximum void width and void roundness, and obtain fitting curve parameters of image indicator distribution; Roadbed compaction detection model: used to convert the gradation, dosage and image corresponding parameters into vectors, and process them using neural networks to obtain the compaction detection model of the construction solid waste recycled aggregate roadbed.

14. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is run, the device where the non-volatile storage medium is located is controlled to execute the method for compaction detection of recycled aggregate from construction solid waste based on image processing as described in any one of claims 1 to 12.

15. An electronic device for detecting compaction of recycled aggregate from construction solid waste based on image processing, characterized in that: It comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions, when executed, execute the method for compaction detection of recycled aggregate from construction solid waste based on image processing as described in any one of claims 1 to 12.

Images