Vehicle gluing quality detection method and device, medium and electronic equipment

Abstract

The invention provides a vehicle gluing quality detection method and device, a medium and electronic equipment. According to the embodiment of the invention, a first segmentation mask of a region-of-interest image in a glued adhesive tape image is applied to a trained preset adhesive tape convolutional neural network model, and a second segmentation mask marked with target feature information is obtained; the second segmentation mask is applied to a trained preset adhesive tape flow state neural network model, a target binary image is obtained, and a foreground adhesive tape area is marked in the target binary image; and comparing the foreground adhesive tape region in the target binary image with the foreground adhesive tape region in the target binary image by taking a foreground adhesive tape region in a preset reference binary image as a reference, and determining whether the quality of the adhesive tape coated in the region-of-interest image is qualified or not based on a comparison result. The target feature information of the adhesive tape region is accurately and efficiently extracted, and the accurate instance segmentation of the adhesive tape region is realized.

Description

Technical Field

[0001] The present application relates to the field of image recognition technology, and in particular to a method, device, medium and electronic equipment for detecting the quality of vehicle glue coating. Background Art

[0002] In an industrial production environment, the shape of the rubber strip will change due to the gluing process, mechanical movement and environmental factors, resulting in defects such as uneven rubber strip width and glue breakage.

[0003] With the increasing application of gluing technology in the automobile production process, quality control in the gluing process has become particularly important. At present, visual inspection technology is widely used for real-time detection of gluing quality, which is highly dependent on imaging quality. When the imaging quality cannot achieve satisfactory results, misjudgment or missed judgment often occurs. Therefore, gluing inspection is a common quality control task in industrial production processes.

[0004] However, the use of traditional neural networks for glue coating detection is difficult to adapt to the dynamic changes in the shape of the glue strips and is easily affected by background interference, resulting in insufficient detection accuracy.

[0005] Therefore, the present application provides a method for detecting the quality of vehicle glue coating to solve the above technical problems. Summary of the invention

[0006] The purpose of this application is to provide a method, device, medium and electronic device for detecting the quality of vehicle glue coating, which can solve at least one of the above-mentioned technical problems. The specific solution is as follows:

[0007] According to a specific implementation of the present application, in a first aspect, the present application provides a method for detecting the quality of vehicle glue coating, comprising:

[0008] Acquire an image of a region of interest in an image of a rubber strip of a vehicle;

[0009] Preprocessing the region of interest image to obtain a first segmentation mask;

[0010] Applying the first segmentation mask to the trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information;

[0011] Applying the second segmentation mask to the trained preset rubber strip flow neural network model to obtain a target binary image, wherein the foreground rubber strip area is marked in the target binary image;

[0012] The foreground glue strip area in the preset reference binary image is used as a reference and compared with the foreground glue strip area in the target binary image, and whether the quality of the glue strip in the image of the region of interest is qualified is determined based on the comparison result.

[0013] Optionally, taking the foreground glue strip area in the preset reference binary image as a reference and comparing it with the foreground glue strip area in the target binary image, and determining whether the quality of the glue strip in the image of the region of interest is qualified based on the comparison result, includes:

[0014] Taking the foreground rubber strip area in the preset reference binary image as a reference, performing rubber break detection on the foreground rubber strip area in the target binary image to obtain a rubber break detection result, and taking the foreground rubber strip area in the preset reference binary image as a reference, performing width detection on the foreground rubber strip area in the target binary image to obtain a width detection result;

[0015] When the glue break detection result is normal connectivity information, and the width detection result is normal width information, it is determined that the quality of the glue strip in the image of the region of interest is qualified.

[0016] Optionally, the foreground glue strip area in the preset reference binary image is used as a reference to perform glue break detection on the foreground glue strip area in the target binary image to obtain the glue break detection result, including:

[0017] Determine at least one first connected region in the foreground rubber strip region in the target binary image, and determine at least one second connected region in the foreground rubber strip region in the preset reference binary image;

[0018] Calculating the number of first connected regions of the at least one first connected region, and calculating the number of second connected regions of the at least one second connected region;

[0019] When the number of the first connected regions is equal to the number of the second connected regions, each first connected region and each second connected region is divided into grids based on a preset first grid specification parameter value;

[0020] Calculating the number of first grids of each first connected region, and calculating the number of second grids of each second connected region;

[0021] Queue the first number of grids of the at least one first connected area according to a preset first queue order to obtain a first queue, and queue the second number of grids of the at least one second connected area according to the preset first queue order to obtain a second queue;

[0022] When the difference between the number of first grids of each queue rank in the first queue and the number of second grids of the corresponding queue rank in the second queue is less than or equal to a preset number difference threshold, the disconnection detection result is determined to be normal connectivity information.

[0023] Optionally, when the glue break detection result is normal connectivity information, taking the foreground glue strip area in the preset reference binary image as a reference, performing width detection on the foreground glue strip area in the target binary image to obtain a width detection result includes:

[0024] Determine at least one third connected region in the foreground rubber strip region in the target binary image, and determine at least one fourth connected region in the foreground rubber strip region in the preset reference binary image;

[0025] Calculating the number of third connected regions of the at least one third connected region, and calculating the number of fourth connected regions of the at least one fourth connected region;

[0026] When the number of the third connected regions is equal to the number of the fourth connected regions, each third connected region and each fourth connected region is divided into grids based on a preset second grid specification parameter value;

[0027] Calculating the number of third grids in each third connected region, and calculating the number of fourth grids in each fourth connected region;

[0028] Queue the at least one third connected region by using the respective third grid numbers according to the preset second queue order to obtain a third queue of the at least one third connected region, and queue the at least one fourth connected region by using the respective fourth grid numbers according to the preset second queue order to obtain a fourth queue of the at least one fourth connected region;

[0029] Extracting a third connected region of each queue rank from the third queue, and extracting a fourth connected region corresponding to the queue rank from the fourth queue;

[0030] Calculating the number of third grids for each width sequence string in the extracted third connected region, and calculating the number of fourth grids for each width sequence string in the extracted fourth connected region;

[0031] Calculate the difference between the number of third grids of the width sequence string of each length ranking in the length direction extracted from the third connected region and the number of fourth grids of the width sequence string of the corresponding length ranking in the length direction extracted from the fourth connected region, and obtain the sequence string difference of the corresponding length ranking in the length direction;

[0032] When the absolute value of the sequence string difference of any length ranking is greater than or equal to a preset normal sequence string difference threshold, the any length ranking is determined to be an abnormal length ranking;

[0033] Count the maximum number of adjacent abnormal length rankings;

[0034] When the maximum adjacent number is less than or equal to a preset normal adjacent number threshold, the width detection result is determined to be normal width information.

[0035] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0036] Acquire multiple sample images of the rubber strip of the vehicle;

[0037] Preprocessing the multiple sample images respectively to obtain multiple first sample segmentation masks;

[0038] The newly created preset rubber strip convolutional neural network model is trained using the multiple first sample segmentation masks until the second sample segmentation mask output by the preset rubber strip convolutional neural network model marks sample target feature information that meets the preset feature conditions.

[0039] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0040] Obtaining a plurality of second sample segmentation masks;

[0041] Preprocessing the plurality of second sample segmentation masks respectively to obtain a plurality of third sample segmentation masks;

[0042] The newly created preset rubber strip flow state neural network model is trained using the multiple third sample segmentation masks until a foreground rubber strip area that meets a preset foreground rubber strip area condition is marked in a target binary image output by the preset rubber strip flow state neural network model.

[0043] Optionally, the loss function of the preset rubber strip convolutional neural network model includes a cross entropy function, and the optimizer of the preset rubber strip flow state neural network model includes a BPTT optimizer.

[0044] According to a specific embodiment of the present application, in a second aspect, the present application provides a vehicle glue coating quality detection device, comprising:

[0045] An acquisition unit, used for acquiring an image of a region of interest in an image of a rubber strip of a vehicle;

[0046] A preprocessing unit, used for preprocessing the region of interest image to obtain a first segmentation mask;

[0047] A first obtaining unit, configured to apply the first segmentation mask to a trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information;

[0048] A second obtaining unit is used for applying the second segmentation mask to a trained preset rubber strip flow neural network model to obtain a target binary image, wherein a foreground rubber strip area is marked in the target binary image;

[0049] The determination unit is used to compare the foreground glue strip area in the preset reference binary image with the foreground glue strip area in the target binary image, and determine whether the quality of the glue strip in the image of the region of interest is qualified based on the comparison result.

[0050] Optionally, taking the foreground glue strip area in the preset reference binary image as a reference and comparing it with the foreground glue strip area in the target binary image, and determining whether the quality of the glue strip in the image of the region of interest is qualified based on the comparison result, includes:

[0051] Taking the foreground rubber strip area in the preset reference binary image as a reference, performing rubber break detection on the foreground rubber strip area in the target binary image to obtain a rubber break detection result, and taking the foreground rubber strip area in the preset reference binary image as a reference, performing width detection on the foreground rubber strip area in the target binary image to obtain a width detection result;

[0052] When the glue break detection result is normal connectivity information, and the width detection result is normal width information, it is determined that the quality of the glue strip in the image of the region of interest is qualified.

[0053] Optionally, the foreground glue strip area in the preset reference binary image is used as a reference to perform glue break detection on the foreground glue strip area in the target binary image to obtain the glue break detection result, including:

[0054] Determine at least one first connected region in the foreground rubber strip region in the target binary image, and determine at least one second connected region in the foreground rubber strip region in the preset reference binary image;

[0055] Calculating the number of first connected regions of the at least one first connected region, and calculating the number of second connected regions of the at least one second connected region;

[0056] When the number of the first connected regions is equal to the number of the second connected regions, each first connected region and each second connected region is divided into grids based on a preset first grid specification parameter value;

[0057] Calculating the number of first grids of each first connected region, and calculating the number of second grids of each second connected region;

[0058] Queue the first number of grids of the at least one first connected area according to a preset first queue order to obtain a first queue, and queue the second number of grids of the at least one second connected area according to the preset first queue order to obtain a second queue;

[0059] When the difference between the number of first grids of each queue rank in the first queue and the number of second grids of the corresponding queue rank in the second queue is less than or equal to a preset number difference threshold, the disconnection detection result is determined to be normal connectivity information.

[0060] Optionally, when the glue break detection result is normal connectivity information, taking the foreground glue strip area in the preset reference binary image as a reference, performing width detection on the foreground glue strip area in the target binary image to obtain a width detection result includes:

[0061] Determine at least one third connected region in the foreground rubber strip region in the target binary image, and determine at least one fourth connected region in the foreground rubber strip region in the preset reference binary image;

[0062] Calculating the number of third connected regions of the at least one third connected region, and calculating the number of fourth connected regions of the at least one fourth connected region;

[0063] When the number of the third connected regions is equal to the number of the fourth connected regions, each third connected region and each fourth connected region is divided into grids based on a preset second grid specification parameter value;

[0064] Calculating the number of third grids in each third connected region, and calculating the number of fourth grids in each fourth connected region;

[0065] Queue the at least one third connected region by using the respective third grid numbers according to the preset second queue order to obtain a third queue of the at least one third connected region, and queue the at least one fourth connected region by using the respective fourth grid numbers according to the preset second queue order to obtain a fourth queue of the at least one fourth connected region;

[0066] Extracting a third connected region of each queue rank from the third queue, and extracting a fourth connected region corresponding to the queue rank from the fourth queue;

[0067] Calculating the number of third grids for each width sequence string in the extracted third connected region, and calculating the number of fourth grids for each width sequence string in the extracted fourth connected region;

[0068] Calculate the difference between the number of third grids of the width sequence string of each length ranking in the length direction extracted from the third connected region and the number of fourth grids of the width sequence string of the corresponding length ranking in the length direction extracted from the fourth connected region, and obtain the sequence string difference of the corresponding length ranking in the length direction;

[0069] When the absolute value of the sequence string difference of any length ranking is greater than or equal to a preset normal sequence string difference threshold, the any length ranking is determined to be an abnormal length ranking;

[0070] Count the maximum number of adjacent abnormal length rankings;

[0071] When the maximum adjacent number is less than or equal to a preset normal adjacent number threshold, the width detection result is determined to be normal width information.

[0072] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0073] Acquire multiple sample images of the rubber strip of the vehicle;

[0074] Preprocessing the multiple sample images respectively to obtain multiple first sample segmentation masks;

[0075] The newly created preset rubber strip convolutional neural network model is trained using the multiple first sample segmentation masks until the second sample segmentation mask output by the preset rubber strip convolutional neural network model marks sample target feature information that meets the preset feature conditions.

[0076] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0077] Obtaining a plurality of second sample segmentation masks;

[0078] Preprocessing the plurality of second sample segmentation masks respectively to obtain a plurality of third sample segmentation masks;

[0079] The newly created preset rubber strip flow state neural network model is trained using the multiple third sample segmentation masks until a foreground rubber strip area that meets a preset foreground rubber strip area condition is marked in a target binary image output by the preset rubber strip flow state neural network model.

[0080] Optionally, the loss function of the preset rubber strip convolutional neural network model includes a cross entropy function, and the optimizer of the preset rubber strip flow state neural network model includes a BPTT optimizer.

[0081] According to the specific implementation of the present application, in a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the method for detecting the vehicle glue coating quality as described in any of the above items is implemented.

[0082] According to the specific implementation of the present application, in a fourth aspect, the present application provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the vehicle glue coating quality detection method as described in any of the above items.

[0083] Compared with the prior art, the above solution of the embodiment of the present application has at least the following beneficial effects:

[0084] The present application provides a method, device, medium and electronic device for detecting the quality of vehicle glue coating. In the embodiment of the present application, the first segmentation mask of the image of the region of interest in the glue-coated rubber strip image is applied to the trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information; the second segmentation mask is applied to the trained preset rubber strip flow neural network model to obtain a target binary image, wherein the foreground rubber strip area is marked in the target binary image; the foreground rubber strip area in the preset reference binary image is used as a reference, and the foreground rubber strip area in the target binary image is compared, and based on the comparison result, it is determined whether the quality of the glue-coated rubber strip in the image of the region of interest is qualified. Through the trained preset rubber strip convolutional neural network model and the preset rubber strip flow neural network model, the target feature information of the rubber strip area is accurately and efficiently extracted, and the accurate instance segmentation of the rubber strip area is achieved. The detection accuracy and robustness are improved, thereby improving the automation level of glue coating quality detection, which is helpful for real-time quality control in industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 A flow chart of a method for detecting the quality of vehicle glue coating according to an embodiment of the present application is shown;

[0086] Figure 2 The image of the glue strip and the target binary image of the vehicle glue quality detection method according to the embodiment of the present application are shown;

[0087] Figure 3 A unit block diagram of a vehicle glue coating quality detection device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0088] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0089] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0090] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0091] It should be understood that although the terms first, second, third, etc. may be used to describe in the embodiments of the present application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first.

[0092] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0093] It should also be noted that the term "includes", "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, so that a commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the commodity or device including the elements.

[0094] It should be particularly noted that any symbols and / or numbers in the specification that are not marked in the accompanying drawings are not drawing marks.

[0095] The optional embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0096] The embodiment provided in this application is an embodiment of a method for detecting the quality of vehicle glue coating.

[0097] Combine the following Figure 1 The embodiments of the present application are described in detail.

[0098] Step S101, obtaining an image of a region of interest in an image of a rubber strip of a vehicle.

[0099] In machine vision and image processing, the area to be processed is outlined from the image to be processed in the form of a box, circle, ellipse, irregular polygon, etc., which is called the region of interest (ROI for short).

[0100] Step S102: preprocess the region of interest image to obtain a first segmentation mask.

[0101] Preprocessing includes but is not limited to scaling and normalization.

[0102] Segmentation mask is a technique in computer vision that is used to accurately separate objects in an image from the background. It achieves fine-grained segmentation of image regions by classifying and labeling each pixel. Each pixel is assigned a label to indicate whether it belongs to the foreground or background, or to a different object category. Such label information forms a two-dimensional matrix, namely the segmentation mask.

[0103] Step S103: Apply the first segmentation mask to the trained preset tape convolutional neural network model to obtain a second segmentation mask marked with target feature information.

[0104] The preset tape convolutional neural network model is built based on the convolutional neural network model (full name Convolutional Neural Network, referred to as CNN). CNN consists of neurons with learnable weights and bias constants. Each neuron receives some inputs and performs some dot product calculations, and the output is the score of each classification. It is a feedforward neural network whose artificial neurons can respond to surrounding units within a part of the coverage range, and has excellent performance for large image processing. CNN is characterized by neurons with multi-dimensional volumes. Taking advantage of the characteristics of the input being a picture, neurons are designed into multiple dimensions. The convolutional neural network consists of one or more convolutional layers and a fully connected layer at the top, as well as associated weights and pooling layers. This structure enables the convolutional neural network to utilize the two-dimensional structure of the input data.

[0105] The embodiment of the present application utilizes the excellent feature extraction capability of the preset tape convolutional neural network model to extract the target feature information in the first segmentation mask. For example, the target feature information includes edge information and / or texture information in the first segmentation mask. The preset tape convolutional neural network model must be trained with a large number of sample images, and the output sample target feature information can only be effectively used after it meets the preset feature conditions.

[0106] Step S104: applying the second segmentation mask to the trained preset rubber strip flow neural network model to obtain a target binary image.

[0107] Wherein, the foreground rubber strip area is marked in the target binary image.

[0108] The preset strip fluid neural network model is built based on the fluid neural network model (Liquid NeuralNets, or LNNs). LNNs are an exciting research direction in the field of artificial intelligence and machine learning, enabling more compact and dynamic time series prediction neural networks. The main advantage of LNNs is that they can continue to adapt to new stimuli after training. In addition, LNNs are robust under noisy conditions and are smaller and more interpretable than traditional networks.

[0109] The embodiment of the present application utilizes the time-varying characteristics of LNNs, so that it can adapt to the morphological changes of the rubber strips, and accurately segment the foreground rubber strip area in the target binary image by dynamically and adaptively updating parameters in time and space, and adapting to the morphological changes of different rubber strips. The preset rubber strip convolutional neural network model must be trained with a large number of sample segmentation masks, and the output foreground rubber strip area can only be effectively used after it meets the preset foreground rubber strip area conditions.

[0110] like Figure 2 As shown in FIG. 1 , the target binary images are generated from two adhesive strip images. The dark brown area surrounded by the blue marking line in the target binary image is the foreground adhesive strip area.

[0111] Step S105 , taking the foreground rubber strip area in the preset reference binary image as a reference, and comparing it with the foreground rubber strip area in the target binary image, and determining whether the quality of the rubber strip in the region of interest image is qualified based on the comparison result.

[0112] The foreground rubber strip area in the preset reference binary image meets the qualified quality standard. The foreground rubber strip area in the preset reference binary image is used as a reference to check whether the quality of the foreground rubber strip area in the target binary image is qualified.

[0113] In some specific embodiments, the foreground glue strip area in the preset reference binary image is compared with the foreground glue strip area in the target binary image, and the quality of the glue strip in the image of the region of interest is determined based on the comparison result, including:

[0114] Step S105-1, taking the foreground rubber strip area in the preset reference binary image as a reference, performing rubber break detection on the foreground rubber strip area in the target binary image to obtain a rubber break detection result, and taking the foreground rubber strip area in the preset reference binary image as a reference, performing width detection on the foreground rubber strip area in the target binary image to obtain a width detection result.

[0115] Glue break detection refers to identifying the continuity of the glue strip in the segmented image, detecting the discontinuous area, and determining whether there is a glue break defect in the glue strip in the image.

[0116] Width detection refers to calculating the width of the rubber strip along the length direction of the foreground rubber strip area in the target binary image and comparing it with the width of the foreground rubber strip area in the preset reference binary image. If the width difference exceeds the preset width difference threshold, it is determined that there is a defect in the width of the rubber strip in the foreground rubber strip area in the target binary image.

[0117] Step S105-2: when the glue break detection result is normal connectivity information and the width detection result is normal width information, it is determined that the quality of the glue strip in the image of the region of interest is qualified.

[0118] In this specific embodiment, whether the quality of the glue strip in the image of the region of interest is qualified is determined by the glue breaking detection result of the glue breaking detection and the width detection result of the width detection. When the glue breaking detection result is normal connectivity information and the width detection result is normal width information, the quality of the glue strip in the image of the region of interest can be determined to be qualified. When the glue breaking detection result is abnormal connectivity information or the width detection result is abnormal width information, the quality of the glue strip in the image of the region of interest is determined to be unqualified.

[0119] In some specific embodiments, the foreground glue strip area in the preset reference binary image is used as a reference, and the glue break detection is performed on the foreground glue strip area in the target binary image to obtain the glue break detection result, including:

[0120] Step S105 - 1 a - 1 , determining at least one first connected region in the foreground rubber strip region in the target binary image, and determining at least one second connected region in the foreground rubber strip region in the preset reference binary image.

[0121] Extract the RGB value of each pixel in the foreground rubber strip area, determine the pixels whose RGB values ​​are within the preset RGB value range as target pixels, and determine the largest area composed of multiple adjacent target pixels as the connected area. Figure 2 As described above, the area composed of dark brown pixels within the blue marked lines in the two target binary images is the connected area.

[0122] Step S105 - 1 a - 2 , calculating the number of first connected regions of the at least one first connected region, and calculating the number of second connected regions of the at least one second connected region.

[0123] Step S105-1a-3, when the number of the first connected areas is equal to the number of the second connected areas, each first connected area and each second connected area is divided into grids based on a preset first grid specification parameter value.

[0124] In this specific embodiment, when the number of the first connected areas is equal to the number of the second connected areas, it indicates that the number of the detected glue-coated rubber strips is consistent with the number of benchmark requirements; when the number of the first connected areas is not equal to the number of the second connected areas, it indicates that the number of the detected glue-coated rubber strips is inconsistent with the number of benchmark requirements, and the glue break detection result is determined to be connectivity abnormality information.

[0125] When the number is the same, each first connected region and each second connected region are divided into grids respectively, and the specifications of each grid are consistent.

[0126] Step S105 - 1 a - 4 , calculating the number of first grids in each first connected region, and calculating the number of second grids in each second connected region.

[0127] Step S105-1a-5, queue the first grid quantity of each of the at least one first connected area according to the preset first queue order to obtain a first queue, and queue the second grid quantity of each of the at least one second connected area according to the preset first queue order to obtain a second queue.

[0128] The preset first queue order includes a preset first descending order or a preset first ascending order.

[0129] The first queue stores the number of first grids in each first connected region; the second queue stores the number of second grids in each second connected region.

[0130] The purpose of queuing is to compare the number of first grids and the number of second grids of the same queue ranking in two queues. For example, the number of first grids of the first queue ranking in the first queue is compared with the number of second grids of the first queue ranking in the second queue; the number of first grids of the second queue ranking in the first queue is compared with the number of second grids of the second queue ranking in the second queue; the number of first grids of the third queue ranking in the first queue is compared with the number of second grids of the third queue ranking in the second queue, and so on.

[0131] Step S105-1a-6, when the difference between the number of first grids of each queue rank in the first queue and the number of second grids of the corresponding queue rank in the second queue is less than or equal to a preset number difference threshold, determine that the disconnection detection result is normal connectivity information.

[0132] If the difference between the ranks of each of the two queues is less than or equal to the preset quantity difference threshold, indicating that each first connected area meets the benchmark requirement, the glue breaking detection result is determined to be normal connectivity information.

[0133] If the difference in the ranking of any queue of the two queues is greater than the preset quantity difference threshold, indicating that the first connected area corresponding to the first grid quantity of any queue ranking does not meet the benchmark requirement, the disconnection detection result is determined to be connectivity abnormality information.

[0134] This specific embodiment identifies whether there is a glue-breaking defect in the glue-coated rubber strip in the foreground rubber strip area by the number of grids. The smaller the preset first grid specification parameter value is, the higher the recognition accuracy is. This reduces the time delay and resource occupation caused by complex image calculations, can efficiently and accurately identify glue-breaking defects, and improves detection efficiency.

[0135] In some specific embodiments, when the glue break detection result is normal connectivity information, a foreground glue strip area in a preset reference binary image is used as a reference to perform width detection on the foreground glue strip area in the target binary image to obtain a width detection result, including:

[0136] Step S105 - 1 b - 1 , determining at least one third connected region in the foreground rubber strip region in the target binary image, and determining at least one fourth connected region in the foreground rubber strip region in the preset reference binary image.

[0137] Step S105-1b-2, calculating the number of third connected regions of the at least one third connected region, and calculating the number of fourth connected regions of the at least one fourth connected region.

[0138] Step S105-1b-3, when the number of the third connected regions is equal to the number of the fourth connected regions, each third connected region and each fourth connected region is respectively divided into grids based on a preset second grid specification parameter value.

[0139] In this specific embodiment, when the number of the third connected areas is equal to the number of the fourth connected areas, it indicates that the number of the detected glue-coated rubber strips is consistent with the number of benchmark requirements; when the number of the third connected areas is not equal to the number of the fourth connected areas, it indicates that the number of the detected glue-coated rubber strips is inconsistent with the number of benchmark requirements, and the glue break detection result is determined to be connectivity abnormality information.

[0140] When the number is the same, each third connected region and each fourth connected region are divided into grids respectively, and the specifications of each grid are consistent.

[0141] Step S105-1b-4, calculating the number of third grids in each third connected region, and calculating the number of fourth grids in each fourth connected region.

[0142] Step S105-1b-5, queue the at least one third connected area according to the preset second queue order by the respective third grid quantities to obtain the third queue of the at least one third connected area, and queue the at least one fourth connected area according to the preset second queue order by the respective fourth grid quantities to obtain the fourth queue of the at least one fourth connected area.

[0143] The preset second queue order includes a preset second descending order or a preset second ascending order.

[0144] The third queue stores each third connected area, and the queue ranking of each third connected area is determined by the sorting of the number of third grids corresponding to the third connected area arranged in the preset second queue order; the fourth queue stores each fourth connected area, and the queue ranking of each fourth connected area is determined by the sorting of the number of fourth grids corresponding to the fourth connected area arranged in the preset second queue order.

[0145] The purpose of queuing is to compare the third connected area with the fourth connected area of ​​the same queue ranking in two queues. For example, the third connected area of ​​the first queue ranking in the third queue is compared with the fourth connected area of ​​the first queue ranking in the fourth queue; the third connected area of ​​the second queue ranking in the third queue is compared with the fourth connected area of ​​the second queue ranking in the fourth queue, the third connected area of ​​the third queue ranking in the third queue is compared with the fourth connected area of ​​the third queue ranking in the fourth queue, and so on.

[0146] Step S105-1b-6, extracting the third connected area of ​​each queue rank from the third queue, and extracting the fourth connected area of ​​the corresponding queue rank from the fourth queue.

[0147] Step S105 - 1b - 7 , calculating the number of third grids of each width sequence string in the extracted third connected region, and calculating the number of fourth grids of each width sequence string in the extracted fourth connected region.

[0148] The calculation of the third grid number of each width sequence string in the extracted third connected area can be understood as selecting a width sequence string composed of multiple parallel grids in the same direction in each width direction along the length direction of the third connected area, and calculating the number of grids in the width sequence string (i.e., the third grid number), that is, calculating the width value along the length direction of the rubber strip.

[0149] Step S105-1b-8, calculate the difference between the number of third grids of the width sequence string of each length ranking in the length direction in the extracted third connected area and the number of fourth grids of the width sequence string of the corresponding length ranking in the length direction in the extracted fourth connected area, and obtain the difference of the sequence strings of the corresponding length ranking in the length direction.

[0150] That is, the number of width sequence strings of the extracted third connected regions and the fourth connected regions is compared at each length ranking in the length direction, and the sequence string difference of the corresponding length ranking is obtained.

[0151] Step S105-1b-9: when the absolute value of the sequence string difference of any length ranking is greater than or equal to a preset normal sequence string difference threshold, the any length ranking is determined to be an abnormal length ranking.

[0152] When the absolute value of the sequence string difference of any length ranking is less than a preset normal sequence string difference threshold, the any length ranking is determined to be a normal length ranking.

[0153] Step S105-1b-10, counting the maximum number of adjacent abnormal length rankings.

[0154] For example, if the abnormal length ranking is: 20-29 length ranking, the number of adjacent nodes is 10; if the abnormal length ranking is: 40-45 length ranking, the number of adjacent nodes is 6; the maximum number of adjacent nodes is 10.

[0155] Step S105-1b-11: when the maximum adjacent number is less than or equal to a preset normal adjacent number threshold, determine that the width detection result is normal width information.

[0156] When the maximum adjacent number is greater than the preset normal adjacent number threshold, the width detection result is determined to be width abnormality information, that is, the width of the rubber strip is abnormal.

[0157] This specific embodiment identifies whether there is a width abnormality in the glue-coated rubber strip in the foreground rubber strip area through the number of grids. For two connected areas with the same queue ranking in the two queues, width sequence strings with the same length ranking are selected from the two connected areas along the length direction, and the two width sequence strings are compared for abnormality to determine whether each length ranking is an abnormal length ranking. If the maximum number of consecutive abnormal length rankings exceeds the preset normal adjacent number threshold, the width detection result is determined to be width abnormality information. The smaller the preset second grid specification parameter value, the higher the recognition accuracy. The time delay and resource occupation caused by complex image calculations are reduced, and width defects can be identified efficiently and accurately, thereby improving detection efficiency.

[0158] In some specific embodiments, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0159] Step S100a-1, obtaining a plurality of sample images of the rubber strip of the vehicle.

[0160] Step S100a-2: pre-process the multiple sample images respectively to obtain multiple first sample segmentation masks.

[0161] Step S100a-3, using the multiple first sample segmentation masks to train the newly created preset rubber strip convolutional neural network model until the second sample segmentation mask output by the preset rubber strip convolutional neural network model marks the sample target feature information that meets the preset feature conditions.

[0162] In some specific embodiments, the loss function of the preset tape convolutional neural network model includes a cross entropy function.

[0163] In this specific embodiment, by training the preset glue strip convolutional neural network model, the target feature information of the glue strip area is accurately and efficiently extracted, which improves the detection accuracy and robustness, thereby improving the automation level of glue coating quality detection and facilitating real-time quality control in industrial production.

[0164] In some specific embodiments, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0165] Step S100b-1, obtaining a plurality of second sample segmentation masks.

[0166] Step S100b-2: pre-process the plurality of second sample segmentation masks respectively to obtain a plurality of third sample segmentation masks.

[0167] Step S100b-3, using the plurality of third sample segmentation masks to train a newly created preset rubber strip flow neural network model until a foreground rubber strip area that meets a preset foreground rubber strip area condition is marked in a target binary image output by the preset rubber strip flow neural network model.

[0168] In some specific embodiments, the optimizer of the preset strip flow neural network model includes a BPTT optimizer.

[0169] In this specific embodiment, by training a preset rubber strip flow neural network model, accurate instance segmentation of the rubber strip area is achieved, which is suitable for rubber strip detection scenarios with complex shapes and backgrounds. Compared with traditional methods, the detection accuracy and robustness are improved, and the rubber strip and background can be efficiently separated, thereby improving the automation level of glue coating quality detection, which is helpful for real-time quality control in industrial production.

[0170] The embodiment of the present application applies the first segmentation mask of the image of the region of interest in the glue-coated rubber strip image to the trained preset rubber strip convolutional neural network model to obtain the second segmentation mask marked with the target feature information; applies the second segmentation mask to the trained preset rubber strip flow neural network model to obtain the target binary image, wherein the foreground rubber strip region is marked in the target binary image; takes the foreground rubber strip region in the preset reference binary image as a reference, and compares it with the foreground rubber strip region in the target binary image, and determines whether the quality of the glue-coated rubber strip in the image of the region of interest is qualified based on the comparison result. Through the trained preset rubber strip convolutional neural network model and the preset rubber strip flow neural network model, accurate and efficient extraction of the target feature information of the rubber strip region and accurate instance segmentation of the rubber strip region are achieved. The detection accuracy and robustness are improved, thereby improving the automation level of glue coating quality detection, which is helpful for real-time quality control in industrial production.

[0171] The present application also provides a device embodiment that is based on the above embodiment, which is used to implement the method steps described in the above embodiment. The explanation based on the same name meaning is the same as the above embodiment, and has the same technical effect as the above embodiment, which will not be repeated here.

[0172] like Figure 3 As shown, the present application provides a vehicle glue coating quality detection device 300, comprising:

[0173] An acquisition unit 301 is used to acquire an image of a region of interest in an image of a rubber strip of a vehicle;

[0174] A preprocessing unit 302 is used to preprocess the region of interest image to obtain a first segmentation mask;

[0175] A first obtaining unit 303 is used to apply the first segmentation mask to a trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information;

[0176] A second obtaining unit 304 is used to apply the second segmentation mask to the trained preset rubber strip flow neural network model to obtain a target binary image, wherein the foreground rubber strip area is marked in the target binary image;

[0177] The determination unit 305 is used to compare the foreground rubber strip area in the preset reference binary image with the foreground rubber strip area in the target binary image, and determine whether the quality of the rubber strip in the image of the region of interest is qualified based on the comparison result.

[0178] Optionally, taking the foreground glue strip area in the preset reference binary image as a reference and comparing it with the foreground glue strip area in the target binary image, and determining whether the quality of the glue strip in the image of the region of interest is qualified based on the comparison result, includes:

[0179] Taking the foreground rubber strip area in the preset reference binary image as a reference, performing rubber break detection on the foreground rubber strip area in the target binary image to obtain a rubber break detection result, and taking the foreground rubber strip area in the preset reference binary image as a reference, performing width detection on the foreground rubber strip area in the target binary image to obtain a width detection result;

[0180] When the glue break detection result is normal connectivity information, and the width detection result is normal width information, it is determined that the quality of the glue strip in the image of the region of interest is qualified.

[0181] Optionally, the foreground glue strip area in the preset reference binary image is used as a reference to perform glue break detection on the foreground glue strip area in the target binary image to obtain the glue break detection result, including:

[0182] Determine at least one first connected region in the foreground rubber strip region in the target binary image, and determine at least one second connected region in the foreground rubber strip region in the preset reference binary image;

[0183] Calculating the number of first connected regions of the at least one first connected region, and calculating the number of second connected regions of the at least one second connected region;

[0184] When the number of the first connected regions is equal to the number of the second connected regions, each first connected region and each second connected region is divided into grids based on a preset first grid specification parameter value;

[0185] Calculating the number of first grids of each first connected region, and calculating the number of second grids of each second connected region;

[0186] Queue the first number of grids of the at least one first connected area according to a preset first queue order to obtain a first queue, and queue the second number of grids of the at least one second connected area according to the preset first queue order to obtain a second queue;

[0187] When the difference between the number of first grids of each queue rank in the first queue and the number of second grids of the corresponding queue rank in the second queue is less than or equal to a preset number difference threshold, the disconnection detection result is determined to be normal connectivity information.

[0188] Optionally, when the glue break detection result is normal connectivity information, taking the foreground glue strip area in the preset reference binary image as a reference, performing width detection on the foreground glue strip area in the target binary image to obtain a width detection result includes:

[0189] Determine at least one third connected region in the foreground rubber strip region in the target binary image, and determine at least one fourth connected region in the foreground rubber strip region in the preset reference binary image;

[0190] Calculating the number of third connected regions of the at least one third connected region, and calculating the number of fourth connected regions of the at least one fourth connected region;

[0191] When the number of the third connected regions is equal to the number of the fourth connected regions, each third connected region and each fourth connected region is divided into grids based on a preset second grid specification parameter value;

[0192] Calculating the number of third grids in each third connected region, and calculating the number of fourth grids in each fourth connected region;

[0193] Queue the at least one third connected region by using the respective third grid numbers according to the preset second queue order to obtain a third queue of the at least one third connected region, and queue the at least one fourth connected region by using the respective fourth grid numbers according to the preset second queue order to obtain a fourth queue of the at least one fourth connected region;

[0194] Extracting a third connected region of each queue rank from the third queue, and extracting a fourth connected region corresponding to the queue rank from the fourth queue;

[0195] Calculating the number of third grids for each width sequence string in the extracted third connected region, and calculating the number of fourth grids for each width sequence string in the extracted fourth connected region;

[0196] Calculate the difference between the number of third grids of the width sequence string of each length ranking in the length direction extracted from the third connected region and the number of fourth grids of the width sequence string of the corresponding length ranking in the length direction extracted from the fourth connected region, and obtain the sequence string difference of the corresponding length ranking in the length direction;

[0197] When the absolute value of the sequence string difference of any length ranking is greater than or equal to a preset normal sequence string difference threshold, the any length ranking is determined to be an abnormal length ranking;

[0198] Count the maximum number of adjacent abnormal length rankings;

[0199] When the maximum adjacent number is less than or equal to a preset normal adjacent number threshold, the width detection result is determined to be normal width information.

[0200] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0201] Acquire multiple sample images of the rubber strip of the vehicle;

[0202] Preprocessing the multiple sample images respectively to obtain multiple first sample segmentation masks;

[0203] The newly created preset rubber strip convolutional neural network model is trained using the multiple first sample segmentation masks until the second sample segmentation mask output by the preset rubber strip convolutional neural network model marks sample target feature information that meets the preset feature conditions.

[0204] Optionally, before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes:

[0205] Obtaining a plurality of second sample segmentation masks;

[0206] Preprocessing the plurality of second sample segmentation masks respectively to obtain a plurality of third sample segmentation masks;

[0207] The newly created preset rubber strip flow state neural network model is trained using the multiple third sample segmentation masks until a foreground rubber strip area that meets a preset foreground rubber strip area condition is marked in a target binary image output by the preset rubber strip flow state neural network model.

[0208] Optionally, the loss function of the preset rubber strip convolutional neural network model includes a cross entropy function, and the optimizer of the preset rubber strip flow state neural network model includes a BPTT optimizer.

[0209] The embodiment of the present application applies the first segmentation mask of the image of the region of interest in the glue-coated rubber strip image to the trained preset rubber strip convolutional neural network model to obtain the second segmentation mask marked with the target feature information; applies the second segmentation mask to the trained preset rubber strip flow neural network model to obtain the target binary image, wherein the foreground rubber strip region is marked in the target binary image; takes the foreground rubber strip region in the preset reference binary image as a reference, and compares it with the foreground rubber strip region in the target binary image, and determines whether the quality of the glue-coated rubber strip in the image of the region of interest is qualified based on the comparison result. Through the trained preset rubber strip convolutional neural network model and the preset rubber strip flow neural network model, accurate and efficient extraction of the target feature information of the rubber strip region and accurate instance segmentation of the rubber strip region are achieved. The detection accuracy and robustness are improved, thereby improving the automation level of glue coating quality detection, which is helpful for real-time quality control in industrial production.

[0210] Example 3

[0211] This embodiment provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method steps described in the above embodiment.

[0212] Example 4

[0213] An embodiment of the present application provides a non-volatile computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the method steps described in the above embodiment.

[0214] Finally, it should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other. For the system or device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0215] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting the quality of vehicle glue coating, characterized in that: include: Acquire an image of a region of interest in an image of a rubber strip of a vehicle; Preprocessing the region of interest image to obtain a first segmentation mask; Applying the first segmentation mask to the trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information; Applying the second segmentation mask to the trained preset rubber strip flow neural network model to obtain a target binary image, wherein the foreground rubber strip area is marked in the target binary image; The foreground glue strip area in the preset reference binary image is used as a reference and compared with the foreground glue strip area in the target binary image, and whether the quality of the glue strip in the image of the region of interest is qualified is determined based on the comparison result.

2. The method according to claim 1, characterized in that The method of comparing the foreground rubber strip area in the preset reference binary image with the foreground rubber strip area in the target binary image, and determining whether the quality of the rubber strip in the image of the region of interest is qualified based on the comparison result, includes: Taking the foreground rubber strip area in the preset reference binary image as a reference, performing rubber break detection on the foreground rubber strip area in the target binary image to obtain a rubber break detection result, and taking the foreground rubber strip area in the preset reference binary image as a reference, performing width detection on the foreground rubber strip area in the target binary image to obtain a width detection result; When the glue break detection result is normal connectivity information, and the width detection result is normal width information, it is determined that the quality of the glue strip in the image of the region of interest is qualified.

3. The method according to claim 2, characterized in that The method of performing glue break detection on the foreground glue strip area in the target binary image based on the foreground glue strip area in the preset reference binary image to obtain the glue break detection result includes: Determine at least one first connected region in the foreground rubber strip region in the target binary image, and determine at least one second connected region in the foreground rubber strip region in the preset reference binary image; Calculating the number of first connected regions of the at least one first connected region, and calculating the number of second connected regions of the at least one second connected region; When the number of the first connected regions is equal to the number of the second connected regions, each first connected region and each second connected region is divided into grids based on a preset first grid specification parameter value; Calculating the number of first grids of each first connected region, and calculating the number of second grids of each second connected region; Queue the first number of grids of the at least one first connected area according to a preset first queue order to obtain a first queue, and queue the second number of grids of the at least one second connected area according to the preset first queue order to obtain a second queue; When the difference between the number of first grids of each queue rank in the first queue and the number of second grids of the corresponding queue rank in the second queue is less than or equal to a preset number difference threshold, the disconnection detection result is determined to be normal connectivity information.

4. The method according to claim 2, characterized in that: When the glue break detection result is normal connectivity information, taking the foreground glue strip area in the preset reference binary image as a reference, performing width detection on the foreground glue strip area in the target binary image to obtain a width detection result includes: Determine at least one third connected region in the foreground rubber strip region in the target binary image, and determine at least one fourth connected region in the foreground rubber strip region in the preset reference binary image; Calculating the number of third connected regions of the at least one third connected region, and calculating the number of fourth connected regions of the at least one fourth connected region; When the number of the third connected regions is equal to the number of the fourth connected regions, each third connected region and each fourth connected region is divided into grids based on a preset second grid specification parameter value; Calculating the number of third grids in each third connected region, and calculating the number of fourth grids in each fourth connected region; Queue the at least one third connected region by using the respective third grid numbers according to the preset second queue order to obtain a third queue of the at least one third connected region, and queue the at least one fourth connected region by using the respective fourth grid numbers according to the preset second queue order to obtain a fourth queue of the at least one fourth connected region; Extracting a third connected region of each queue rank from the third queue, and extracting a fourth connected region corresponding to the queue rank from the fourth queue; Calculating the number of third grids for each width sequence string in the extracted third connected region, and calculating the number of fourth grids for each width sequence string in the extracted fourth connected region; Calculate the difference between the number of third grids of the width sequence string of each length ranking in the length direction extracted from the third connected region and the number of fourth grids of the width sequence string of the corresponding length ranking in the length direction extracted from the fourth connected region, and obtain the sequence string difference of the corresponding length ranking in the length direction; When the absolute value of the sequence string difference of any length ranking is greater than or equal to a preset normal sequence string difference threshold, the any length ranking is determined to be an abnormal length ranking; Count the maximum number of adjacent abnormal length rankings; When the maximum adjacent number is less than or equal to a preset normal adjacent number threshold, the width detection result is determined to be normal width information.

5. The method according to claim 1, characterized in that Before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes: Acquire multiple sample images of the rubber strip of the vehicle; Preprocessing the multiple sample images respectively to obtain multiple first sample segmentation masks; The newly created preset rubber strip convolutional neural network model is trained using the multiple first sample segmentation masks until the second sample segmentation mask output by the preset rubber strip convolutional neural network model marks sample target feature information that meets the preset feature conditions.

6. The method according to claim 5, characterized in that Before acquiring the image of the region of interest in the image of the rubber strip of the vehicle, the method further includes: Obtaining a plurality of second sample segmentation masks; Preprocessing the plurality of second sample segmentation masks respectively to obtain a plurality of third sample segmentation masks; The newly created preset rubber strip flow state neural network model is trained using the multiple third sample segmentation masks until a foreground rubber strip area that meets a preset foreground rubber strip area condition is marked in a target binary image output by the preset rubber strip flow state neural network model.

7. The method according to claim 6, characterized in that The loss function of the preset rubber strip convolutional neural network model includes a cross entropy function, and the optimizer of the preset rubber strip flow state neural network model includes a BPTT optimizer.

8. A vehicle glue coating quality detection device, characterized in that: include: An acquisition unit, used for acquiring an image of a region of interest in an image of a rubber strip of a vehicle; A preprocessing unit, used for preprocessing the region of interest image to obtain a first segmentation mask; A first obtaining unit, configured to apply the first segmentation mask to a trained preset rubber strip convolutional neural network model to obtain a second segmentation mask marked with target feature information; A second obtaining unit is used for applying the second segmentation mask to a trained preset rubber strip flow neural network model to obtain a target binary image, wherein a foreground rubber strip area is marked in the target binary image; The determination unit is used to compare the foreground glue strip area in the preset reference binary image with the foreground glue strip area in the target binary image, and determine whether the quality of the glue strip in the image of the region of interest is qualified based on the comparison result.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 7.

Images