A method and system for identifying a concrete test block
By acquiring high-dimensional feature vectors of concrete test blocks and performing deep learning algorithm matching and recognition, the problem of information loss and confusion in concrete test block management was solved, and the accurate identification of test blocks and the standardization of the testing process were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA ACAD OF BUILDING RES
- Filing Date
- 2026-03-13
- Publication Date
- 2026-06-26
AI Technical Summary
In existing technologies, the management of concrete test blocks mainly relies on manual marking and paper records, which makes the information easy to wear out, become blurred and fall off, and the records easy to be lost or tampered with. It is difficult to achieve rapid collection and digital management of test block information, and cannot meet the requirements of modern engineering construction for the efficiency and accuracy of material quality control.
A concrete specimen identification method is adopted. By acquiring high-dimensional feature vectors of sampling and verification images, deep learning algorithms are used for matching and identification. Similarity is calculated by combining Euclidean distance, and a similarity threshold is set to determine the identity of the concrete specimen. A concrete specimen feature database is established for information association and storage.
This method enables accurate identification of concrete test blocks, solves the problem of test block confusion, promotes the standardization of the testing process, and improves the accuracy and efficiency of testing.
Smart Images
Figure CN122289736A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of concrete testing, specifically relating to a method and system for identifying concrete test blocks. Background Technology
[0002] Concrete test blocks are crucial materials in building construction and quality acceptance. Their sampling, identification, and tracking management are essential for ensuring project quality and achieving accountability. Currently, the management of concrete test blocks mainly relies on manual marking and paper records, where information such as numbers, project locations, and dates are handwritten on the surface of the test blocks. This method suffers from problems such as easily worn-out and faded writing, and records being easily lost or altered, leading to the loss or confusion of test block identification information and causing difficulties in subsequent strength testing, quality traceability, and liability determination. Furthermore, traditional methods are insufficient for the rapid collection and digital management of test block information, failing to meet the efficiency and accuracy requirements of modern engineering construction for material quality control. Summary of the Invention
[0003] Therefore, the technical problem to be solved by the present invention is to provide a method and system for identifying concrete test blocks, so as to verify the concrete test blocks and ensure the consistency of the concrete test blocks in the processes of sampling, inspection and storage.
[0004] This invention proposes a method for identifying concrete test blocks. In Scheme 1, the method for identifying concrete test blocks includes: S1: During the sampling stage, obtain sampling images of concrete test blocks and obtain high-dimensional feature vectors of the sampling images; S2: Associate the high-dimensional feature vector of the sampled image obtained in step S1 with the identity information of the sampled image and store it in the concrete test block feature database; S3: During the verification stage, obtain the verification image of the concrete test block to be verified, and obtain the high-dimensional feature vector of the verification image. S4: Match the high-dimensional feature vector of the verification image with the high-dimensional feature vector of the sampling image in the concrete test block feature database, and determine the identity of the concrete test block based on the matching result.
[0005] In Scheme 2 based on Scheme 1, step S4 includes: S41: Calculate the similarity between the high-dimensional feature vector of the verification image and the high-dimensional feature vector of the sampling image in the concrete test block feature database based on Euclidean distance; S42: Determine the matching result based on the similarity; S43: Determine the identity of the concrete test block based on the matching results.
[0006] In Scheme 3 based on Scheme 2, step S42 includes: A preset similarity threshold is set. If the calculated similarity is greater than the threshold, the match is considered successful; if the calculated similarity is less than the threshold, the match is considered unsuccessful.
[0007] In Scheme 4, which is based on Scheme 3, step S43 includes: If a match is successful, the identity information corresponding to the high-dimensional feature vector of the matched sampled image will be output. If the matching fails, the high-dimensional feature vector corresponding to the verification image is associated with the identity information and stored in the concrete test block feature database.
[0008] In Scheme 5, which is based on any one of Schemes 1 to 4, step S1 includes: S11: During the sampling stage, obtain sampling images of concrete test blocks; S12: Perform grayscale and noise reduction processing on the sampled image; S13: Based on the deep learning algorithm, obtain the high-dimensional feature vector based on the sampled image processed in step S12.
[0009] In Scheme 6, which is based on Scheme 5, step S3 includes: S31: During the verification phase, obtain verification images of the concrete test blocks to be verified; S32: Perform grayscale and noise reduction processing on the verification image; S33: Based on the deep learning algorithm, obtain the high-dimensional feature vector from the verification image processed in step S32.
[0010] In Scheme 7, which is based on any one of Schemes 1 to 4, the concrete test block is formed by pouring concrete into a mold with at least one open side and solidifying it, the concrete test block having a first surface formed on the open side of the mold. The sampled image includes at least an image of the first surface; and The verification image includes at least an image of the first surface.
[0011] In Scheme 8, which is based on any one of Schemes 1 to 4, the concrete test block identification method further includes: S01: During the sampling phase, acquire all images of the surface of the concrete test block; S02: Based on the image recognition algorithm, the concrete test blocks are screened using the images obtained in step S01, and concrete test blocks with surface defects are excluded.
[0012] The present invention also proposes a concrete test block identification system. In Scheme 9, the concrete test block identification system of the present invention applies the concrete test block identification method described in any one of Schemes 1 to 8 above.
[0013] In Scheme 10, which is based on Scheme 9, the identification system includes: A structured light lens is an optical element with structured light projection and imaging functions, wherein the structured light lens acquires sampling images and verification images; The processor processes the sampled image and the verification image; The memory contains the concrete test block feature database.
[0014] Beneficial effects: In the sampling stage, sampling images of concrete test blocks are acquired, and in the verification stage, verification images of concrete test blocks to be verified are acquired. By matching the high-dimensional feature vectors of the sampling images and the high-dimensional feature vectors of the verification images, it is possible to identify whether the concrete test block to be verified is consistent with the concrete test block in the sampling stage. This invention effectively solves the problem of easy confusion of concrete test blocks in the modern building testing industry by accurately confirming the identity of concrete test blocks, and significantly promotes the standardization of the testing process. Attached Figure Description
[0015] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.
[0016] Figure 1 This is a flowchart of a concrete test block identification method according to an embodiment of the present invention; Figure 2 This is a flowchart of a concrete test block feature verification and matching method according to an embodiment of the present invention; Figure 3 This is a flowchart of a concrete test block feature extraction method according to an embodiment of the present invention; Figure 4 This is a flowchart of a method for obtaining high-dimensional feature vectors of concrete test blocks according to an embodiment of the present invention; Figure 5 This is a flowchart of a concrete test block identification method with a test block screening step according to an embodiment of the present invention; Figure 6 This is a schematic diagram of a concrete test block image capture according to an embodiment of the present invention; Figure 7 This is a schematic diagram of a concrete test block identification system according to an embodiment of the present invention; Figure 8 This is a schematic diagram of a concrete specimen feature extraction method according to an embodiment of the present invention.
[0017] Figure label: 1. Binocular lens; 2. Concrete test block; 3. Background board; 4. Structured light projection element. Detailed Implementation
[0018] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. The principles and features of the present invention are described below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other. The embodiments given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0019] Concrete test blocks are concrete samples prepared according to standard manufacturing processes. They are usually cubes, cylinders, or prisms and are used to test the mechanical properties and durability of concrete. They are one of the core bases for quality control in construction projects.
[0020] Concrete test blocks are typically randomly selected at the concrete pouring site, where concrete is poured into molds to form the test blocks. These on-site sampled blocks need to be taken back for curing before testing. A single project requires the collection of a large number of concrete test blocks, and accidental mixing or human intervention during transportation, storage, and testing can lead to inconsistencies between the sampled and tested blocks, resulting in inaccurate test results. To address these issues, this invention proposes a method and system for identifying concrete test blocks.
[0021] This invention provides a method for identifying concrete test blocks. Figure 1 This is a flowchart of a concrete test block identification method according to an embodiment of the present invention. In some embodiments of the present invention, such as... Figure 1 As shown, the concrete test block identification method includes: S1: During the sampling stage, obtain sampling images of concrete test blocks and obtain high-dimensional feature vectors of the sampling images. S2: Associate the high-dimensional feature vector of the sampled image obtained in step S1 with the identity information of the sampled image and store it in the concrete test block feature database; S3: During the verification phase, obtain the verification image of the concrete test block to be verified, and obtain the high-dimensional feature vector of the verification image. S4: Match the high-dimensional feature vector of the verification image with the high-dimensional feature vector of the sampled image in the concrete test block feature database, and determine the identity of the concrete test block based on the matching result.
[0022] In the embodiments of the present invention, images of concrete test blocks (i.e., sampling images and verification images) are acquired at the sampling stage and the verification stage, respectively. Based on the images, digitized high-dimensional feature vectors are obtained. By comparing the high-dimensional feature vectors of the sampling image and the verification image, it can be determined whether the sampling image and the verification image come from the same concrete test block, thereby improving the consistency of concrete test blocks throughout the entire testing cycle. This effectively solves the problem of easy confusion of concrete test blocks in the modern building testing industry and significantly promotes the standardization of the testing process.
[0023] As one specific application scenario, during the on-site sampling of concrete test blocks (i.e., the sampling stage), the identification method of this embodiment of the invention is used to acquire sampling images. Before the concrete test blocks are sent to the laboratory for testing, a verification image of the concrete test block is acquired using the sampling method of this embodiment of the invention. The verification image and the sampling image are then matched and identified using the identification method of this embodiment of the invention to confirm the identity of the concrete test block and avoid confusion. This approach allows for convenient retrieval of the concrete test block's identity information, simplifying the testing process, and also prevents the concrete test blocks from being switched, improving testing accuracy.
[0024] The sampling stage refers to the stage of taking samples of concrete test blocks on site, which is the initial stage of obtaining concrete test blocks.
[0025] The verification phase includes, but is not limited to, all stages of transporting, testing, and storing concrete test blocks to verify their consistency.
[0026] The concrete test block feature database consists of high-dimensional feature vectors (and corresponding identity information) of one or more concrete test blocks obtained in step S1. One possible storage method is to associate the high-dimensional feature vector of the same sampled image with its identity information and store both the high-dimensional feature vector and the identity information. Another possible storage method is to associate a sampled image with its high-dimensional feature vector and identity information and store both the sampled image, the high-dimensional feature vector, and the identity information. The data stored in the concrete test block feature database may include high-dimensional feature vectors and identity information, or it may include sampled images, high-dimensional feature vectors, and identity information, and may also include additional information with one-to-one correspondence.
[0027] Identification information is used to distinguish concrete test blocks. It may include, but is not limited to, information such as the image shooting angle, the geographical location at the time of shooting, the shooting time, the serial number of the concrete test block, and the source of the concrete test block.
[0028] High-dimensional feature vectors refer to data features obtained from images, and the specific acquisition method will be described in detail in subsequent embodiments.
[0029] In step S4, the "matching" method can be to compare the data in the feature database one by one with the high-dimensional feature vector of the verification image to obtain the matching result.
[0030] Figure 2 This is a flowchart of a concrete test block feature verification and matching method according to an embodiment of the present invention. In some embodiments of the present invention, such as... Figure 2 As shown, step S4 includes: S41: Calculate the similarity between the high-dimensional feature vector of the verification image and the high-dimensional feature vector of the sampled image in the concrete block feature database based on Euclidean distance; S42: Determine the matching result based on similarity; S43: Determine the identity of the concrete test block based on the matching results.
[0031] In the embodiments of the present invention, similarity is determined by Euclidean distance, and matching results are judged based on similarity. This provides a data-driven verification of the comparison and matching between the sampled image and the verification image, thereby improving the accuracy and reliability of the verification results.
[0032] The calculation method for Euclidean distance will be described in detail in subsequent embodiments of the present invention.
[0033] Images of concrete test blocks acquired under different lighting conditions, angles, and image acquisition devices will exhibit certain differences. Furthermore, concrete test blocks experience wear and tear during storage and transportation, leading to discrepancies between sampling and verification images of the same concrete test block. To address this issue, in some embodiments of the present invention, a preset similarity threshold can be implemented. If the calculated similarity is greater than the threshold, the match is considered successful; if the calculated similarity is less than the threshold, the match is considered unsuccessful.
[0034] In embodiments of the present invention, the verification image and the sampling image are not completely identical as the criterion for successful matching. Instead, a similarity threshold is set to avoid the error of judging the sampling image and the verification image of the same concrete test block as mismatched due to image differences caused by natural factors such as light, angle, equipment and wear.
[0035] One way to select the similarity threshold is by obtaining it through experiments.
[0036] In some embodiments of the present invention, the similarity threshold can be optimized during the use of the identification method of the present invention. For example, if the initial similarity threshold is set to 80%, and after a period of use it is found that the similarity calculation results during the matching process are concentrated in the two ranges of 20%-60% and 90%-95%, it can be said that the similarity between the sampling image and the verification image of the same concrete test block is basically above 90%. Therefore, the similarity threshold can be increased by a certain margin, for example, from 80% to 90% (or 85%).
[0037] For cases where the feature database contains multiple sets of concrete test block data, in step S41, it is necessary to calculate the Euclidean distance between the verification image and all sampled images, and select the sampled image with the smallest Euclidean distance (i.e., the highest similarity) as the candidate matching object; in step S42, the matching result is determined based on the similarity with the candidate matching object, that is, if the similarity between the verification image and the candidate matching object is greater than the threshold, it is judged as a successful match, and if the similarity between the verification image and the candidate matching object is less than the threshold, it is judged as a failed match.
[0038] There are various ways to output and process the matching results. In some embodiments of the present invention, step S43 includes: If the match is successful, the identity information corresponding to the high-dimensional feature vector of the matched sampled image is output; if the match fails, the high-dimensional feature vector corresponding to the verification image and the identity information are associated and stored in the concrete test block feature database.
[0039] In embodiments of the present invention, upon successful matching, the identity information of the corresponding concrete test block is output, serving as an identity retrieval and verification function, facilitating rapid confirmation of the identity of the concrete test block to be verified. Upon failed matching, the concrete test block to be verified is directly used as a new sample, storing its high-dimensional feature vector and identity information. This facilitates the recording of new concrete test blocks appearing during the verification phase.
[0040] In some embodiments of the present invention, the verification result can be associated with subsequent procedures (e.g., experimental testing). For example, when the output result is a matching failure (or a matching success, but the matching result is not the target test block), subsequent procedures (e.g., experimental testing) are aborted.
[0041] In some embodiments of the present invention, a target test block can be set, namely the concrete test block to be tested. If the matching result shows that the matched identity information does not match the target test block, even if the verification image obtained in the verification stage is successfully matched in the feature database, it can be considered that the concrete test block to be verified is not the target test block, and it can be indicated that there is a possibility of confusion between the concrete test blocks. For example: there are four sets of high-dimensional feature vectors, namely A, B, C, and D, and their corresponding identity information in the feature database. In the verification stage, concrete test block A is to be tested (i.e., the target test block is A). After identification by the identification method of the present invention, the matching result shows that the concrete test block actually sent for testing (i.e., to be verified) matches the sampling image of C. It is then considered that the concrete test block to be verified does not match the target test block, and there is a possibility of confusion between concrete test block C and concrete test block A. As one implementation method, subsequent testing can be stopped accordingly until the identification result of the concrete test block sent for testing is concrete test block A.
[0042] Figure 3 This is a flowchart of a concrete specimen feature extraction method according to an embodiment of the present invention. In some embodiments of the present invention, such as... Figure 3 As shown, step S1 includes: S11: During the sampling stage, obtain sampling images of concrete test blocks; S12: Perform grayscale and noise reduction processing on the sampled image; S13: Based on the deep learning algorithm, obtain the high-dimensional feature vector based on the sampled image processed in step S12.
[0043] In embodiments of the present invention, the grayscale processing in step S12 can highlight the main features of the concrete test block image (such as texture, pores, etc.) and reduce data dimensionality and computational complexity, while the noise reduction processing can suppress related noise in the image. Through the processing in step S12, efficiency and accuracy can be significantly improved.
[0044] Figure 8 This is a schematic diagram of a concrete specimen feature extraction method according to an embodiment of the present invention, as shown below. Figure 3 A specific embodiment of the feature extraction method shown is as follows: Figure 8 As shown, firstly, by photographing the concrete test blocks and performing grayscale processing, the following results were obtained: Figure 8 The original image is then processed. To facilitate efficient subsequent processing, the original image is binarized, defining the contour-related parts as black and assigning a value of 1, while defining the background as white and assigning a value of 0, thus obtaining a binary image of the concrete block. Based on this, a denoising algorithm is applied to generate the final contour image. Finally, the contour features of the concrete are visualized, and a Cartesian coordinate system is established to quantitatively describe its contour features for subsequent identification and comparison. Figure 8 The coordinate scatter plot, pure coordinate axis display, and contour overlay plot demonstrate the process of contour feature comparison, visualize it, and establish a rectangular coordinate system to quantitatively describe it, resulting in... Figure 8 The last quantitative chart in the middle.
[0045] During the sampling stage, images of multiple concrete test blocks can be sampled simultaneously. The identity information and associated high-dimensional feature vectors of these sampled images are stored in a concrete test block feature database. The concrete test block feature database also stores historical data from previous sampling sessions. In some embodiments of the present invention, step S1 may further include: S14: Match the high-dimensional feature vector of the sampled image obtained in step S13 with the high-dimensional feature vector in the feature database. If the match is successful, exclude this concrete test block.
[0046] During the sampling phase, if the sampled image matches the data in the feature database, it indicates that the concrete specimen is highly similar to a previous concrete specimen and is difficult to distinguish using the deep learning algorithm of this invention. Excluding this concrete specimen during the sampling phase serves two purposes: firstly, it directly eliminates concrete specimens that cannot be distinguished from previous specimens, avoiding potential confusion; secondly, it prevents the reuse of previous concrete specimens, improving the reliability of the entire testing process.
[0047] "Exclusion" means discarding this concrete test block. Depending on the actual needs, a new concrete test block can be made or another concrete test block can be selected.
[0048] Figure 4 This is a flowchart of a method for obtaining high-dimensional feature vectors of concrete test blocks according to an embodiment of the present invention. Based on similar considerations, in some embodiments of the present invention, such as... Figure 4 As shown, step S3 includes: S31: During the verification phase, obtain verification images of the concrete test blocks to be verified; S32: Perform grayscale and noise reduction processing on the verification image; S33: Based on the deep learning algorithm, obtain the high-dimensional feature vector from the verification image processed in step S32.
[0049] In particular, the grayscale processing and noise reduction processing in steps S12 and S22 are carried out in the same way to avoid the reduction in the similarity of the high-dimensional feature vectors between the sampled image and the verification image due to algorithm differences.
[0050] During storage and transportation, concrete test blocks may experience normal wear and tear, leading to subtle changes on their primary surface. This reduces the similarity between the verification image and the sampling image of the same concrete test block. In some embodiments of the present invention, the direction of deviation should also be considered when determining the similarity between the verification image and the sampling image. Specifically, the change in surface features from the sampling image to the verification image should be unidirectional. For example, if the sampling image surface has four protrusions, and the verification image surface has three and five protrusions, the similarity obtained based on Euclidean distance calculation should be similar. However, it is obvious that the number of protrusions on the surface of the same concrete test block may decrease due to wear, but will not increase. Similarly, the height of the protrusions may decrease due to wear, but will not increase. Before determining similarity based on Euclidean distance calculation, objects in the feature database can be excluded based on the direction of deviation. This exclusion process can be performed using image recognition software.
[0051] The grayscale processing methods in steps S12 and S22 will be described in detail in subsequent embodiments.
[0052] In some embodiments of the invention, the concrete specimen is formed by pouring concrete into a mold with at least one open side and allowing it to solidify. The concrete specimen has a first surface formed on the open side of the mold. The sampling image includes at least an image of the first surface, and the verification image also includes at least an image of the first surface.
[0053] During the molding process, the surfaces of concrete specimens in contact with the mold tend to be more regular and similar, with minimal differences, making it difficult to distinguish different concrete specimens based on surface features. The first surface, however, corresponds to the open side of the mold, and its surface formation has a degree of freedom and randomness. Therefore, the first surface of different concrete specimens will exhibit more differentiated features, making it easier to distinguish between them.
[0054] In one implementation method, the concrete test block is a cube (such as a cuboid or cubic prism), and the corresponding mold is a cubic mold with one side open. With the open side of the mold facing upwards, concrete is poured into the mold through the open side. After the test block is formed, the upward-facing side is the first surface.
[0055] After concrete test blocks are formed and cured, they need to be tested for strength, impermeability, and other properties. If the concrete test blocks have defects during the forming stage, the test results will deviate from the actual quality level of the concrete due to these defects. Therefore, concrete test blocks with surface defects cannot be used for testing. If concrete test blocks with surface defects are transported to the laboratory for testing, on the one hand, new concrete test blocks need to be made, prolonging the testing cycle and wasting time; on the other hand, it is impossible to determine whether the surface defects were inherent in the concrete test blocks themselves or were damaged during delivery or testing. Figure 5 This is a flowchart of a concrete test block identification method with a test block screening step according to an embodiment of the present invention. To solve this problem, in the embodiments of the present invention, as follows... Figure 5 As shown, the concrete test block identification method also includes: S01: During the sampling phase, acquire all images of the surface of the concrete test block; S02: Based on the image recognition algorithm, the concrete test blocks are screened using the images obtained in step S01, and concrete test blocks with surface defects are excluded.
[0056] In an embodiment of the present invention, all images of the concrete test block are acquired during the sampling stage, and then surface defects are identified according to the image recognition algorithm. Concrete test blocks with surface defects can be directly excluded during the sampling stage, thereby improving detection efficiency.
[0057] The identification of surface defects in step S02 can be achieved based on the YOLO algorithm. This algorithm and its applications are relatively mature, and will not be described in detail here.
[0058] like Figure 5 As shown, steps S01 and S02 are executed before step S1.
[0059] Surface defects include, but are not limited to, cracks, honeycombing, pitting, chipped edges and corners, and voids.
[0060] In some embodiments of the present invention, for concrete test blocks without surface defects (i.e., concrete test blocks not excluded in step S01), the first surface is automatically identified from all images of the concrete test blocks obtained in step S01, and the first surface is cropped as the sampling image in step S1.
[0061] In an embodiment of the present invention, the first surface is cropped from the image obtained in step S01, eliminating the need for repeated shooting or manual selection of the first surface, thus simplifying the operation steps.
[0062] The identification and selection of the first surface can be based on various principles. One implementation involves comparing the various surfaces of the concrete specimen, with the surface showing the greatest difference being designated as the first surface. Another implementation involves forming the first surface without the aid of a mold, while the other surfaces come into contact with the mold during forming. Therefore, the first surface will possess unique surface features, which can be used to train a corresponding recognition algorithm for direct identification.
[0063] When acquiring images during the sampling and verification stages, it's difficult to ensure that the angles and distances of the acquired images are exactly the same, which may adversely affect the matching efficiency and accuracy of the verification image in the feature database. To address this issue, one possible implementation is to set a standard image shape. Both the captured (or cropped) sampling images and verification images are corrected to the standard image shape, thereby avoiding interference from irrelevant factors on image recognition and matching. The standard image shape can be a rectangle or a parallelogram, and can be selected and adjusted according to the actual effect. Correction of the sampling and verification images refers to correcting the outline shape of the images. For example, if the captured (or cropped) image is a parallelogram, it can be corrected to a rectangle.
[0064] The differences in the first surface of different concrete test blocks stem from the fact that the formation of the first surface is independent of molds, thus its formation has a certain degree of randomness. Therefore, if all surfaces of a concrete test block are formed using molds, it may be impossible to effectively identify different concrete test blocks. To solve this problem, in some embodiments of the present invention, during the sampling stage, image recognition technology is used to determine whether the first surface in the sampling image has demolding marks. If demolding marks are found, the concrete test block is excluded. In this way, the possibility of the first surface being formed by a mold can be ruled out, avoiding confusion or substitution of concrete test blocks that would go undetected.
[0065] For the first surface that does not naturally form using a mold, it is possible to create highly similar concrete test blocks by casting the first surface and demolding it with concrete of different qualities. To avoid this risk, after obtaining the verification image during the verification phase, image recognition technology is used to determine whether the first surface in the verification image has demolding marks. If demolding marks are detected, matching is stopped. In this way, the original sampled concrete test block can be prevented from being copied and replaced.
[0066] The identification of demolding features can be achieved by training an image recognition model, which will not be elaborated upon in this disclosure.
[0067] In some embodiments of the present invention, the sampling image and the verification image can be stereoscopic images with three-dimensional information of the concrete specimen surface, thereby providing more accurate and effective surface feature information. As one implementation, to facilitate surface feature calculation, the image can be obtained from the front, perpendicular to the concrete specimen surface; as another implementation, to highlight the surface features of the concrete specimen, the image of the concrete surface can be obtained along a direction at an angle to the concrete specimen surface.
[0068] In some embodiments of the present invention, the grayscale processing in steps S12 and S32 may be implemented as follows: The acquired colored concrete specimen images are converted from RGB color to grayscale images using the image grayscale weighted averaging method. The formula for the image grayscale weighted averaging method is as follows:
[0069] in, The horizontal coordinates of a pixel in an image represent its two-dimensional coordinates. The vertical coordinates of a pixel in an image. This indicates the position of the image points at the pixel level after grayscale conversion. grayscale value at that location This indicates that the red channel in the original color image is... Pixel value at that location, This indicates that the green channel in the original color image is... Pixel value at that location, This indicates that the blue channel in the original color image is... The pixel value at that location. One way to select the weighting coefficients is... b > a > c ,For example, a =0.299, b =0.587, c =0.114.
[0070] In steps S41 and S42, if there are two or more matching results with similarity, the weight coefficients (i.e.) can be adjusted. a , b and c Adjustments are made, and the above similar matching results are recalculated based on the grayscale formula after adjusting the weight coefficients, in order to determine the final unique matching result.
[0071] In some embodiments of the present invention, the noise reduction processing in steps S12 and S32 may be implemented as follows: The image is processed using Gaussian filtering. The formula for Gaussian filtering is as follows:
[0072] in, Represents the coordinates inside the Gaussian kernel. Indicates coordinates within the kernel The corresponding Gaussian kernel function weight values at that location. It represents the standard deviation.
[0073]
[0074] in, This indicates that the output image after Gaussian filtering is in coordinates The new pixel value of the location, and This represents the coordinate values used for traversal within the Gaussian kernel. Indicates relative to the center pixel Offset The original grayscale value of the pixel at the location.
[0075] The deep learning algorithm in step S13 (and step S33) can be a model trained on data. One possible training method for this model is as follows: Training Step 1: Establish an image database of concrete test blocks on a data platform (such as the Roboflow open-source database platform). (The images in this database can be grayscaled and denoised.) Annotate the contours and features of each concrete test block in this database to generate standardized annotation files, facilitating the construction of a structured dataset for model training. Annotation can be done using existing tools, such as Labelimg.
[0076] Training Step Two: Based on object detection and segmentation algorithms, the structured dataset constructed in Step One is input into the model for training. Iterative optimization using deep learning enables the model to automatically extract high-dimensional feature vectors from images, ultimately obtaining a feature extraction deep learning model with generalization capabilities. Existing algorithms, such as the YOLO series, can be used for object detection and segmentation.
[0077] The formula for calculating the Euclidean distance in step S41 is:
[0078] in, and This represents two high-dimensional feature vectors to be compared. express and The Euclidean distance between them and They represent and In the Component values in each dimension This represents the dimension of the feature vector.
[0079] One possible implementation method for acquiring the sampling image and verification image in steps S1 and S3 includes: Images are acquired using a structured light lens with binoculars, and the complete outline of the concrete test block in the image is extracted and filtered based on the Canny edge detection algorithm, thereby locating and cropping the target area to provide high-quality images for subsequent processing.
[0080] Figure 6 This is a schematic diagram of a concrete test block image capture according to an embodiment of the present invention, as shown below. Figure 6 As shown, the left side is the image directly acquired by the binocular camera, and the right side is the target area after processing based on the Canny edge detection algorithm.
[0081] This invention also proposes a concrete specimen identification system, which applies the concrete specimen identification method of any embodiment of this invention. Correspondingly, the identification system using the aforementioned identification method also possesses the same advantages, which will not be elaborated further here.
[0082] In some embodiments of the present invention, the identification system further includes a structured light lens, a processor, and a memory. The structured light lens is an optical element with structured light projection and imaging functions, and acquires sampling images and verification images. The processor processes the sampling images and verification images (but is not limited thereto). The concrete test block feature database is stored in the memory.
[0083] In an embodiment of the present invention, a structured light lens is used to acquire sampling images and verification images, which can obtain images with three-dimensional features of concrete test blocks and obtain the surface features of concrete test blocks.
[0084] In embodiments of the present invention, the structured light lens may have a binocular lens, which simultaneously captures images from different perspectives to improve the surface features of the concrete specimen.
[0085] Figure 7 This is a schematic diagram of a concrete test block identification system according to an embodiment of the present invention. The structured light projection element of the structured light lens and the binocular lens can be arranged in various ways. As one arrangement, such as... Figure 7 As shown, the structured light projection element 4 can be placed between the two lenses of the binocular lens 1, and the binocular lens 1 and the structured light projection element 4 are arranged in a horizontal straight line during shooting. However, it is not limited to this.
[0086] like Figure 7As shown, when using a structured light lens to acquire images of concrete specimen 2, a background plate 3 can be placed on the concrete specimen to make the image background uniform and easier to process. As one option, the background plate 3 can be white.
[0087] In some embodiments of the present invention, the concrete specimen identification system can be integrated into a mobile phone, using the mobile phone's lens as a structured light lens.
[0088] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.
[0089] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.
[0090] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.
[0091] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.
[0092] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0093] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for identifying concrete test blocks, characterized in that, include: S1: During the sampling stage, obtain sampling images of concrete test blocks and obtain high-dimensional feature vectors of the sampling images; S2: Associate the high-dimensional feature vector of the sampled image obtained in step S1 with the identity information of the sampled image and store it in the concrete test block feature database; S3: During the verification stage, obtain the verification image of the concrete test block to be verified, and obtain the high-dimensional feature vector of the verification image. S4: Match the high-dimensional feature vector of the verification image with the high-dimensional feature vector of the sampling image in the concrete test block feature database, and determine the identity of the concrete test block based on the matching result.
2. The concrete test block identification method according to claim 1, characterized in that, Step S4 includes: S41: Calculate the similarity between the high-dimensional feature vector of the verification image and the high-dimensional feature vector of the sampling image in the concrete test block feature database based on Euclidean distance; S42: Determine the matching result based on the similarity; S43: Determine the identity of the concrete test block based on the matching results.
3. The concrete test block identification method according to claim 2, characterized in that, Step S42 includes: A preset similarity threshold is set. If the calculated similarity is greater than the threshold, the match is considered successful; if the calculated similarity is less than the threshold, the match is considered unsuccessful.
4. The concrete test block identification method according to claim 3, characterized in that, Step S43 includes: If a match is successful, the identity information corresponding to the high-dimensional feature vector of the matched sampled image will be output. If the matching fails, the high-dimensional feature vector corresponding to the verification image is associated with the identity information and stored in the concrete test block feature database.
5. The method for identifying concrete test blocks according to any one of claims 1-4, characterized in that, Step S1 includes: S11: During the sampling stage, obtain sampling images of concrete test blocks; S12: Perform grayscale and noise reduction processing on the sampled image; S13: Based on the deep learning algorithm, obtain the high-dimensional feature vector based on the sampled image processed in step S12.
6. The concrete test block identification method according to claim 5, characterized in that, Step S3 includes: S31: During the verification phase, obtain verification images of the concrete test blocks to be verified; S32: Perform grayscale and noise reduction processing on the verification image; S33: Based on the deep learning algorithm, obtain the high-dimensional feature vector from the verification image processed in step S32.
7. The method for identifying concrete test blocks according to any one of claims 1-4, characterized in that, The concrete test block is formed by pouring concrete into a mold with at least one open side and solidifying it, the concrete test block having a first surface formed on the open side of the mold; The sampled image includes at least an image of the first surface; and The verification image includes at least an image of the first surface.
8. The method for identifying concrete test blocks according to any one of claims 1-4, characterized in that, The concrete test block identification method also includes: S01: During the sampling phase, acquire all images of the surface of the concrete test block; S02: Based on the image recognition algorithm, the concrete test blocks are screened using the images obtained in step S01, and concrete test blocks with surface defects are excluded.
9. A concrete test block identification system, characterized in that, The concrete test block identification method according to any one of claims 1-8 is applied.
10. The concrete test block identification system according to claim 9, characterized in that, The identification system includes: A structured light lens is an optical element with structured light projection and imaging functions, wherein the structured light lens acquires sampling images and verification images; The processor processes the sampled image and the verification image; The memory contains the concrete test block feature database.