Pipeline liquid state identification method and device, computer device and storage medium

Abstract

The present application relates to the technical field of biological analysis, and discloses a pipeline liquid state recognition method and device, computer equipment and a storage medium. The pipeline liquid state recognition method obtains a pipeline enhanced image through an image acquisition unit; the pipeline enhanced image includes a pipeline image element of a target pipeline for liquid flow; a pipeline image deformation result of the pipeline image element is recognized by analyzing and processing the pipeline enhanced image through a preset image analysis model; and a pipeline liquid state corresponding to the target pipeline is obtained based on the pipeline image deformation result. By introducing the pipeline image element as an image background in the pipeline enhanced image and recognizing the pipeline image deformation result based on the preset image analysis model, different liquid states in the pipeline can be more effectively distinguished, the efficiency and accuracy of pipeline liquid state recognition are improved, and the stability of state recognition is improved.

Description

Technical Field

[0001] This invention relates to the field of bioanalytical technology, and in particular to a method, apparatus, computer equipment, and storage medium for identifying the liquid state in pipelines. Background Technology

[0002] In the cell therapy industry, to ensure a sterile environment for the cell solution, the liquid in the tubing must be kept sealed at every stage of the cell processing flow. Leaks, dead spots, or loose connections in the tubing can allow external air to enter and create air bubbles. These air bubbles can affect liquid transport and disrupt the sterile environment of the cell solution. Therefore, it is essential to monitor the presence of liquid and air bubbles in the tubing.

[0003] Image recognition is a method that uses trained deep learning models to identify the state of actual liquid pipeline images. The key to image recognition lies in the acquisition and analysis of liquid pipeline images. Current detection methods often acquire liquid pipeline images directly against the environmental background or against a solid-color background. In practical applications, transparent pipelines primarily transport transparent liquids, making images of liquid pipelines with environmental backgrounds susceptible to interference, leading to misjudgments of the liquid's state. Conversely, images of liquid pipelines with a solid-color background can actually increase the difficulty of image recognition for liquids of similar colors, resulting in unstable liquid state identification results. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, device, computer equipment, and storage medium for identifying the liquid state of pipelines, in order to solve the problems of easy deviation and unstable results in identifying the liquid state of pipelines.

[0005] A method for identifying the liquid state of a pipeline includes: Enhanced pipeline images are acquired through an image acquisition unit; the enhanced pipeline images include pipeline image elements for the target pipeline through which the liquid flows. The pipeline enhancement image is analyzed and processed using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements; The liquid state of the pipeline corresponding to the target pipeline is obtained based on the pipeline image deformation results.

[0006] A pipeline liquid status identification device, comprising: An image acquisition module is used to acquire enhanced pipeline images through an image acquisition unit; the enhanced pipeline images include pipeline image elements for a target pipeline through which liquid flows. The image analysis module is used to analyze and process the enhanced pipeline image using a preset image analysis model, and to identify the pipeline image deformation results of the pipeline image elements. The state determination module is used to determine the liquid state of the pipeline corresponding to the target pipeline based on the pipeline image deformation result.

[0007] A computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the above-described pipeline liquid state identification method when executing the computer-readable instructions.

[0008] A computer-readable storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the pipeline liquid state identification method described above.

[0009] In the aforementioned pipeline liquid state recognition method, apparatus, computer equipment, and storage medium, the pipeline liquid state recognition method acquires an enhanced pipeline image through an image acquisition unit. The enhanced pipeline image includes pipeline image elements of the target pipeline through which the liquid flows. A preset image analysis model is used to analyze and process the enhanced pipeline image, identifying the pipeline image deformation results of the pipeline image elements. Based on the pipeline image deformation results, the pipeline liquid state corresponding to the target pipeline is determined. This invention introduces pipeline image elements as the image background in the enhanced pipeline image, which does not interfere with the subsequent image recognition process, thus helping to improve the stability of state recognition. Simultaneously, identifying the pipeline image deformation results based on the preset image analysis model can more effectively distinguish different liquid states in the pipeline, improving the efficiency and accuracy of pipeline liquid state recognition. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic flowchart of a pipeline liquid state identification method according to an embodiment of the present invention; Figure 2 This is a pipeline enhanced image of a pipeline liquid state recognition method in one embodiment of the present invention; Figure 3 This is a schematic diagram of a pipeline liquid state identification device according to an embodiment of the present invention; Figure 4This is a schematic diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0012] 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, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] In one embodiment, such as Figure 1 As shown, a method for identifying the liquid state of a pipeline is provided, including the following steps S10-S30: S10. Obtain pipeline enhancement images through the image acquisition unit; the pipeline enhancement images include pipeline image elements for the target pipeline through which the liquid flows.

[0014] Understandably, an image acquisition unit is a device used to capture images of a target pipeline. It may include only a camera, or it may include both a camera and a light source for auxiliary imaging. The target pipeline is the pipeline whose state of whether there is liquid inside needs to be determined. A background board with a preset background pattern is placed on the side of the target pipeline away from the image acquisition unit. The preset background pattern is a specific pattern that serves as the background for the target pipeline when capturing images. The preset background pattern can be silkscreened or pasted onto the surface of the background board, or it can be set via a panel on the background board. The pipeline enhancement image is an image obtained by simultaneously capturing images of the target pipeline and the preset background pattern using the image acquisition unit. The pipeline enhancement image includes pipeline image elements corresponding to the target pipeline. The pipeline image elements refer to the basic pattern elements of the preset background pattern, which can be basic graphics distributed according to rules (such as circles, rhombuses, triangles, etc.), or a combination of a colored background and basic graphics. For example, when the preset background pattern is a dot matrix pattern of multiple dots, the pipeline image element is dots; when the preset background pattern is a dot matrix pattern of multiple rhombuses, the pipeline image element is rhombuses; when the preset background pattern is a dot matrix pattern of multiple dots on a blue background, the pipeline image element is dots and a blue background.

[0015] In one embodiment, such as Figure 2As shown, the preferred preset background pattern is a pattern of multiple dots arranged in a dot matrix, and the corresponding pipeline image element is a dot. In this case, the pipeline enhancement image includes a background pattern composed of a dot array 101 and a pipeline image of the target pipeline 102. The dot array 101 is set on the background plate surface on the side of the target pipeline 102 facing away from the image acquisition unit, used to enhance the visual display effect of the transparent pipeline. Since the dots are regular circular patterns, when observed through the transparent pipeline, they will be deformed into elliptical patterns due to refraction. Furthermore, when dots are arranged continuously, two dots are connected by a point of tangency. When the point of tangency shifts due to deformation, it is easier to observe or identify compared to solid color patterns or other patterns (such as striped patterns).

[0016] S20. The pipeline enhancement image is analyzed and processed using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements.

[0017] Understandably, a pre-trained image analysis model is a neural network model used to perform pixel-by-pixel analysis on input pipeline enhancement images and output deformation results. Pre-trained image analysis models include, but are not limited to, one of the following: Convolutional Neural Networks (CNN), Residual Networks (ResNet), and Visual Geometry Groups (VGG). Before using the pre-trained image analysis model to analyze and process pipeline enhancement images, the model needs to be trained. Multiple pipeline enhancement images in different states (e.g., no liquid, liquid with bubbles, and liquid without bubbles) are collected against the same dot array background as a set of recognition samples for the image analysis model to be trained. Each recognition sample corresponds to a pipeline liquid state label that the model needs to identify. For example, no liquid in the pipeline is labeled 0, liquid with no bubbles is labeled 1, and liquid with bubbles is labeled 2. The image analysis model to be trained has initial parameters and is a model that has not yet been trained using deep learning. After analyzing a recognition sample, the image analysis model to be trained outputs a training recognition result. The initial parameters are adjusted based on the deviation between the output training recognition result and the corresponding pipeline liquid state marker, until a pre-trained image analysis model is finally obtained. That is, when a recognition sample is input into the image analysis model to be trained, if the deviation between the output training recognition result and the corresponding pipeline liquid state marker is controlled within a preset deviation threshold (e.g., 5%), the training of the image analysis model to be trained is complete, and the pre-trained image analysis model is obtained.

[0018] The preset image analysis model takes a pipeline enhancement image as input and outputs the pipeline image element deformation results under the influence of the target pipeline. The pipeline image deformation result is a symbol representing the degree of deformation of pipeline image elements caused by the pipeline in the enhanced image. The pipeline image deformation result can be a number, a letter, or a combination of one or more of these. Different liquid and bubble contents in the pipeline result in different degrees of deformation of pipeline image elements in the enhanced image, and consequently, different pipeline image deformation results. When the target pipeline is empty, it is filled with air, and the deformation of the pipeline image elements is minimal. When the target pipeline is full, it is filled with liquid. The refractive index of liquid is greater than that of air, creating a convex lens effect, resulting in significant deformation of the pipeline image elements. When the target pipeline is a gas-liquid mixture, the bubble portion will show minimal deformation, while the liquid portion will show significant deformation.

[0019] S30. Based on the deformation result of the pipeline image, the liquid state of the pipeline corresponding to the target pipeline is obtained.

[0020] Understandably, the liquid state of a pipeline is information used to characterize whether there is liquid and whether there are air bubbles in the liquid within the target pipeline, including an empty pipeline, a full pipeline, and a gas-liquid mixture. Different degrees of deformation of pipeline image elements caused by the pipeline result in different pipeline image deformations, and correspondingly, different pipeline liquid states. For example, a pipeline image deformation result of 0 indicates that there is no liquid in the target pipeline, corresponding to an empty pipeline state. A pipeline image deformation result of 1 indicates that the target pipeline is full of liquid and has no air bubbles, corresponding to a full pipeline state. A pipeline image deformation result of 2 indicates that both liquid and air bubbles are present in the target pipeline, corresponding to a gas-liquid mixture state.

[0021] This embodiment acquires enhanced pipeline images through an image acquisition unit. The enhanced pipeline images include pipeline image elements of the target pipeline through which the liquid flows. A preset image analysis model is used to analyze and process the enhanced pipeline images, identifying the pipeline image deformation results of the pipeline image elements. Based on the pipeline image deformation results, the liquid state of the pipeline corresponding to the target pipeline is determined. This embodiment introduces pipeline image elements as the image background in the enhanced pipeline images, which does not interfere with the subsequent image recognition process and helps improve the stability of state recognition. Simultaneously, identifying the pipeline image deformation results based on the preset image analysis model can more effectively distinguish different liquid states in the pipeline, improving the efficiency and accuracy of pipeline liquid state recognition.

[0022] In one embodiment, step S20, namely, analyzing and processing the enhanced pipeline image using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements, includes: S201. Analyze the pipeline enhancement image pixel by using a preset image analysis model to identify the background pattern area and the overlapping area of ​​the pipeline background in the pipeline enhancement image. S202. Determine whether the pipeline enhancement image meets the preset pipeline enhancement conditions based on the overlapping area of ​​the background pattern area and the pipeline background. S203. When it is confirmed that the pipeline enhancement image meets the preset pipeline enhancement conditions, the pixel deformation features of the pipeline image elements in the pipeline background overlap area are extracted, and the pipeline image deformation result of the pipeline enhancement image is determined according to the pixel deformation features.

[0023] Understandably, based on whether the pipeline image elements are covered by the transparent target pipeline, the pipeline enhancement image can be divided into two regions: the background pattern region and the pipeline-background overlap region. The background pattern region refers to the image area where the pipeline image elements are not covered by the target pipeline. The pipeline-background overlap region refers to the image area where the pipeline image elements are covered by the target pipeline. By performing pixel analysis on the pipeline enhancement image using a preset image analysis model, the pixels of the pipeline image elements in the background pattern region and the pipeline-background overlap region are different. Based on the pixel distribution, the background pattern region and the pipeline-background overlap region in the pipeline enhancement image can be determined. After determining the background pattern region and the pipeline-background overlap region, it is necessary to determine whether the pipeline enhancement image meets the preset pipeline enhancement conditions. Subsequent image analysis steps can only proceed if the pipeline enhancement image meets the preset pipeline enhancement conditions. The preset pipeline enhancement conditions are pre-defined critical conditions used to evaluate whether the deformation of pipeline image elements can be effectively compared between the background pattern region and the pipeline-background overlap region in the pipeline enhancement image.

[0024] Furthermore, after performing pixel-level analysis on the pipeline enhancement image using a preset image analysis model, the pixel deformation characteristics of pipeline image elements in the pipeline background overlap region compared to the background pattern region can be analyzed. Pixels of pipeline image elements in the background pattern region remain undeformed, while the presence or absence of liquid in the target pipeline, and even the presence or absence of air bubbles in the liquid, will cause pixel deformation in the pipeline image elements in the pipeline background overlap region. Pixel deformation features characterize the degree of deformation of pixels corresponding to pipeline image elements. The pipeline background overlap region includes several pixels corresponding to pipeline image elements, and each pixel has corresponding pixel deformation features, including empty pipe deformation features and full pipe deformation features. When the pipeline enhancement image is confirmed to meet preset pipeline enhancement conditions, the pipeline image deformation result of the pipeline enhancement image can be determined based on the pixel deformation features of all pixels corresponding to pipeline image elements.

[0025] The preset image analysis model in this embodiment determines whether the pipeline enhancement image meets the preset pipeline enhancement conditions based on the overlapping area of ​​the background pattern region and the pipeline background, ensuring the effectiveness and analyzability of the pipeline enhancement image. Simultaneously, it determines the pipeline image deformation result by comprehensively considering the pixel deformation features in the overlapping area of ​​the pipeline background, improving the accuracy of the preset image analysis model in analyzing pipeline enhancement images.

[0026] In one embodiment, the pipeline image deformation result includes a first state identifier, a second state identifier, and a third state identifier; step S30, namely, determining the pipeline liquid state corresponding to the target pipeline based on the pipeline image deformation result, includes: S301. When all the pixel deformation features are empty pipe deformation features, the deformation result of the pipeline image is determined as the first state identifier, and the liquid state of the pipeline is determined as the empty pipe state. S302. When all the pixel deformation features are full pipe deformation features, the deformation result of the pipeline image is determined as the second state identifier, and the liquid state of the pipeline is determined as full pipe state. S303. When the pixel deformation feature includes both empty pipe deformation feature and full pipe deformation feature, the pipeline image deformation result is determined as a third state identifier, the pipeline liquid state is determined as a gas-liquid mixture state, and the output gas-liquid ratio is determined.

[0027] Understandably, the pipeline image deformation results include a first state identifier, a second state identifier, and a third state identifier. The first state identifier indicates that the deformation in the overlapping region of the pipeline background is entirely caused by an empty pipe, represented by "0". The second state identifier indicates that the deformation in the overlapping region of the pipeline background is entirely caused by a full pipe, represented by "1". The third state identifier indicates that the deformation in the overlapping region of the pipeline background is caused by both partially empty and partially full pipes, represented by "2". The overlapping region of the pipeline background includes several pixels corresponding to pipeline image elements. Each pixel has a corresponding pixel deformation feature, including empty pipe deformation features and full pipe deformation features. The empty pipe deformation feature indicates that the deformation degree of the pixel corresponding to the pipeline image element conforms to the deformation caused by a pipe without liquid, i.e., the pixel deformation is caused by refraction between the pipe and air. The full pipe deformation feature indicates that the deformation degree of the pixel corresponding to the pipeline image element conforms to the deformation caused by a pipe filled with liquid, i.e., the pixel deformation is caused by refraction between the pipe and liquid.

[0028] In one embodiment, when the pipeline image deformation result is 0, it indicates that the pixel deformation features of all pixels corresponding to pipeline image elements in the overlapping region of the pipeline background are empty pipe deformation features, and the corresponding pipeline liquid state is an empty pipe state. When the pipeline image deformation result is 1, it indicates that the pixel deformation features of all pixels corresponding to pipeline image elements in the overlapping region of the pipeline background are full pipe deformation features, and the corresponding pipeline liquid state is a full pipe state. When the pipeline image deformation result is 2, it indicates that the pixel deformation features of all pixels corresponding to pipeline image elements in the overlapping region of the pipeline background include both empty pipe deformation features and full pipe deformation features, and the corresponding pipeline liquid state is a gas-liquid mixed state. In this case, the gas-liquid mixing ratio can be determined based on the ratio between the number of empty pipe pixels and the number of full pipe pixels. The number of empty pipe pixels is the number of pixels corresponding to the empty pipe deformation feature, and the number of full pipe pixels is the number of pixels corresponding to the full pipe deformation feature. The gas-liquid mixing ratio is a parameter used to characterize the amount of bubbles when bubbles exist in the liquid of the target pipeline.

[0029] The pipeline image deformation results in this embodiment reflect various scenarios of pixel deformation characteristics at different pixels, allowing for accurate determination of the corresponding pipeline liquid state. Furthermore, for pipelines in a gas-liquid mixed state, the gas-liquid mixing ratio can be further determined, achieving quantitative analysis based on qualitative analysis.

[0030] In one embodiment, step S202, namely determining whether the pipeline enhancement image meets the preset pipeline enhancement conditions based on the overlapping area of ​​the background pattern area and the pipeline background, includes: S2021. When the background element pattern in the background pattern area is identified, the element coverage of the background element pattern in the pipeline enhancement image is obtained, and it is determined whether the element coverage reaches a preset coverage threshold. S2022. When it is confirmed that the element coverage reaches the preset coverage threshold, element anomaly detection and pipeline direction detection are performed on the overlapping area of ​​the background pattern area and the pipeline background to obtain the detection results. S2023. When the detection result is qualified, confirm that the pipeline enhancement image meets the preset pipeline enhancement conditions. S2024. When the detection result is unqualified, it is confirmed that the pipeline enhancement image does not meet the preset pipeline enhancement conditions.

[0031] Understandably, in the process of determining whether the pipeline enhancement image meets the preset pipeline enhancement conditions, it is necessary to make judgments simultaneously from both the overall dimension and the partial dimension. In the overall dimension, when the background element pattern in the background pattern area is recognized, the element coverage of the background element pattern in the pipeline enhancement image is obtained, and it is judged whether the element coverage reaches the preset coverage threshold. The pipeline enhancement image can be selected in different sizes and shapes according to needs, such as rectangles, rhombuses, and squares of specific sizes. As Figure 2 shown, the pipeline enhancement image is preferably a square image of a specific size, that is, the pipeline enhancement image input into the preset image analysis model is a unified square image. The length and width of the square are the same, which is beneficial to the design and update of the deep learning algorithm itself. When training the neural network model, the error caused by the change of the image length and width can be avoided. The pipeline image elements distributed in the pipeline background coincidence area and the background pattern area in the pipeline enhancement image are the same basic pattern element, and the basic pattern element of the preset background pattern can also be selected according to needs. The actually selected basic pattern element is called the pipeline image element, and the background element pattern is a pattern formed by arranging the actually selected basic pattern elements regularly in the background pattern area. The element coverage of the background element pattern in the pipeline enhancement image refers to the ratio of the pipeline image elements distributed in the background pattern area to all the pipeline image elements in the pipeline enhancement image. For example, when the pipeline image element in Figure 2 is a dot, the element coverage represents the ratio of the number of dots distributed in the background pattern area to the total number of dots in the pipeline enhancement image. The preset coverage threshold is a critical value preset for determining whether the pipeline image elements are distributed reasonably. When the element coverage reaches the preset coverage threshold, it indicates that the division of the pipeline background coincidence area and the background pattern area in the pipeline enhancement image is reasonable, avoiding the pipeline image deformation result where an invalid background appears and the pipeline image elements cannot be recognized. When the element coverage has not reached the preset coverage threshold, it is confirmed that the pipeline enhancement image does not meet the preset pipeline enhancement conditions.

[0032] When the element coverage reaches the preset coverage threshold, subsequent judgments are further made in the partial dimension. At this time, two aspects of detection, namely element anomaly detection and pipeline direction detection, need to be carried out in the background pattern area and the pipeline background coincidence area to obtain the final detection result. Element anomaly detection is a process used to verify whether the distribution of pipeline image elements in the pipeline enhancement image can effectively compare the deformation effects between the background pattern area and the pipeline background coincidence area. Pipeline direction detection refers to a process used to verify whether the pipeline direction in the pipeline background coincidence area can significantly cause deformation of the pipeline image elements. When the final detection result is qualified, it is confirmed that the pipeline enhancement image meets the preset pipeline enhancement conditions; when the final detection result is unqualified, it is confirmed that the pipeline enhancement image does not meet the preset pipeline enhancement conditions.

[0033] This embodiment comprehensively judges whether the pipeline enhancement image meets the preset pipeline enhancement conditions from both overall and partial dimensions. This ensures that the pipeline enhancement image meets specific standards and requirements, which helps to accurately and quickly analyze the pipeline image deformation results in subsequent pipeline enhancement.

[0034] In one embodiment, step S2022, namely, performing element anomaly detection and pipeline direction detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain the detection result, includes: S20221. Perform element anomaly detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain the element anomaly detection result. S20222, Perform pipeline direction detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain pipeline anomaly detection results; S20223. Determine the detection results based on the element anomaly detection results and the pipeline anomaly detection results.

[0035] Understandably, after performing element anomaly detection and pipeline orientation detection on the overlapping areas of the background pattern region and the pipeline background, respectively, two detection results are obtained: element anomaly detection result and pipeline anomaly detection result. Pipeline image elements in the pipeline enhancement image are distributed according to specific rules, exhibiting features such as distribution interval, size, and arrangement direction. The element anomaly detection result characterizes whether the distribution of pipeline image elements meets requirements, including distribution interval, size, and boundary distance. The pipeline anomaly detection result characterizes whether the distribution orientation of the pipeline image elements meets requirements with the placement orientation of the target pipeline. Both element anomaly detection results and pipeline anomaly detection results may be either qualified or unqualified. When both element anomaly detection results and pipeline anomaly detection results are qualified, the final detection result can be confirmed as qualified. When either element anomaly detection result or pipeline anomaly detection result is unqualified, the final detection result can be confirmed as unqualified.

[0036] This embodiment comprehensively considers the influencing factors of both pipeline image elements and target pipeline, ensuring the rigor and rationality of the detection results, improving the quality of pipeline enhanced images, and helping to improve the accuracy of pipeline liquid state identification.

[0037] In one embodiment, step S20221, namely, performing element anomaly detection on the overlapping area of ​​the background pattern region and the pipeline background to obtain the element anomaly detection result, further includes: S202211. Perform element spacing recognition on the background element pattern in the background pattern area to obtain the spacing recognition result; S202212. Identify the element boundary spacing of the background element pattern in the background pattern area to obtain the boundary spacing identification result; S202213. Identify the element size of the background element pattern in the background pattern area to obtain the size result, identify the pipe diameter of the target pipe in the pipeline background overlapping area to obtain the pipe diameter result, and obtain the diameter ratio result based on the size result and the pipe diameter result; S202214. Determine the element abnormality detection result according to the interval identification result, the boundary spacing identification result and the diameter ratio result.

[0038] Understandably, the regular distribution of pipeline image elements in the background pattern area forms the background element pattern and there is no deformation, so it is helpful to perform element interval identification, element boundary spacing identification and diameter value analysis to obtain the corresponding interval identification result, boundary spacing identification result and diameter ratio result respectively. The interval identification result is used to characterize whether the distribution interval of pipeline image elements meets the requirement of effectively comparing the deformation effects between the background pattern area and the pipeline background overlapping area. The boundary spacing identification result is used to characterize whether the distance of the distribution interval of pipeline image elements meets the requirement of effectively comparing the deformation effects between the background pattern area and the pipeline background overlapping area. The diameter ratio result is used to characterize whether the size of pipeline image elements meets the requirement of effectively comparing the deformation effects between the background pattern area and the pipeline background overlapping area. These three results, namely the interval identification result, the boundary spacing identification result and the diameter ratio result, jointly determine the element abnormality detection result. When all three results are normal, it can be determined that the element abnormality detection result is qualified for detection. When at least one of these three results is abnormal, it can be determined that the element abnormality detection result is unqualified for detection.

[0039] In one embodiment, as Figure 2 shown, the pipeline enhanced image is preferably a square image of a specific size, and the pipeline image elements are preferably dots, which are arranged regularly in the horizontal (row direction) and vertical (column direction) to form a dot array. From the perspective of pixel point analysis of the pipeline enhanced image through a preset image analysis model, the pixel values of the pixels corresponding to the dots and the pixel values of the pixels corresponding to the blank intervals are different. The pixels corresponding to the dots in the pipeline background overlapping area will be deformed compared with the pixels corresponding to the dots in the background pattern area, and there are also differences in the pixel deformation characteristics between the empty pipe deformation characteristics and the full pipe deformation characteristics. This embodiment uses the dot array as the background, which helps to magnify the refraction deformation characteristics in different pipeline liquid states, makes it easier to identify whether there is liquid and whether there are bubbles in the pipeline, improves the efficiency and accuracy of pipeline liquid state identification, and at the same time can avoid background interference and helps to improve the stability of pipeline liquid state identification.

[0040] Furthermore, in this embodiment, when identifying the element spacing of the background pattern in the background pattern area, the distribution of the pixels corresponding to the dots and the pixels corresponding to the blank spaces is used to determine whether there are gaps between the dots in the column direction and the row direction of the dot array in the background pattern area. When there are no gaps in the column direction (adjacent dots in the same column are arranged continuously) and there are gaps in the row direction (adjacent dots in the same row are arranged with gaps), the gap identification result is normal. Figure 2 As shown, continuous arrangement of dots in the column direction helps identify continuous deformations, while spacing of dots in the row direction prevents them from deforming simultaneously and interfering with each other. When there are gaps in the column direction and / or in the row direction, the gap identification result is abnormal.

[0041] Furthermore, in this embodiment, when identifying the element boundary spacing of background elements in the background pattern area, the diameter of the dot (the distance between the two farthest pixels corresponding to the same dot) can be obtained by mapping the corresponding pixels of the dots. Given a row-wise interval, the element boundary spacing (the distance between the two closest pixels between two adjacent dots in the same row) can be obtained by mapping the corresponding pixels of two adjacent dots in the same row. The relationship between the dot diameter and the element boundary spacing is then determined. When the element boundary spacing is greater than or equal to the dot diameter, the boundary spacing identification result is normal. When the element boundary spacing is less than the dot diameter, the boundary spacing identification result is abnormal. An excessively small element boundary spacing may cause adjacent dots to easily influence each other during deformation, leading to inaccurate identification of the pipeline image deformation results.

[0042] Furthermore, in this embodiment, when identifying the element size of the background pattern in the background pattern area, the diameter of the dot (the distance between the two farthest pixels corresponding to the same dot) can be obtained by mapping the pixels corresponding to the dots, and the dot diameter is determined as the size result. Simultaneously, in this embodiment, when identifying the pipe diameter in the overlapping area of ​​the pipe background, the diameter of the target pipe (the distance between the two farthest pixels corresponding to the deformed dots in the pipe diameter direction) can be obtained by mapping the pixels corresponding to the dots in the overlapping area of ​​the pipe background, and the target pipe diameter is determined as the pipe diameter result. The ratio between the size result and the pipe diameter result is calculated to obtain the diameter ratio, and it is determined whether the diameter ratio is less than or equal to 1 (e.g., the diameter ratio is 1 / N, where N is a positive integer greater than or equal to 1, preferably N is 2 or 3). When the diameter ratio is less than or equal to 1, the diameter ratio result is normal. In this case, the pipe diameter can at least accommodate one dot, and the degree of deformation of the dot can be easily identified by the change in the arc of the tangent dot. Figure 2In the enhanced pipeline image shown, the pipe diameter can accommodate 2 dots. When the diameter ratio is greater than 1, the diameter ratio result is abnormal, the dot diameter exceeds the pipe diameter, and it is not easy to identify the deformation degree of the dots.

[0043] Based on the interval recognition result, the boundary spacing recognition result, and the diameter ratio result, this embodiment jointly determines the element abnormality detection result, which can ensure the quality of the enhanced pipeline image, help effectively distinguish the deformation of the pipeline image between the background pattern area and the pipeline background overlapping area, and thus improve the state recognition accuracy.

[0044] In one embodiment, in step S20222, that is, performing pipeline direction detection on the background pattern area and the pipeline background overlapping area to obtain a pipeline abnormality detection result, including: S202221. Extract the included angle feature between the element arrangement direction of the background element pattern in the background pattern area and the extension direction of the target pipeline in the pipeline background overlapping area, and perform included angle recognition based on the extracted included angle feature to obtain an included angle recognition result; S202222. Determine the pipeline abnormality detection result based on the included angle recognition result.

[0045] Understandably, in the embodiment shown in Figure 2 , the element arrangement direction of the background element pattern in the background pattern area refers to the direction of continuous arrangement of dots, that is, the column direction of the dot array. The extension direction of the target pipeline refers to the extension direction of the liquid channel. After performing pixel point analysis on the enhanced pipeline image through a preset image analysis model, the included angle feature between the element arrangement direction and the extension direction of the target pipeline in the pipeline background overlapping area can also be extracted, and included angle recognition is performed based on the extracted included angle feature to obtain an included angle recognition result. The included angle recognition result is specifically used to represent whether there is an included angle between the extension line of the column direction of the dot array and the extension line of the extension direction of the target pipeline, which can be represented by a specific angle or a character. For example, "1" indicates that there is an included angle, and "0" indicates that there is no included angle. When the included angle recognition result is that there is an included angle, it indicates that the extension line of the column direction of the dot array intersects with the extension line of the extension direction of the target pipeline. At this time, the pipeline abnormality detection result is qualified. When the included angle recognition result is that there is an included angle, it indicates that the extension line of the column direction of the dot array is parallel to the extension line of the extension direction of the target pipeline. At this time, the pipeline abnormality detection result is unqualified. Particularly preferably, when the included angle recognition result is that there is an included angle and the included angle is 90 degrees, it indicates that the extension line of the column direction of the dot array is perpendicular to the extension line of the extension direction of the target pipeline. At this time, the pipeline abnormality detection result is qualified and meets the best requirement for increasing the deformation effect between the background pattern area and the pipeline background overlapping area.

[0046] This embodiment identifies the target pipeline based on the angle between the element arrangement direction of the background pattern in the background pattern area and the extension direction of the target pipeline in the overlapping area of ​​the pipeline background. This can eliminate abnormal pipeline direction of the target pipeline, ensure the quality of the pipeline enhancement image, avoid invalid image analysis, and help improve the accuracy of state recognition.

[0047] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0048] In one embodiment, a pipeline liquid state identification device is provided, which corresponds one-to-one with the pipeline liquid state identification method described in the above embodiments. For example... Figure 3 As shown, the pipeline liquid status identification device includes an image acquisition module 10, an image analysis module 20, and a status determination module 30. Detailed descriptions of each functional module are as follows: Image acquisition module 10 is used to acquire pipeline enhancement images through image acquisition unit; the pipeline enhancement images include pipeline image elements for the target pipeline through which liquid flows; Image analysis module 20 is used to analyze and process the enhanced pipeline image through a preset image analysis model, and identify the pipeline image deformation results of the pipeline image elements; The state determination module 30 is used to determine the liquid state of the pipeline corresponding to the target pipeline based on the pipeline image deformation result.

[0049] In one embodiment, the image analysis module 20 includes: The pixel analysis unit is used to perform pixel analysis on the pipeline enhancement image using a preset image analysis model, and to identify the background pattern area and the overlapping area of ​​the pipeline background in the pipeline enhancement image. The condition judgment unit is used to determine whether the pipeline enhancement image meets the preset pipeline enhancement conditions based on the overlapping area of ​​the background pattern area and the pipeline background. The pipeline image deformation result determination unit is used to extract the pixel deformation features of pipeline image elements in the pipeline background overlap area when it is confirmed that the pipeline enhancement image meets the preset pipeline enhancement conditions, and determine the pipeline image deformation result of the pipeline enhancement image based on the pixel deformation features.

[0050] In one embodiment, the state determination module 30 includes: The empty pipe state determination unit is used to determine the deformation result of the pipeline image as a first state identifier when the deformation features of the pixels are all empty pipe deformation features, and to determine the liquid state of the pipeline as an empty pipe state. The full pipe state determination unit is used to determine the deformation result of the pipeline image as a second state identifier when the deformation features of the pixels are all full pipe deformation features, and to determine the liquid state of the pipeline as full pipe state. The mixed state determination unit is used to determine the deformation result of the pipeline image as a third state identifier when the deformation feature of the pixel point simultaneously includes the deformation feature of an empty pipe and the deformation feature of a full pipe, and to determine the liquid state of the pipeline as a gas-liquid mixed state, and to output the gas-liquid ratio.

[0051] In one embodiment, the image analysis module 20 further includes: The element coverage determination unit is used to obtain the element coverage of the background element pattern in the pipeline enhancement image when the background element pattern in the background pattern area is identified, and to determine whether the element coverage reaches a preset coverage threshold. The area detection unit is used to perform element anomaly detection and pipeline direction detection on the overlapping area of ​​the background pattern area and the pipeline background when it is confirmed that the element coverage has reached a preset coverage threshold, and obtain the detection result. The inspection pass determination unit is used to confirm that the pipeline enhancement image meets the preset pipeline enhancement conditions when the inspection result is qualified. The non-compliance determination unit is used to confirm that the pipeline enhancement image does not meet the preset pipeline enhancement conditions when the detection result is non-compliance.

[0052] In one embodiment, the image analysis module 20 further includes: An element anomaly detection unit is used to perform element anomaly detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain the element anomaly detection result. The pipeline orientation detection unit is used to perform pipeline orientation detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain pipeline anomaly detection results. The detection result determination unit is used to determine the detection result based on the element anomaly detection result and the pipeline anomaly detection result.

[0053] In one embodiment, the image analysis module 20 further includes: An element spacing recognition unit is used to perform element spacing recognition on the background element pattern in the background pattern area to obtain the spacing recognition result. An element boundary spacing recognition unit is used to recognize the element boundary spacing of the background element pattern in the background pattern area and obtain the boundary spacing recognition result. The diameter ratio determination unit is used to identify the element size of the background element pattern in the background pattern area to obtain the size result, identify the pipe diameter of the target pipe in the overlapping area of ​​the pipe background to obtain the pipe diameter result, and obtain the diameter ratio result based on the size result and the pipe diameter result. An element anomaly detection result determination unit is used to determine the element anomaly detection result based on the interval identification result, the boundary spacing identification result, and the diameter ratio result.

[0054] In one embodiment, the image analysis module 20 further includes: Angle recognition unit is used to extract the angle features between the element arrangement direction of the background element pattern in the background pattern area and the extension direction of the target pipeline in the overlapping area of ​​the pipeline background, and to perform angle recognition based on the extracted angle features to obtain the angle recognition result. The pipeline anomaly detection result determination unit is used to determine the pipeline anomaly detection result based on the included angle identification result.

[0055] Specific limitations regarding the pipeline liquid condition identification device can be found in the limitations of the pipeline liquid condition identification method described above, and will not be repeated here. Each module in the aforementioned pipeline liquid condition identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0056] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a readable storage medium and internal memory. The readable storage medium stores an operating system, computer-readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The database stores data related to the pipeline liquid state identification method. The network interface communicates with external terminals via a network connection. When the computer-readable instructions are executed by the processor, a pipeline liquid state identification method is implemented. The readable storage medium provided in this embodiment includes both non-volatile and volatile readable storage media.

[0057] In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor performs the following steps when executing the computer-readable instructions: Enhanced pipeline images are acquired through an image acquisition unit; the enhanced pipeline images include pipeline image elements for the target pipeline through which the liquid flows. The pipeline enhancement image is analyzed and processed using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements; The liquid state of the pipeline corresponding to the target pipeline is obtained based on the pipeline image deformation results.

[0058] In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided. The readable storage media provided in this embodiment include non-volatile readable storage media and volatile readable storage media. The readable storage media stores computer-readable instructions, which, when executed by one or more processors, perform the following steps: Enhanced pipeline images are acquired through an image acquisition unit; the enhanced pipeline images include pipeline image elements for the target pipeline through which the liquid flows. The pipeline enhancement image is analyzed and processed using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements; The liquid state of the pipeline corresponding to the target pipeline is obtained based on the pipeline image deformation results.

[0059] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When executed, these computer-readable instructions can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0060] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0061] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for identifying the liquid state of a pipeline, characterized in that, include: Enhanced pipeline images are acquired through an image acquisition unit; the enhanced pipeline images include pipeline image elements for the target pipeline through which the liquid flows. The pipeline enhancement image is analyzed and processed using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements; The liquid state of the pipeline corresponding to the target pipeline is obtained based on the pipeline image deformation results.

2. The pipeline liquid state identification method as described in claim 1, characterized in that, The step of analyzing and processing the enhanced pipeline image using a preset image analysis model to identify the pipeline image deformation results of the pipeline image elements includes: The pipeline enhancement image is analyzed pixel by pixel using a preset image analysis model to identify the background pattern area and the overlapping area of ​​the pipeline background in the pipeline enhancement image. Based on the overlapping area of ​​the background pattern area and the pipeline background, determine whether the pipeline enhancement image meets the preset pipeline enhancement conditions; When it is confirmed that the pipeline enhancement image meets the preset pipeline enhancement conditions, the pixel deformation features of the pipeline image elements in the pipeline background overlap area are extracted, and the pipeline image deformation result of the pipeline enhancement image is determined according to the pixel deformation features.

3. The pipeline liquid state identification method as described in claim 2, characterized in that, The pipeline image deformation result includes a first state identifier, a second state identifier, and a third state identifier; The step of determining the fluid state of the pipeline corresponding to the target pipeline based on the pipeline image deformation result includes: When all the pixel deformation features are empty tube deformation features, the deformation result of the pipeline image is determined as the first state identifier, and the liquid state of the pipeline is determined as the empty tube state. When all the pixel deformation features are full-pipe deformation features, the deformation result of the pipeline image is determined as the second state identifier, and the liquid state of the pipeline is determined as the full-pipe state; When the pixel deformation features include both empty pipe deformation features and full pipe deformation features, the pipeline image deformation result is determined as a third state identifier, the pipeline liquid state is determined as a gas-liquid mixture state, and the output gas-liquid ratio is determined.

4. The pipeline liquid state identification method as described in claim 2, characterized in that, The step of determining whether the pipeline enhancement image meets the preset pipeline enhancement conditions based on the overlapping area of ​​the background pattern area and the pipeline background includes: When the background element pattern in the background pattern area is identified, the element coverage of the background element pattern in the pipeline enhancement image is obtained, and it is determined whether the element coverage reaches a preset coverage threshold. When it is confirmed that the element coverage reaches the preset coverage threshold, element anomaly detection and pipeline direction detection are performed on the overlapping area of ​​the background pattern area and the pipeline background to obtain the detection results; When the detection result is qualified, it is confirmed that the pipeline enhancement image meets the preset pipeline enhancement conditions; If the detection result is unqualified, it is confirmed that the pipeline enhancement image does not meet the preset pipeline enhancement conditions.

5. The pipeline liquid state identification method as described in claim 4, characterized in that, The step of performing element anomaly detection and pipeline direction detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain the detection results includes: Element anomaly detection is performed on the overlapping area of ​​the background pattern area and the pipeline background to obtain the element anomaly detection result; Pipeline direction detection is performed on the overlapping area of ​​the background pattern area and the pipeline background to obtain pipeline anomaly detection results; The detection results are determined based on the element anomaly detection results and the pipeline anomaly detection results.

6. The pipeline liquid state identification method as described in claim 5, characterized in that, The step of performing element anomaly detection on the overlapping area of ​​the background pattern area and the pipeline background to obtain element anomaly detection results includes: Element spacing recognition is performed on the background element patterns in the background pattern area to obtain the spacing recognition result; The boundary spacing of background elements in the background pattern area is identified to obtain the boundary spacing identification result. The background element pattern in the background pattern area is identified by element size to obtain size result, and the target pipe in the overlapping area of ​​the pipe background is identified by pipe diameter to obtain pipe diameter result. Based on the size result and the pipe diameter result, the diameter ratio result is obtained. The element anomaly detection result is determined based on the interval identification result, the boundary spacing identification result, and the diameter ratio result.

7. The pipeline liquid state identification method as described in claim 5, characterized in that, The step of detecting the pipeline direction in the overlapping area of ​​the background pattern region and the pipeline background to obtain pipeline anomaly detection results includes: Extract the angle features between the element arrangement direction of the background element pattern in the background pattern area and the extension direction of the target pipeline in the overlapping area of ​​the pipeline background, and perform angle recognition based on the extracted angle features to obtain the angle recognition result; The pipeline anomaly detection result is determined based on the angle recognition result.

8. A pipeline liquid status identification device, characterized in that, include: An image acquisition module is used to acquire enhanced pipeline images through an image acquisition unit; the enhanced pipeline images include pipeline image elements for a target pipeline through which liquid flows. The image analysis module is used to analyze and process the enhanced pipeline image using a preset image analysis model, and to identify the pipeline image deformation results of the pipeline image elements. The state determination module is used to determine the liquid state of the pipeline corresponding to the target pipeline based on the pipeline image deformation result.

9. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that, When the processor executes the computer-readable instructions, it implements the pipeline liquid state identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-readable instructions, characterized in that, When the computer-readable instructions are executed by one or more processors, the one or more processors cause the pipeline liquid state identification method as described in any one of claims 1 to 7.

Images