Concrete steel fiber content detection method and device and storage medium
By using image processing and deep learning models to extract features from concrete cross-sectional images and train them, the problem of detecting the steel fiber content in ultra-high performance concrete has been solved, achieving efficient and accurate automatic detection.
Patent Information
- Application Number
- CN202410764479.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-13
- Publication Date
- 2025-11-18
AI Technical Summary
In existing technologies, it is difficult to detect the steel fiber content in ultra-high performance concrete. Sampling is difficult and poses personal dangers. The work efficiency is low and it is impossible to detect the set concrete.
Image capture, image preprocessing, information annotation, and deep learning models (YOLOv5 and CNN convolutional neural network) are used to extract and train features from concrete cross-section images to achieve automatic detection of steel fiber content in concrete.
This improves the accuracy and efficiency of concrete steel fiber content detection and avoids the dangers associated with manual sampling.
Smart Images
Figure CN120971409A_ABST
Abstract
Description
Technical Field
[0001] This invention mainly relates to the field of concrete steel fiber content detection technology, specifically to a method, device, and storage medium for detecting concrete steel fiber content. Background Technology
[0002] Ultra-high performance concrete has a high steel fiber content, making accurate and efficient detection of this content challenging. Currently, the steel fiber content in concrete is determined before the concrete sets by separating various materials. This method is difficult for workers to sample, carries the risk of injury during sampling, and is inefficient. Furthermore, it cannot test the steel fiber content of set concrete. Summary of the Invention
[0003] The technical problem to be solved by the present invention is to provide a method, device and storage medium for detecting the steel fiber content in concrete, in order to address the shortcomings of the prior art.
[0004] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for detecting the steel fiber content in concrete, comprising the following steps:
[0005] Multiple original concrete cross-section images are obtained by taking images of the concrete cross-section from multiple angles using a preset image capturing device.
[0006] Based on the image specification requirements, multiple original concrete cross-section images were preprocessed to obtain multiple optimized concrete cross-section images.
[0007] Information is annotated on each of the optimized concrete cross-section images, and an XML file is generated from the annotated multiple concrete cross-section images.
[0008] A YOLOv5 detection model is constructed, and the XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers.
[0009] An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network.
[0010] The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content.
[0011] Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
[0012] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: A concrete steel fiber content detection device, comprising:
[0013] The imaging module is used to capture images of the concrete cross-section from multiple angles using a preset image imaging device, thereby obtaining multiple original concrete cross-section images.
[0014] The image preprocessing module is used to preprocess multiple original concrete cross-section images according to image specification requirements to obtain multiple optimized concrete cross-section images.
[0015] The annotation module is used to annotate the information of each of the optimized concrete cross-section images and generate an XML file from the annotated multiple concrete cross-section images.
[0016] The steel fiber content detection module is used to construct a YOLOv5 detection model. The XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers.
[0017] An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network.
[0018] The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content.
[0019] Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
[0020] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: a concrete steel fiber content detection device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the concrete steel fiber content detection method described above is implemented.
[0021] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the concrete steel fiber content detection method as described above.
[0022] The beneficial effects of this invention are as follows: multiple angles of the concrete cross-section are captured to obtain multiple original concrete cross-section images. These images are then optimized, labeled with information, and an XML file is generated. Feature extraction is performed on the XML file using the YOLOv5 detection model. The extracted preliminary features of the concrete steel fibers are used to initially train a CNN convolutional neural network, enabling the CNN to obtain preliminary parameters. The CNN is then trained again using the XML file, significantly improving its recognition accuracy. The CNN convolutional neural network in this invention can automatically detect and analyze concrete cross-section images, improving the efficiency of steel fiber content detection. Attached Figure Description
[0023] Figure 1 This is a schematic flowchart of a concrete steel fiber content detection method according to an embodiment of the present invention;
[0024] Figure 2 This is a block diagram of a concrete steel fiber content detection device provided in an embodiment of the present invention. Detailed Implementation
[0025] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0027] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0028] like Figure 1 As shown in the figure, this invention provides a method for detecting the steel fiber content in concrete, comprising the following steps:
[0029] Multiple original concrete cross-section images are obtained by taking images of the concrete cross-section from multiple angles using a preset image capturing device.
[0030] Based on the image specification requirements, multiple original concrete cross-section images were preprocessed to obtain multiple optimized concrete cross-section images.
[0031] Information is annotated on each of the optimized concrete cross-section images, and an XML file is generated from the annotated multiple concrete cross-section images.
[0032] A YOLOv5 detection model is constructed, and the XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers.
[0033] An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network.
[0034] The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content.
[0035] Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
[0036] Specifically, the requirements for cutting concrete sections are: using high-strength cutting equipment, cut no less than 3 sections from the concrete specimen.
[0037] The requirements for capturing images of concrete cross-sections from multiple angles are: use a high-definition camera to capture images of each cross-section of the concrete to obtain high-resolution images.
[0038] In the above embodiments, multiple angles of the concrete cross-section are captured to obtain multiple original concrete cross-section images. The images are then optimized, labeled with information, and an XML file is generated. The YOLOv5 detection model is used to extract features from the XML file. The extracted preliminary features of the concrete steel fibers are used to initially train the CNN convolutional neural network, enabling the CNN to obtain preliminary parameters. The CNN is then trained again using the XML file, which significantly improves its recognition accuracy. The CNN convolutional neural network in this invention can automatically detect and analyze concrete cross-section images, improving the efficiency of steel fiber content detection.
[0039] Preferably, the step of preprocessing multiple original concrete cross-section images according to image specification requirements to obtain multiple optimized concrete cross-section images specifically involves:
[0040] Multiple original concrete cross-section images are cropped according to a set size to obtain multiple cropped concrete cross-section images.
[0041] Multiple cropped concrete cross-section images were normalized to obtain multiple normalized concrete cross-section images.
[0042] Data augmentation was performed on multiple normalized concrete cross-section images to obtain multiple optimized concrete cross-section images.
[0043] The image preprocessing process is as follows:
[0044] 1. Resizing: Adjust the image to the standard size required by the neural network. Generally, neural networks accept fixed-size image inputs, so the images to be recognized need to be resized to the same size.
[0045] 2. Cropping: Cropping the image removes irrelevant parts or centers the main content. This reduces noise and draws attention to the image.
[0046] 3. Normalization: To eliminate the influence of different dimensions between indicators, data normalization is necessary to ensure comparability. After normalization, the raw data brings all indicators to the same order of magnitude, making them suitable for comprehensive comparative evaluation. Using non-linear normalization reduces the distribution distance of pixel values, making subsequent processing more convenient and accurate.
[0047] 4. Data augmentation: By rotating, flipping, scaling, and other operations on the original image, the diversity of the training data is increased, thereby improving the model's generalization ability.
[0048] In the preprocessing stage, normalization methods include linear normalization and nonlinear normalization. Linear normalization scales the pixel values by dividing them by 255, while nonlinear normalization methods, such as logarithmic transformation, are better suited to different situations. Since the grayscale difference between the steel fiber and the ultra-high performance concrete cross-section background is relatively small, nonlinear normalization is chosen here for image preprocessing to improve model performance and accelerate convergence.
[0049] Preferably, the step of annotating the optimized images of each concrete cross-section and generating an XML file from the annotated images specifically involves:
[0050] Using the Labelimg tool, category and location information are labeled in the optimized images of each concrete section using rectangular bounding boxes. The labeled information is saved in VOC format and an XML file is generated.
[0051] In the above embodiments, based on the collected classification network dataset, the category and location information are labeled on the image with rectangles using Labellimg software, and the labeling information is saved in VOC format to generate an XML file, which can improve the accuracy of feature extraction training.
[0052] Preferably, the CNN convolutional neural network includes convolutional layers and activation functions. The convolutional kernels of the convolutional layers are changed from 3×3 to 7×7 to obtain dilated convolutional layers, and the activation function is introduced into the dilated convolutional layers.
[0053] The CNN convolutional neural network further includes a branch pooling layer and a fully connected layer, with the dilated convolutional layer sequentially connected to the branch pooling layer and the fully connected layer.
[0054] Preferably, the step of inputting the XML file into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network for concrete steel fiber content specifically involves:
[0055] The XML file is input into the trained CNN convolutional neural network. Feature extraction training is performed on the images in the XML file using the receptive field expression of the dilated convolutional layer. The receptive field calculation formula is:
[0056] ri=d(k-1)+ri-1,
[0057] Where k represents the kernel size, ri represents the dilation rate, and d represents the stride.
[0058] The semantic information of the features is fused through the fusion expression of the branch pooling layer. The fusion expression is as follows:
[0059]
[0060] Among them, y p The output of the branch pooling layer is represented by B, where B represents the number of parallel branches. It should be understood that the branch pooling layer is used to fuse semantic information from different parallel branches and avoid introducing additional parameters.
[0061] Finally, the output of the branch pooling layer is mapped to the output category through the fully connected layer to obtain the CNN convolutional neural network for concrete steel fiber content.
[0062] It should be understood that in a CNN (Convolutional Neural Network), the convolutional layer is the key component used to extract image features. The convolution operation extracts features by sliding a convolutional kernel across the image. The kernel calculates a weighted sum for each local region in the input image, thereby generating a feature map.
[0063] After the convolutional layer, an activation function (such as ReLU) is usually applied to introduce non-linear characteristics to increase the expressive power of the network.
[0064] Pooling layers are used to reduce the dimensionality of feature maps, decrease computation, and enhance the robustness of the model. Average pooling layers fuse receptive field information from three branches, improving multi-scale prediction and recognition accuracy. Multi-branch convolutional layers include dilated convolutions, batch normalization (BN) layers, and ReLU activation layers. The dilated convolutions in the three parallel branches have the same kernel size but different dilation rates. Each dilated convolution kernel is 3×3, and the dilation rates d for different branches are 1, 3, and 5, respectively. Dilated convolutions support exponential expansion of the receptive field without loss of resolution. In the dilated convolution operation, the kernel elements are spaced, and the spatial size depends on the dilation rate.
[0065] Finally, after multiple convolutional and pooling layers, a fully connected layer is typically added to map the extracted features to the final output category. The fully connected layer achieves a non-linear combination of features through weight matrices and activation functions. Features of the image are gradually extracted through multiple layers of convolution and pooling operations. Once the concrete steel fiber content CNN (i.e., the model) can identify the location of the reinforcing bars, relevant information such as length, diameter, and cross-sectional area can be extracted. This typically involves drawing bounding boxes on the feature map output by the model or extracting this information directly from the model's predictions.
[0066] Analyze the model output to identify regions labeled "steel fibers," which represent the location and quantity of steel fibers. If the model's predicted length does not match the actual measured length, the model or dataset needs adjustment. Based on the preliminary analysis results, further adjustments and optimizations to the model or dataset may be necessary to ensure the accuracy of the final output.
[0067] Preferably, after obtaining the CNN convolutional neural network for the steel fiber content of concrete, the method further includes the following steps:
[0068] The CNN convolutional neural network containing the steel fiber content of the concrete is deployed in a specified application.
[0069] It should be understood that this process involves application and deployment: after all steps are completed, the trained model can be deployed to practical applications, such as concrete structure design or other related fields.
[0070] Finally, the results output by the model can be used for data analysis and comparison. The images can be compared with template sample images or images of different steel fibers to determine their distribution. The results can also be compared with the performance data of steel fiber content in concrete to determine the impact of the test data and the content of steel fibers in concrete on its slump, flexural strength and compressive strength.
[0071] like Figure 2 As shown, an embodiment of the present invention provides a concrete steel fiber content detection device, comprising:
[0072] The imaging module is used to capture images of the concrete cross-section from multiple angles using a preset image imaging device, thereby obtaining multiple original concrete cross-section images.
[0073] The image preprocessing module is used to preprocess multiple original concrete cross-section images according to image specification requirements to obtain multiple optimized concrete cross-section images.
[0074] The annotation module is used to annotate the information of each of the optimized concrete cross-section images and generate an XML file from the annotated multiple concrete cross-section images.
[0075] The steel fiber content detection module is used to construct a YOLOv5 detection model. The XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers.
[0076] An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network.
[0077] The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content.
[0078] Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
[0079] Preferably, in the image preprocessing module, multiple original concrete cross-section images are preprocessed according to image specification requirements to obtain multiple optimized concrete cross-section images, specifically as follows:
[0080] Multiple original concrete cross-section images are cropped according to a set size to obtain multiple cropped concrete cross-section images.
[0081] Multiple cropped concrete cross-section images were normalized to obtain multiple normalized concrete cross-section images.
[0082] Data augmentation was performed on multiple normalized concrete cross-section images to obtain multiple optimized concrete cross-section images.
[0083] This invention provides a concrete steel fiber content detection device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the concrete steel fiber content detection method as described above.
[0084] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the concrete steel fiber content detection method as described above.
[0085] Optionally, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the concrete steel fiber content detection method as described above.
[0086] Optionally, another embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the vulnerability scanning method described above.
[0087] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0088] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0089] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.
[0090] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.
[0091] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0092] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. This is understood to mean that the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0093] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for detecting the steel fiber content in concrete, characterized in that, Includes the following steps: Multiple original concrete cross-section images are obtained by taking images of the concrete cross-section from multiple angles using a preset image capturing device. Based on the image specification requirements, multiple original concrete cross-section images were preprocessed to obtain multiple optimized concrete cross-section images. Information is annotated on each of the optimized concrete cross-section images, and an XML file is generated from the annotated multiple concrete cross-section images. A YOLOv5 detection model is constructed, and the XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers. An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network. The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content. Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
2. The method for detecting the steel fiber content in concrete according to claim 1, characterized in that, The process involves preprocessing multiple original concrete cross-section images according to image specification requirements to obtain multiple optimized concrete cross-section images. Specifically: Multiple original concrete cross-section images are cropped according to a set size to obtain multiple cropped concrete cross-section images. Multiple cropped concrete cross-section images were normalized to obtain multiple normalized concrete cross-section images. Data augmentation was performed on multiple normalized concrete cross-section images to obtain multiple optimized concrete cross-section images.
3. The method for detecting the steel fiber content in concrete according to claim 1, characterized in that, The step of annotating the optimized images of each concrete cross-section and generating an XML file from the annotated images is as follows: Using the Labelimg tool, category and location information are labeled in the optimized images of each concrete section using rectangular bounding boxes. The labeled information is saved in VOC format and an XML file is generated.
4. The method for detecting the steel fiber content in concrete according to claim 1, characterized in that, The CNN convolutional neural network includes convolutional layers and activation functions. The convolutional kernel of the convolutional layer is changed from 3×3 to 7×7 to obtain an expanded convolutional layer, and the activation function is introduced into the expanded convolutional layer. The CNN convolutional neural network further includes a branch pooling layer and a fully connected layer, with the dilated convolutional layer sequentially connected to the branch pooling layer and the fully connected layer.
5. The method for detecting the steel fiber content in concrete according to claim 4, characterized in that, The step of inputting the XML file into the trained CNN convolutional neural network for training to obtain the concrete steel fiber content CNN convolutional neural network is as follows: The XML file is input into the trained CNN convolutional neural network. Feature extraction training is performed on the images in the XML file using the receptive field expression of the dilated convolutional layer. The receptive field calculation formula is: ri=d(k-1)+ri-1, Where k represents the kernel size, ri represents the dilation rate, and d represents the convolution stride; The semantic information of the features is fused through the fusion expression of the branch pooling layer. The fusion expression is as follows: Among them, y p This represents the output of the branch pooling layer, where B represents the number of parallel branches. The output of the branch pooling layer is mapped to the output category through the fully connected layer to obtain the CNN convolutional neural network for concrete steel fiber content.
6. The method for detecting the steel fiber content in concrete according to claim 1, characterized in that, After obtaining the CNN convolutional neural network for the steel fiber content of concrete, the following steps are also included: The CNN convolutional neural network containing the steel fiber content of the concrete is deployed in a specified application.
7. A device for detecting the steel fiber content in concrete, characterized in that, include: The imaging module is used to capture images of the concrete cross-section from multiple angles using a preset image imaging device, thereby obtaining multiple original concrete cross-section images. The image preprocessing module is used to preprocess multiple original concrete cross-section images according to image specification requirements to obtain multiple optimized concrete cross-section images. The annotation module is used to annotate the information of each of the optimized concrete cross-section images and generate an XML file from the annotated multiple concrete cross-section images. The steel fiber content detection module is used to construct a YOLOv5 detection model. The XML file is input into the YOLOv5 detection model for feature extraction to obtain preliminary features of concrete steel fibers. An initial CNN convolutional neural network is constructed, and the preliminary features of the concrete steel fiber are input into the CNN convolutional neural network for training, resulting in a trained CNN convolutional neural network. The XML file is input into the trained CNN convolutional neural network for training to obtain the CNN convolutional neural network with concrete steel fiber content. Import the concrete cross-section image to be identified, input the concrete cross-section image to be identified into the concrete steel fiber content CNN convolutional neural network, and output the concrete steel fiber content information of the concrete cross-section image through the concrete steel fiber content CNN convolutional neural network.
8. The concrete steel fiber content detection device according to claim 7, characterized in that, In the image preprocessing module, multiple original concrete cross-section images are preprocessed according to image specification requirements to obtain multiple optimized concrete cross-section images, specifically: Multiple original concrete cross-section images are cropped according to a set size to obtain multiple cropped concrete cross-section images. Multiple cropped concrete cross-section images were normalized to obtain multiple normalized concrete cross-section images. Data augmentation was performed on multiple normalized concrete cross-section images to obtain multiple optimized concrete cross-section images.
9. A device for detecting the steel fiber content in concrete, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the concrete steel fiber content detection method as described in any one of claims 1 to 6.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the concrete steel fiber content detection method as described in any one of claims 1 to 6.