Bottled liquid detection system and method

Abstract

The invention relates to the technical field of bottled liquid detection, and particularly provides a bottled liquid detection system and method. A rotating assembly; the heating assembly is arranged on the periphery of the bottled liquid to be detected; the cooling assembly is arranged on the periphery of the bottled liquid to be detected; a thermal imaging camera; the controller is used for carrying out controlled convection induction on the to-be-detected bottled liquid placed on the objective table, acquiring thermal imaging images of the to-be-detected bottled liquid at multiple moments through the thermal imaging camera in the controlled convection induction process, acquiring convection velocity field information corresponding to each thermal imaging image based on the thermal imaging images, and sending the convection velocity field information to the thermal imaging camera; the dynamic characteristics of the convection velocity field information are related to the physicochemical properties of the bottled liquid to be detected; the system can effectively solve the problem that the predicted parameter index cannot accurately reflect the actual parameter index of the bottled liquid due to the fact that the influence of the convective dynamic data on the internal temperature distribution of the bottled liquid cannot be sensed.

Description

Technical Field

[0001] This application relates to the field of bottled liquid detection technology, and more specifically, to a bottled liquid detection system and method. Background Technology

[0002] Bottled liquids inevitably undergo changes in volume and density during storage due to evaporation and other factors. For liquid foods, these changes in volume and density can alter texture, consistency, and other aspects of their mouthfeel.

[0003] Currently, the testing of bottled liquids typically involves opening the bottle and taking samples, which inevitably damages the appearance and packaging. To avoid this damage, related technologies use infrared thermal imaging to measure the infrared radiation energy of objects and convert it into a visual thermal image. The temperature distribution on the bottled liquid surface is then identified based on the thermal image, and finally, bottled liquid parameters are predicted based on this temperature distribution. However, these technologies only focus on the static surface temperature distribution when predicting bottled liquid parameters. Convection within the bottled liquid can cause uneven temperature distribution, meaning the surface temperature distribution obtained from the thermal image cannot accurately reflect the actual internal temperature distribution. Furthermore, these technologies lack effective means to deeply analyze how the temperature distribution of the bottled liquid is determined by its intrinsic properties and how these temperature changes reflect the liquid's flow state. Therefore, these technologies suffer from low accuracy and reliability in predicting bottled liquid parameters because they cannot perceive the impact of dynamic convection data on the internal temperature distribution of the bottled liquid, resulting in inaccurate predictions.

[0004] Currently, there is no effective technical solution to the above-mentioned problems. It should be noted that the information disclosed in this section is only for understanding the background of the present invention and therefore may include information that does not constitute prior art. Summary of the Invention

[0005] The purpose of this application is to provide a bottled liquid detection system and method that can effectively solve the problem that the predicted parameters cannot accurately reflect the actual parameters of the bottled liquid due to the inability to perceive the influence of convection dynamic data on the internal temperature distribution of the bottled liquid, resulting in low accuracy and reliability of bottled liquid parameter prediction.

[0006] In a first aspect, this application provides a bottled liquid detection system, comprising: The stage is used to hold the bottled liquid to be tested. A rotating assembly connected to the stage; A heating element is positioned around the outer periphery of the bottled liquid to be tested. The cooling component is located on the outer periphery of the bottled liquid to be tested, and its installation height is different from that of the heating component. A thermal imaging camera is mounted on one side of the stage; The controller is used to induce controlled convection in a bottled liquid to be tested placed on a stage, so as to generate convection within the bottled liquid. The controlled convection induction includes heating the bottled liquid using a heating component, cooling the bottled liquid using a cooling component, and rotating the bottled liquid using a rotating component. It is also used to acquire thermal imaging images of the bottled liquid at multiple moments during the controlled convection induction process using a thermal imaging camera, to acquire convection velocity field information corresponding to each thermal imaging image, and to predict the parameter indicators of the bottled liquid based on all the convection velocity field information.

[0007] This application provides a bottled liquid detection system that obtains convection velocity field information directly related to the physical and chemical properties of the liquid by actively inducing convection and using a thermal imaging camera to capture dynamic thermal images of the convection process. In other words, this application is equivalent to introducing dynamic convection data when predicting the parameters of bottled liquids. Therefore, this application can effectively solve the problem that the predicted parameters cannot accurately reflect the actual parameters of the bottled liquid because the influence of dynamic convection data on the internal temperature distribution of the bottled liquid cannot be perceived, resulting in low accuracy and reliability of bottled liquid parameter prediction.

[0008] Optionally, the steps for obtaining the convection velocity field information corresponding to each thermal imaging image include: A1. Based on the preset division rules, all thermal imaging images are divided into regions so that all thermal imaging images are evenly divided into multiple test regions. A2. For each test area, obtain the velocity vector of the test area based on the gradient of all pixels in the test area and the difference between the center pixel and the non-center pixel in the test area. A3. For each thermal imaging image, integrate all the corresponding velocity vectors to obtain the convective velocity field information corresponding to that thermal imaging image.

[0009] This technical solution divides the thermal imaging image into regions and calculates the velocity vector of each region to be measured, treating the velocity vector of all pixels within the same region as consistent. In other words, this technical solution is equivalent to performing neighborhood pooling and downsampling on the thermal imaging image, using the velocity vector of the region to be measured instead of calculating the velocity vector of each pixel, thereby reducing the amount of data processing required to obtain convective velocity field information and improving the efficiency of obtaining convective velocity field information.

[0010] Optionally, step A1 includes: A11. Perform edge detection on all thermal imaging images to obtain the bottle outline of the liquid to be detected. A12. Determine the internal liquid region of the bottled liquid to be detected in all thermal imaging images based on the bottle outline and bottle thickness information. A13. Divide the internal liquid region into regions based on preset division rules so that all thermal imaging images are divided into multiple regions to be tested.

[0011] This technical solution divides the thermal imaging image into regions and calculates the velocity vector of each region to be measured, treating the velocity vector of all pixels within the same region as consistent. In other words, this technical solution is equivalent to performing neighborhood pooling and downsampling on the thermal imaging image, using the velocity vector of the region to be measured instead of calculating the velocity vector of each pixel, thereby reducing the amount of data processing required to obtain convective velocity field information and improving the efficiency of obtaining convective velocity field information.

[0012] Optionally, step A13 includes: A131. Divide the internal liquid region into regions based on preset division rules so that all thermal imaging images are divided into multiple initial division regions. A132. For each initially divided region, obtain its internal gradient information of temperature distribution, dynamic boundary change information, and motion trajectory continuity information in different thermal imaging images; A133. For each initially defined region, analyze whether the initially defined region is a non-convective region based on the corresponding internal gradient information of temperature distribution, dynamic change information of boundary, and continuity information of motion trajectory. If it is, remove the initially defined region; otherwise, take the initially defined region as the region to be measured. Non-convective regions include bubble regions and sediment regions.

[0013] This technical solution accurately captures the unique thermal and kinematic characteristics of bubble and sediment regions by acquiring internal gradient information of temperature distribution, dynamic change information of boundaries, and continuity information of motion trajectory within the initially divided region. Therefore, this technical solution can accurately classify the initially divided region based on this multi-dimensional information to effectively distinguish between convective and non-convective regions. In other words, this technical solution can ensure that the input data for subsequent calculation of convective velocity field information only includes the liquid portion that actually participates in convection, thereby avoiding noise and errors introduced by non-convective regions, and thus effectively improving the accuracy and reliability of convective velocity field information acquisition and parameter prediction.

[0014] Optionally, step A2 includes: A21. For each test area, obtain the spatial temperature gradient information of each pixel in the test area, and obtain the temporal temperature difference information between the non-center pixel and the center pixel in the test area. A22. For each test area, analyze whether the temperature field distribution of the test area is uniform based on the spatial temperature gradient information and the temporal temperature difference information. If so, obtain the velocity vector of the test area based on the gradient of all pixels in the test area and the difference between the center pixel and the non-center pixel in the test area. If not, proceed to step A23. The condition for judging the uniformity of the temperature field distribution is: the spatial temperature gradient information is less than the preset temperature gradient and the temporal temperature difference information is less than the preset temperature difference. A23. Divide the test area with uneven temperature field distribution into multiple local sub-regions; A24. Obtain the local velocity vector corresponding to each local sub-region based on the gradient of all pixels in the local sub-region and the difference between the center pixel and non-center pixels in the local sub-region. Integrate all local velocity vectors corresponding to the same test region to obtain the velocity vector corresponding to the test region.

[0015] Optionally, the step of acquiring the convective velocity field information corresponding to each thermal imaging image also includes steps performed before step A1: A4. Obtain the liquid type information of the bottled liquid to be tested; A5. Determine the preset division rules based on the liquid type information.

[0016] Optionally, the step of predicting the parameter indicators of the bottled liquid to be tested based on all convection velocity field information includes: B1. Obtain the brand and formula information of the bottled liquid to be tested; B2. Determine parameter prediction strategies based on brand and formula information; B3. Utilize parameter prediction strategies to predict the parameter indicators of the bottled liquid to be tested based on all convective velocity field information.

[0017] Optionally, the step of inducing controlled convection in the bottled liquid to be tested placed on the stage to generate convection within the bottled liquid includes: C1. Obtain the actual heating temperature of the heating component, the actual cooling temperature of the cooling component, and the actual rotation speed of the stage; C2. Based on the first PID control algorithm, the heating power of the heating component is adjusted according to the actual heating temperature and the preset target heating temperature. Based on the second PID control algorithm, the cooling power of the cooling component is adjusted according to the actual cooling temperature and the preset target cooling temperature. Based on the third PID control algorithm, the operating parameters of the rotating component are adjusted according to the actual rotation speed and the preset target rotation speed, so as to induce controlled convection and generate convection within the bottled liquid to be tested placed on the stage.

[0018] Optionally, the parameters include density, viscosity, and mouthfeel.

[0019] Secondly, this application also provides a method for detecting bottled liquids, which is applied to a bottled liquid detection system provided in the first aspect above. The method includes the following steps: S1. Controlled convection induction is performed on the bottled liquid to be tested placed on the stage to generate convection within the bottled liquid. Controlled convection induction includes heating the bottled liquid to be tested using a heating component, cooling the bottled liquid to be tested using a cooling component, and rotating the bottled liquid to be tested using a rotating component. S2. During the controlled convection induction process, thermal imaging images of the bottled liquid to be detected at multiple moments are acquired using a thermal imaging camera. S3. Obtain convection velocity field information corresponding to each thermal imaging image based on the thermal imaging images; S4. Predict the parameters of the bottled liquid to be tested based on all convection velocity field information.

[0020] This application provides a method for detecting bottled liquids. By actively inducing convection and using a thermal imaging camera to capture dynamic thermal images during the convection process, the method obtains convection velocity field information that is directly related to the physical and chemical properties of the liquid. In other words, this application is equivalent to introducing dynamic convection data when predicting the parameters of bottled liquids. Therefore, this application can effectively solve the problem that the predicted parameters cannot accurately reflect the actual parameters of the bottled liquid because the influence of dynamic convection data on the internal temperature distribution of the bottled liquid cannot be perceived, resulting in low accuracy and reliability of the predicted parameters of bottled liquids.

[0021] As can be seen from the above, the bottled liquid detection system and method provided in this application obtains convection velocity field information directly related to the physical and chemical properties of the liquid by actively inducing convection and using a thermal imaging camera to capture dynamic thermal imaging images during the convection process. In other words, this application is equivalent to introducing dynamic convection data when predicting the parameters of bottled liquids. Therefore, this application can effectively solve the problem that the predicted parameters cannot accurately reflect the actual parameters of the bottled liquid due to the inability to perceive the influence of dynamic convection data on the internal temperature distribution of the bottled liquid, resulting in low accuracy and reliability of the predicted parameters of bottled liquids. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a bottled liquid detection system provided in an embodiment of this application.

[0023] Figure 2 This is a flowchart of a method for detecting bottled liquids provided in an embodiment of this application.

[0024] Reference numerals: 1. Stage; 2. Bottled liquid to be tested; 3. Rotating assembly; 4. Heating assembly; 5. Cooling assembly; 6. Thermal imaging camera; 7. Controller. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] Firstly, such as Figure 1 As shown, this application provides a bottled liquid detection system, which includes: Stage 1, used to place the bottled liquid 2 to be tested; Rotating component 3 is connected to stage 1; Heating component 4 is disposed on the outer periphery of the bottled liquid 2 to be tested; The cooling component 5 is located on the outer periphery of the bottled liquid 2 to be tested, and its setting height is different from that of the heating component 4; A thermal imaging camera 6 is mounted on one side of the stage 1; The controller 7 is used to induce controlled convection in the bottled liquid 2 to be tested placed on the stage 1, so as to generate convection within the bottled liquid 2. The controlled convection induction includes heating the bottled liquid 2 to be tested using the heating component 4, cooling the bottled liquid 2 to be tested using the cooling component 5, and rotating the bottled liquid 2 to be tested using the rotating component 3. It is also used to acquire thermal imaging images of the bottled liquid 2 to be tested at multiple moments during the controlled convection induction process using the thermal imaging camera 6. It is also used to acquire convection velocity field information corresponding to each thermal imaging image based on the thermal imaging images. It is also used to predict the parameter indicators of the bottled liquid 2 to be tested based on all the convection velocity field information.

[0028] This application aims to effectively solve the problem of low accuracy in non-destructive testing of bottled liquids in the prior art by integrating automated convection induction, multi-time thermal imaging image acquisition, and parameter prediction based on convection velocity field information, thereby improving the accuracy and reliability of bottled liquid parameter prediction. The bottled liquid 2 to be tested in this embodiment is a bottled liquid for which parameter prediction is required. The bottled liquid detection system provided in this embodiment can generate observable convection phenomena inside the bottled liquid without damaging its appearance and packaging. Specifically, the controlled convection induction in this embodiment refers to the precise control of the operation of the heating component 4, the cooling component 5, and the rotating component 3 to form a stable and obvious convection pattern inside the bottled liquid. As can be seen from the Navier-Stokes equations, the velocity field of a fluid is correlated with the parameters of the fluid. Therefore, this convection pattern is closely related to the parameters of the bottled liquid 2 to be tested. That is, the dynamic characteristics of the convection velocity field information are related to the physicochemical properties of the bottled liquid to be tested. This means that liquids with different densities, viscosities, thermal conductivity, and specific heat capacities will exhibit different convection patterns and dynamic changes in the velocity field under the same controlled convection induction conditions. Therefore, this embodiment can predict the parameter indicators of the bottled liquid 2 to be tested based on all convection temperature field information by first obtaining the dynamic change of the convection velocity field with respect to time based on all convection velocity field information, and then making parameter predictions based on this dynamic change. The thermal imaging camera 6 in this embodiment is a device capable of capturing infrared radiation from the surface of an object and converting it into a visual thermal imaging image. This embodiment can obtain the temperature distribution information inside the bottled liquid 2 by analyzing these thermal imaging images. The convective velocity field information in this embodiment refers to a vector field describing the velocity and direction of fluid movement inside the bottled liquid. Since the optical flow equation and the thermal convection equation are similar in form—that is, the temperature field obtained based on the thermal imaging image is equivalent to the liquid convective velocity field—this embodiment can obtain the convective velocity field information corresponding to each thermal imaging image by generating a temperature field based on the thermal imaging image using an existing optical flow algorithm and using this temperature field as the convective velocity field. Preferably, taking the calculation of the convective velocity field information corresponding to a certain thermal imaging image as an example, the specific process is as follows: calculate the velocity vector of each pixel in the thermal imaging image, and then integrate all velocity vectors into the convective velocity field information. The formula for calculating the velocity vector is: Where I(x,y) represents the pixel value of the pixel at coordinates (x,y) in the current thermal image, and t represents time. This represents the image gradient of the pixel at coordinates (x, y) in the current thermal image, and v represents the velocity vector of the pixel at coordinates (x, y) in the current thermal image. This represents the difference between the current thermal imaging image and the thermal imaging images at adjacent time points. Since the pixel value of a pixel remains unchanged within the infinitesimally variable dt, that is... =0, while the image gradient and difference value can be obtained through image processing algorithms. Therefore, this embodiment can obtain the velocity vector by substituting the image gradient and difference value into the formula. The set of all velocity vectors corresponding to the current thermal imaging image is the convective velocity field information of the current thermal imaging image. The parameter indicators in this embodiment refer to the key physicochemical properties of the bottled liquid 2 to be tested, such as density, viscosity, and taste indicators. These indicators are crucial for evaluating the quality and storage status of the bottled liquid.

[0029] In this embodiment, the stage 1 is used to stably place the bottled liquid 2 to be tested. The stage 1 can be a flat circular platform with a non-slip surface to ensure that the bottled liquid does not shift during the testing process. The rotating assembly 3 is connected to the stage 1 and is used to drive the bottled liquid 2 through the stage 1. The rotating assembly 3 can consist of a motor and a transmission mechanism. The heating assembly 4 is disposed around the outer periphery of the bottled liquid 2 to heat it. The heating assembly 4 can be an existing resistance wire heater or a PTC heater. Preferably, the heating assembly 4 is arranged around the outer wall of the bottle to ensure uniform heating of the bottled liquid 2. In this embodiment, the cooling component 5 is disposed on the outer periphery of the bottled liquid 2 to be tested. This cooling component 5 is used to cool the bottled liquid 2. Since the height of the cooling component 5 differs from that of the heating component 4 (the cooling component 5 is preferably located above the heating component 4), this embodiment effectively utilizes the cooling component 5 and the heating component 4 to create a significant temperature gradient within the bottled liquid 2, causing spontaneous flow of the liquid within the bottled liquid 2, thereby generating convection within the bottled liquid 2. The cooling component 5 can be an existing semiconductor cooling chip or a circulating water cooling device. In this embodiment, the thermal imaging camera 6 is used to acquire thermal images of the bottled liquid at multiple moments during the controlled convection induction process. The thermal imaging camera 6 can be an existing infrared thermal imager. In this embodiment, the controller 7 can be an embedded system or an industrial PC, running specialized control software and image processing algorithms. The controller 7 can precisely adjust the heating power of the heating component 4, the cooling power of the cooling component 5, and the rotation speed of the rotating component 3 through a PID control algorithm to achieve precise control of the convection.

[0030] The bottled liquid detection system of this application places the bottled liquid 2 to be detected on a stage 1, and uses a rotating component 3, a heating component 4, and a cooling component 5 to induce controlled convection in the bottled liquid. Specifically, the heating component 4 and the cooling component 5 work together at different heights around the bottled liquid to create a temperature gradient inside the liquid, thereby inducing thermal convection. Simultaneously, the rotating component 3 rotates the bottled liquid, further enhancing and stabilizing the convection pattern. During the convection induction process, a thermal imaging camera 6 continuously acquires thermal images of the bottled liquid at multiple moments, recording the dynamic changes in the internal temperature field of the liquid. A controller 7 receives these thermal images and processes them to obtain the convection velocity field information corresponding to each thermal image. This convection velocity field information reflects the motion state of the fluid inside the bottled liquid and is closely related to parameters such as the liquid's density and viscosity. Finally, based on all the acquired convection velocity field information, the controller 7 predicts the parameters of the bottled liquid using a preset prediction model or algorithm.

[0031] This application provides a bottled liquid detection system that acquires convection velocity field information directly related to the physicochemical properties of the liquid by actively inducing convection and using a thermal imaging camera to capture dynamic thermal images of the convection process. In essence, this application introduces dynamic convection data into the prediction of bottled liquid parameters. Therefore, this application effectively solves the problem of low accuracy and reliability in predicting bottled liquid parameters because the influence of dynamic convection data on the internal temperature distribution of the bottled liquid cannot be perceived, resulting in inaccurate predictions of the actual parameters. Thus, this application not only avoids damage to the appearance and packaging of bottled liquids caused by opening the bottle for sampling, but also significantly improves the accuracy and reliability of bottled liquid parameter prediction. In other words, this application provides a high-precision and non-destructive solution for the quality control and evaluation of bottled liquids.

[0032] In some preferred embodiments, the step of acquiring the convection velocity field information corresponding to each thermal imaging image includes: A1. Based on the preset division rules, all thermal imaging images are divided into regions so that all thermal imaging images are evenly divided into multiple test regions. A2. For each test area, obtain the velocity vector of the test area based on the gradient of all pixels in the test area and the difference between the center pixel and the non-center pixel in the test area. A3. For each thermal imaging image, integrate all the corresponding velocity vectors to obtain the convective velocity field information corresponding to that thermal imaging image.

[0033] The preset division rule in step A1 can be set according to the physical characteristics of the bottled liquid 2 to be detected or the image resolution. For example, the preset division rule can be to divide the thermal imaging image into regular grid-like regions or to make adaptive division according to the temperature gradient distribution. Preferably, the preset division rule indicates that the thermal imaging image is divided into a test area by every 3×3 pixels. Each test area is a local subset of the thermal imaging image, and the pixels inside it will be used to calculate local velocity information (velocity vector of the test area). Step A2 is the core of obtaining the local velocity vector. Specifically, for each region to be measured, the gradient information of all pixels in the region is first obtained, which reflects the rate and direction of temperature change in space. At the same time, the difference between the center pixel and the non-center pixel in the region to be measured is obtained, which reflects the relative change of the temperature field in the local region. By combining these spatial gradient information and relative difference values, the local motion trend and intensity of the fluid in the region to be measured (the velocity vector of the region to be measured) can be derived using the principles of optical flow algorithms in fluid mechanics or image processing. Preferably, the formula used in this embodiment to calculate the velocity vector of the region to be measured is the same as the formula used in the above embodiment to calculate the velocity vector of the pixel. Specifically, when calculating the velocity vector of the region to be measured, this embodiment can use the average value of the gradient of all pixels in the region to be measured to replace the image gradient of the pixel with coordinates (x,y) in the current thermal imaging image of the above embodiment. This embodiment uses the difference between the center pixel and the non-center pixel in the region to be measured to replace the difference between the current thermal imaging image and the thermal imaging image at adjacent time points in the above embodiment. In step A3, for each thermal imaging image, the velocity vectors of all the areas to be measured are integrated. This integration can be achieved by vector superposition, interpolation, or weighted averaging to construct the convective velocity field information of the entire bottled liquid 2 to be tested represented by the thermal imaging image. The integrated convective velocity field information is a vector field, in which each point contains a velocity vector, indicating the direction and magnitude of the fluid movement at that location, thus comprehensively reflecting the convective state inside the bottled liquid. This embodiment treats the velocity vectors of all pixels within the same test area as consistent by dividing the thermal imaging image into regions and calculating the velocity vectors of each test area. Essentially, this embodiment performs neighborhood pooling and downsampling on the thermal imaging image, using the velocity vector of the test area instead of calculating the velocity vectors of individual pixels. This reduces the amount of data processing required to acquire convective velocity field information and improves the efficiency of acquiring such information. It should be understood that those skilled in the art can adjust the preset division rules according to actual needs to avoid situations where the calculated velocity vector cannot accurately reflect the velocity vectors of individual pixels within the test area due to an excessively large test area. In other words, this embodiment improves the calculation efficiency of the velocity vector while ensuring its accuracy.

[0034] In some preferred embodiments, step A1 includes: A11. Perform edge detection on all thermal imaging images to obtain the bottle outline of the bottled liquid 2 to be detected; A12. Determine the internal liquid region of the bottled liquid 2 to be detected in all thermal imaging images based on the bottle outline and bottle thickness information. A13. Divide the internal liquid region into regions based on preset division rules so that all thermal imaging images are divided into multiple regions to be tested.

[0035] Step A11 accurately identifies the bottle outline of the bottled liquid 2 to be detected by performing edge detection on all thermal imaging images. This embodiment can use various mature image processing algorithms (such as Canny operator, Sobel operator, or Prewitt operator) to achieve edge detection. These image processing algorithms identify boundaries in the image by analyzing the grayscale or temperature gradient changes of image pixels. Step A12 determines the internal liquid region of the bottled liquid 2 to be detected in all thermal imaging images based on the acquired bottle outline and pre-set bottle thickness information. Specifically, the bottle thickness information can be obtained based on the bottle's design parameters or through actual measurement. Step A12 can accurately define the internal liquid region of the bottle by shrinking the bottle outline inward by a distance related to the bottle thickness, ensuring that subsequent analysis focuses only on the actual liquid part, thereby eliminating interference from the bottle wall and external space. This embodiment first performs edge detection on the thermal imaging image to obtain the bottle outline, then combines the bottle thickness information to determine the internal liquid region, and finally divides only the internal liquid region to ensure the accuracy of the region division and avoid including non-liquid regions such as the bottle or the external environment in the analysis scope. In other words, this embodiment can effectively solve the problem that directly dividing the entire thermal imaging image may introduce interference from non-liquid regions, so that the subsequent acquisition of convection velocity field information can focus more on the convection activity of the liquid itself, thereby effectively improving the accuracy of the acquisition of convection velocity field information and making the calculation results of the convection velocity field more realistically reflect the convection state inside the bottled liquid, thus effectively improving the accuracy and reliability of parameter prediction.

[0036] In some preferred embodiments, step A13 includes: A131. Divide the internal liquid region into regions based on preset division rules so that all thermal imaging images are divided into multiple initial division regions. A132. For each initially divided region, obtain its internal gradient information of temperature distribution, dynamic boundary change information, and motion trajectory continuity information in different thermal imaging images; A133. For each initially defined region, analyze whether the initially defined region is a non-convective region based on the corresponding internal gradient information of temperature distribution, dynamic change information of boundary, and continuity information of motion trajectory. If it is, remove the initially defined region; otherwise, take the initially defined region as the region to be measured. Non-convective regions include bubble regions and sediment regions.

[0037] For each initially segmented region, step A132 needs to acquire various information about it in different thermal imaging images. Specifically, the internal gradient information of the temperature distribution refers to the degree and direction of temperature change within the region. In this embodiment, the internal gradient information of the temperature distribution can be acquired by calculating the first or second derivative of the temperature of the pixels within the region. The dynamic change information of the boundary refers to the characteristics of the boundary of the region changing over time, such as the contraction, expansion, or shape change of the boundary. In this embodiment, the dynamic change information of the boundary can be acquired by comparing the boundary of the region in thermal imaging images at different times. The motion trajectory continuity information refers to whether the motion path of the region in a continuous sequence of thermal imaging images is smooth and continuous. In this embodiment, the motion trajectory continuity information can be acquired by analyzing it using techniques such as optical flow or feature point tracking. Step A133 is crucial for identifying and excluding non-convective regions, which include bubble regions and sediment regions. Since non-convective regions typically exhibit different thermal and kinematic characteristics compared to normal liquid convection regions—specifically, bubble regions usually show a large internal temperature gradient (due to the gas-liquid interface), irregular boundary dynamics (bubble bursting and merging), and discontinuous or abrupt motion trajectories; sediment regions typically show a small internal temperature gradient (usually with insignificant temperature changes), stable boundary dynamics (fixed or slowly moving), and discontinuous or stationary motion trajectories—step A133 can determine whether an initially defined region belongs to a non-convective region by comprehensively analyzing the internal temperature gradient information, boundary dynamics information, and motion trajectory continuity information corresponding to the initially defined region. For example, if an initially defined region has an extremely high temperature gradient and drastic boundary changes, and a discontinuous motion trajectory, then the initially defined region is identified as a bubble region; if an initially defined region has an extremely low temperature gradient and stable boundaries, and a nearly stationary motion trajectory, then the initially defined region is identified as a sediment region.

[0038] This embodiment accurately captures the unique thermal and kinematic characteristics of the bubble and sediment regions by acquiring internal gradient information of temperature distribution, dynamic change information of boundaries, and continuity information of motion trajectory within the initially divided regions. Therefore, this embodiment can accurately classify the initially divided regions based on this multi-dimensional information to effectively distinguish between convective and non-convective regions. In other words, this embodiment can ensure that the input data for subsequent calculation of convective velocity field information only includes the liquid portion that actually participates in convection, thereby avoiding noise and errors introduced by non-convective regions, and thus effectively improving the accuracy and reliability of convective velocity field information acquisition and parameter prediction.

[0039] In some preferred embodiments, a specific example is given below. Suppose that during the inspection of a bottled beverage, thermal imaging camera 6 acquires a series of thermal images. First, according to steps A11 and A12 above, the internal liquid region of the bottled beverage is determined. Next, in step A131, this internal liquid region is divided into multiple initial regions, each comprising 3×3 pixels. Subsequently, in step A132, for each initial region, the system calculates its average temperature gradient in the continuous thermal images, the displacement variance of the boundary pixels, and the continuity information of the motion trajectory of the region's center point. For example, if the average temperature gradient of an initial region is much higher than the surrounding region, the boundary displacement variance is large, and the motion trajectory changes drastically in a short time, then in step A133, this region may be identified as a bubble region and removed. Conversely, if the average temperature gradient of an initial region is extremely low, the boundary displacement variance is close to zero, and the motion vector is almost zero or very slow and continuous, then this region may be identified as a sediment region and removed. Finally, the initially defined regions that were determined to be non-convective areas are removed, and the remaining initially defined regions are identified as the regions to be measured. In this way, it is ensured that the calculation of convective velocity field information is based on pure liquid convection regions, thereby improving the accuracy of the detection results.

[0040] In some preferred embodiments, step A2 includes: A21. For each test area, obtain the spatial temperature gradient information of each pixel in the test area, and obtain the temporal temperature difference information between the non-center pixel and the center pixel in the test area. A22. For each test area, analyze whether the temperature field distribution of the test area is uniform based on the spatial temperature gradient information and the temporal temperature difference information. If so, obtain the velocity vector of the test area based on the gradient of all pixels in the test area and the difference between the center pixel and the non-center pixel in the test area. If not, proceed to step A23. The condition for judging the uniformity of the temperature field distribution is: the spatial temperature gradient information is less than the preset temperature gradient and the temporal temperature difference information is less than the preset temperature difference. A23. Divide the test area with uneven temperature field distribution into multiple local sub-regions; A24. Obtain the local velocity vector corresponding to each local sub-region based on the gradient of all pixels in the local sub-region and the difference between the center pixel and non-center pixels in the local sub-region. Integrate all local velocity vectors corresponding to the same test region to obtain the velocity vector corresponding to the test region.

[0041] Obtaining the spatial temperature gradient information of each pixel within the test area refers to characterizing the spatial distribution of temperature by calculating the rate of temperature change between adjacent pixels within the test area. This can be achieved using image processing algorithms such as the Sobel operator, Prewitt operator, or Roberts operator. These operators calculate the gradient of the image through convolution operations, thereby obtaining the spatial temperature gradient information. Obtaining the temporal temperature difference information between non-center and center pixels within the test area refers to reflecting the dynamic characteristics of temperature change over time by comparing the temperature values ​​of non-center pixels at different times with the temperature value of the center pixel at the corresponding time. This embodiment can obtain the temporal temperature difference information by directly subtracting the temperature values ​​of the non-center pixels from those of the center pixel. For example, for a non-center pixel P(x,y) and a center pixel C, the temporal temperature difference can be expressed as ΔT(x,y,t) = T(x,y,t) - T(C,t). This embodiment analyzes the uniformity of the temperature field distribution in the test area by comparing spatial temperature gradient information with a preset temperature gradient and temporal temperature difference information with a preset temperature difference. Specifically, if the spatial temperature gradient information is less than the preset temperature gradient and the temporal temperature difference information is less than the preset temperature difference, the temperature field distribution is considered uniform. It should be understood that the preset temperature gradient and preset temperature difference can be set based on empirical values, experimental data, or the needs of specific application scenarios. When the analysis indicates a non-uniform temperature field distribution, this embodiment can use methods such as quadtree decomposition, K-means clustering, or image segmentation algorithms (such as the watershed algorithm) to divide the test area with a non-uniform temperature field distribution into multiple local sub-regions. For example, the division boundary can be determined based on the local maximum value of the temperature gradient or the local non-uniformity of the temperature difference to decompose a large non-uniform region into multiple relatively uniform small regions. The specific process for obtaining the local velocity vector corresponding to each local sub-region in this embodiment is preferably the same as the specific process for obtaining the velocity vector of the test area in the above embodiment, and will not be discussed in detail here. This embodiment can use weighted averaging, interpolation, or fusion methods based on physical models (such as the Navier-Stokes equations) to integrate all local velocity vectors corresponding to the same test area to obtain the velocity vector corresponding to the test area. For example, different weights can be assigned to different local velocity vectors according to the area of ​​each local sub-region or the degree of uniformity of its temperature field, and then weighted averaging or interpolation algorithms can be used to smoothly connect the local velocity vectors to form the velocity vector field of the entire test area.This embodiment is equivalent to effectively improving the accuracy of velocity vector acquisition when the temperature field distribution in the area to be measured is uneven. This is achieved by dividing the area to be measured into multiple local sub-regions, calculating the local velocity vectors separately, and then integrating them. This makes the acquisition of convective velocity field information more accurate and reliable. In other words, this embodiment can further improve the accuracy and reliability of convective velocity field information.

[0042] In some preferred embodiments, the step of acquiring the convective velocity field information corresponding to each thermal imaging image further includes a step performed before step A1: A4. Obtain the liquid type information of the bottled liquid to be tested 2; A5. Determine the preset division rules based on the liquid type information.

[0043] Step A4 can obtain liquid type information by scanning the barcode or QR code on the bottled liquid 2 to query relevant parameter information of the bottled liquid 2. Step A5, determining the preset division rule based on the liquid type information, means that after obtaining the liquid type information, the system will dynamically select or generate the most suitable region division rule for the current liquid type based on the information. For example, a database or lookup table containing multiple liquid types and corresponding division rules can be stored in the controller 7 in advance. When a certain liquid type information is obtained, the controller 7 retrieves the corresponding preset division rule from the database. The preset division rule in this embodiment may include, but is not limited to, the size, shape, division density of the region, and specific physical boundary conditions that need to be considered during the division process.

[0044] This embodiment achieves dynamic determination of preset division rules based on specific liquid types by acquiring the liquid type information of the bottled liquid 2 to be detected before dividing the region, and then dynamically determining preset division rules based on this information. This allows the preset division rules to better adapt to the physical properties and thermal convection behavior of different liquids. For example, for liquids with high viscosity, more detailed region division may be required to capture local slow convection; while for liquids with low viscosity, a coarser division may be used to focus on the overall flow trend. Therefore, this embodiment can effectively avoid errors caused by the mismatch between preset division rules and liquid types, thereby further improving the accuracy and reliability of convection velocity field information acquisition, and thus improving the robustness and applicability of the bottled liquid detection system.

[0045] In some preferred embodiments, a specific example is given below. Assume the bottled liquid 2 to be detected is mineral water or cooking oil. During step A4, the controller 7 scans the product label barcode on the bottled liquid 2 to identify it as mineral water. Subsequently, during step A5, the controller 7 retrieves a region division rule specifically for water-based liquids from a preset database based on the liquid type information of "mineral water." For example, this rule stipulates that because water has a lower viscosity and relatively faster convection velocity, a relatively uniform and slightly larger grid can be used for region division in the thermal imaging image to capture its overall flow field characteristics. If the bottled liquid 2 to be detected is identified as "cooking oil," the controller 7 determines another set of division rules based on the liquid type information of "cooking oil." Specifically, because the viscosity of cooking oil is usually higher than that of water, its convection may be slower and more localized. Therefore, the preset division rule for cooking oil uses a finer grid for region division to capture local convection details more precisely. In this way, the system can flexibly adjust the region division strategy according to the characteristics of different liquids, ensuring that the most accurate convection velocity field information can be obtained regardless of the type of liquid being detected.

[0046] In some preferred embodiments, the step of predicting the parameter parameters of the bottled liquid 2 to be detected based on all convection velocity field information includes: B1. Obtain the brand and formula information of the bottled liquid to be tested 2; B2. Determine parameter prediction strategies based on brand and formula information; B3. Utilize parameter prediction strategies to predict the parameter indicators of the bottled liquid 2 to be tested based on all convective velocity field information.

[0047] Obtaining the brand and formula information of the bottled liquid 2 to be tested can be understood as acquiring specific identification and composition data of the bottled liquid through various methods. For example, the brand and product model can be automatically identified by scanning the barcode or QR code on the bottle, and then the corresponding formula information can be retrieved from a preset database; or, the operator can manually enter the brand and formula details of the bottled liquid. Specifically, brand information refers to the identification of the manufacturer or product series, while formula information refers to the specific composition of the liquid product, production process parameters, or description of its physicochemical properties. Determining a parameter prediction strategy based on the acquired brand and formula information refers to selecting or generating a model or algorithm best suited for predicting parameters of the bottled liquid based on this specific data. For example, the response characteristics of parameters such as density and viscosity may differ for liquids of different brands or formulas when heated or cooled. Specifically, this embodiment can pre-establish a strategy library containing multiple prediction models or parameter adjustment rules. Once a specific brand and formula are identified, the system can automatically match and call the corresponding prediction strategy. This parameter prediction strategy can be a specific machine learning model (such as a neural network, support vector machine, etc.) or a set of weight coefficients or calibration curves optimized for the liquid's characteristics. Since this embodiment uses a parameter prediction strategy determined based on brand and formula information to predict the parameter indicators of the bottled liquid 2 to be tested based on all convection velocity field information, this means that after obtaining the convection velocity field information generated by the bottled liquid 2 to be tested during controlled convection induction, a general model is no longer used for prediction. Instead, this velocity field information is input into a prediction strategy optimized for the specific brand and formula. Therefore, this embodiment can obtain more accurate parameter indicator prediction results. Since liquids of different brands and formulations have unique physicochemical properties and thermodynamic responses, these characteristics directly affect the correlation between the convection patterns formed under controlled convection and the final parameter indicators. Therefore, this embodiment can select or generate a highly matched prediction model or algorithm for each liquid by introducing the brand information and formulation information of the bottled liquid 2 to be tested, and determining a customized parameter prediction strategy accordingly. This effectively solves the problem of insufficient accuracy of general prediction models when dealing with diverse liquid products, thereby effectively improving the accuracy and reliability of parameter indicator prediction.

[0048] In some preferred embodiments, a specific example is given below. Suppose we need to detect two different brands of milk, such as pure milk from brand A and low-fat milk from brand B. After inducing controlled convection and acquiring its convection velocity field information, the system first obtains the brand and formula information of the two types of milk. For pure milk from brand A, the system selects a parameter prediction model specifically optimized for high-fat milk from a pre-defined strategy library based on its brand and formula (e.g., known to have a high fat content and relatively high viscosity). This model may have used a large amount of data from pure milk from brand A during training and considered its unique thermophysical properties. For low-fat milk from brand B, the system selects another parameter prediction model optimized for low-fat milk, which may focus more on capturing subtle changes in low-viscosity liquids during convection. In this way, even if the convection velocity field information of the two types of milk is similar in some aspects, the final predicted density, viscosity, or taste indicators will more accurately reflect the true characteristics of each product due to the use of targeted prediction strategies, thus avoiding prediction bias that may be caused by using a single model.

[0049] In some preferred embodiments, the step of inducing controlled convection within the bottled liquid 2 to be tested, placed on the stage 1, to generate convection within the bottled liquid 2 includes: C1. Obtain the actual heating temperature of the heating component 4, the actual cooling temperature of the cooling component 5, and the actual rotation speed of the platform 1; C2. Based on the first PID control algorithm, the heating power of the heating component 4 is adjusted according to the actual heating temperature and the preset target heating temperature. Based on the second PID control algorithm, the cooling power of the cooling component 5 is adjusted according to the actual cooling temperature and the preset target cooling temperature. Based on the third PID control algorithm, the operating parameters of the rotating component 3 are adjusted according to the actual rotation speed and the preset target rotation speed, so as to induce controlled convection in the bottled liquid 2 to be tested placed on the stage 1 and generate convection within the bottled liquid 2 to be tested.

[0050] The actual heating temperature in this embodiment refers to the current temperature value of the heating component 4 as monitored in real time by a temperature sensor located near the heating component 4. The actual cooling temperature in this embodiment refers to the current temperature value of the cooling component 5 as monitored in real time by a temperature sensor located near the cooling component 5. The actual rotation speed in this embodiment refers to the current rotation speed of the stage 1 as detected in real time by a speed sensor or encoder located on the rotation component 3. The acquisition of these actual parameters is intended to provide real-time feedback data for subsequent control algorithms. The first, second, and third PID control algorithms (proportional-integral-derivative control algorithms) of this embodiment are feedback control methods widely used in industrial control. They calculate the error between the setpoint and the actual value and adjust the control quantity based on the weighted sum of the proportional, integral, and derivative parameters. Specifically, the first PID control algorithm dynamically adjusts the heating power of the heating component 4 based on the deviation between the actual heating temperature and the preset target heating temperature, so that the temperature of the heating component 4 approaches and stabilizes at the target heating temperature. The second PID control algorithm dynamically adjusts the cooling power of the cooling component 5 based on the deviation between the actual cooling temperature and the preset target cooling temperature, so that the temperature of the cooling component 5 approaches and stabilizes at the target cooling temperature. The third PID control algorithm dynamically adjusts the operating parameters (e.g., motor drive voltage or current) of the rotating component 3 based on the deviation between the actual rotation speed and the preset target rotation speed, so that the rotation speed of the stage 1 approaches and stabilizes at the target rotation speed. The target heating temperature, target cooling temperature, and target rotation speed of this embodiment are preferably ideal operating parameters preset according to the type of bottled liquid 2 to be detected, the expected convection mode, and the detection requirements. This embodiment achieves precise and dynamic adjustment of the heating component 4, cooling component 5, and rotating component 3 by introducing a PID control algorithm. Since the PID control algorithm can acquire actual operating parameters in real time and make feedback adjustments based on the deviation between the actual operating parameters and the target parameters, this embodiment enables continuous optimization of the heating power, cooling power, and operating parameters of the rotating component 3. This ensures that the heating, cooling, and rotation processes always proceed according to preset conditions, effectively offsetting the influence of external environmental changes and internal system disturbances on the convection induction process and guaranteeing the stability and repeatability of convection within the bottled liquid 2 under test. By precisely controlling the temperature gradient and shear force, a specific and stable convection pattern can be induced, which is crucial for accurately acquiring thermal imaging images and convection velocity field information subsequently.Through the above technical solution, this application can significantly improve the accuracy and stability of the controlled convection induction process. Since the heating, cooling and rotation conditions are precisely controlled at the preset target values, the convection mode generated inside the bottled liquid 2 to be detected will be more stable and predictable. Therefore, this application helps to obtain clearer and more accurate thermal imaging images, thereby making the extraction of convection velocity field information more reliable, and thus effectively improving the accuracy and reliability of predicting the parameters of the bottled liquid.

[0051] In some preferred embodiments, the parameters include density, viscosity, and mouthfeel indicators. Density refers to the mass of a substance per unit volume and is a fundamental physical property of liquids; its changes may reflect changes in the liquid's composition or state. Viscosity refers to the resistance a fluid exhibits to shear stress, reflecting the ease or difficulty of liquid flow and significantly influencing its processing, storage, and usage performance. Mouthfeel indicators are a comprehensive sensory evaluation, typically involving the liquid's flavor, texture, and body, and are particularly crucial for product quality control in industries such as food and beverage.

[0052] Secondly, such as Figure 2 As shown, this application also provides a method for detecting bottled liquids, which is applied to the bottled liquid detection system provided in the first aspect above. The method includes the following steps: S1. Controlled convection induction is performed on the bottled liquid 2 to be tested placed on the stage 1 to generate convection within the bottled liquid 2. The controlled convection induction includes heating the bottled liquid 2 to be tested using the heating component 4, cooling the bottled liquid 2 to be tested using the cooling component 5, and rotating the bottled liquid 2 to be tested using the rotating component 3. S2. During the controlled convection induction process, thermal imaging images of the bottled liquid 2 to be detected at multiple times are acquired by thermal imaging camera 6. S3. Obtain convection velocity field information corresponding to each thermal imaging image based on the thermal imaging images; S4. Predict the parameters of the bottled liquid 2 to be tested based on all the convection velocity field information.

[0053] The bottled liquid detection method provided in this application is applied to the bottled liquid detection system provided in the first aspect above. The principle of the bottled liquid detection method provided in this embodiment is the same as that of the bottled liquid detection system provided in the first aspect above, and will not be discussed in detail here.

[0054] As can be seen from the above, the bottled liquid detection system and method provided in this application obtains convection velocity field information directly related to the physical and chemical properties of the liquid by actively inducing convection and using a thermal imaging camera to capture dynamic thermal imaging images during the convection process. In other words, this application is equivalent to introducing dynamic convection data when predicting the parameters of bottled liquids. Therefore, this application can effectively solve the problem that the predicted parameters cannot accurately reflect the actual parameters of the bottled liquid due to the inability to perceive the influence of dynamic convection data on the internal temperature distribution of the bottled liquid, resulting in low accuracy and reliability of the predicted parameters of bottled liquids.

[0055] In the embodiments provided in this application, it should be understood that relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0056] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A bottled liquid detection system, characterized in that, The bottled liquid detection system includes: The stage is used to hold the bottled liquid to be tested. A rotating assembly is connected to the stage. A heating element is disposed on the outer periphery of the bottled liquid to be tested; A cooling component is disposed on the outer periphery of the bottled liquid to be tested, and its placement height is different from that of the heating component; A thermal imaging camera is mounted on one side of the stage; The controller is used to induce controlled convection in a bottled liquid to be tested placed on a stage, so as to generate convection within the bottled liquid. The controlled convection induction includes heating the bottled liquid using the heating component, cooling the bottled liquid using the cooling component, and rotating the bottled liquid using the rotating component. The controller is also used to acquire thermal imaging images of the bottled liquid at multiple moments during the controlled convection induction process using a thermal imaging camera, to acquire convection velocity field information corresponding to each thermal imaging image, and to predict the parameter indicators of the bottled liquid based on all the convection velocity field information.

2. The bottled liquid detection system according to claim 1, characterized in that, The step of acquiring the convective velocity field information corresponding to each of the thermal imaging images includes: A1. Based on a preset division rule, all the thermal imaging images are divided into regions so that all the thermal imaging images are divided into multiple regions to be tested. A2. For each region to be tested, obtain the velocity vector of the region to be tested based on the gradient of all pixels in the region to be tested and the difference between the center pixel and the non-center pixel in the region to be tested. A3. For each of the thermal imaging images, integrate all the corresponding velocity vectors to obtain the convective velocity field information corresponding to the thermal imaging image.

3. The bottled liquid detection system according to claim 2, characterized in that, Step A1 includes: A11. Perform edge detection on all the thermal imaging images to obtain the bottle outline of the bottled liquid to be detected; A12. Determine the internal liquid region of the bottled liquid to be detected in all the thermal imaging images based on the bottle outline and bottle thickness information; A13. The internal liquid region is divided into regions based on a preset division rule, so that all the thermal imaging images are divided into multiple regions to be tested.

4. The bottled liquid detection system according to claim 3, characterized in that, Step A13 includes: A131. The internal liquid region is divided into regions based on a preset division rule, so that all the thermal imaging images are divided into multiple initial division regions. A132. For each of the initial division regions, obtain its internal gradient information of temperature distribution, dynamic boundary change information, and motion trajectory continuity information in different thermal imaging images; A133. For each of the initial division regions, analyze whether the initial division region is a non-convective region based on the corresponding internal gradient information of temperature distribution, boundary dynamic change information and motion trajectory continuity information. If it is, remove the initial division region; otherwise, take the initial division region as the region to be measured. The non-convective region includes bubble region and sediment region.

5. The bottled liquid detection system according to claim 2, characterized in that, Step A2 includes: A21. For each of the regions to be tested, obtain the spatial temperature gradient information of each pixel in the region to be tested, and obtain the temporal temperature difference information between the non-center pixel and the center pixel in the region to be tested. A22. For each of the regions to be tested, analyze whether the temperature field distribution of the region to be tested is uniform based on the spatial temperature gradient information and the temporal temperature difference information. If so, obtain the velocity vector of the region to be tested based on the gradient of all pixels in the region to be tested and the difference between the center pixel and the non-center pixel in the region to be tested. If not, proceed to step A23. The condition for determining whether the temperature field distribution is uniform is: the spatial temperature gradient information is less than the preset temperature gradient and the temporal temperature difference information is less than the preset temperature difference. A23. Divide the test area with uneven temperature field distribution into multiple local sub-regions; A24. Obtain the local velocity vector corresponding to each local sub-region based on the gradient of all pixels in the local sub-region and the difference between the center pixel and non-center pixel in the local sub-region. Integrate all the local velocity vectors corresponding to the same test region to obtain the velocity vector corresponding to the test region.

6. The bottled liquid detection system according to claim 2, characterized in that, The step of acquiring the convective velocity field information corresponding to each of the thermal imaging images further includes steps performed before step A1: A4. Obtain the liquid type information of the bottled liquid to be detected; A5. Determine the preset division rules based on the liquid type information.

7. The bottled liquid detection system according to claim 1, characterized in that, The step of predicting the parameter indicators of the bottled liquid to be detected based on all the convective velocity field information includes: B1. Obtain the brand information and formula information of the bottled liquid to be tested; B2. Determine the parameter prediction strategy based on the brand information and the formula information; B3. Using the parameter prediction strategy, predict the parameter indicators of the bottled liquid to be detected based on all the convection velocity field information.

8. The bottled liquid detection system according to claim 1, characterized in that, The step of inducing controlled convection in the bottled liquid to be tested placed on the stage to generate convection within the bottled liquid includes: C1. Obtain the actual heating temperature of the heating component, the actual cooling temperature of the cooling component, and the actual rotation speed of the platform; C2. Based on the first PID control algorithm, the heating power of the heating component is adjusted according to the actual heating temperature and the preset target heating temperature. Based on the second PID control algorithm, the cooling power of the cooling component is adjusted according to the actual cooling temperature and the preset target cooling temperature. Based on the third PID control algorithm, the operating parameters of the rotating component are adjusted according to the actual rotation speed and the preset target rotation speed, so as to induce controlled convection and generate convection within the bottled liquid to be tested placed on the stage.

9. The bottled liquid detection system according to claim 1, characterized in that, The parameters include density, viscosity, and mouthfeel.

10. A method for detecting bottled liquids, characterized in that, The bottled liquid detection method is applied in the bottled liquid detection system as described in any one of claims 1-9, and the bottled liquid detection method includes the following steps: S1. Controlled convection induction is performed on the bottled liquid to be tested placed on the stage to generate convection within the bottled liquid to be tested. The controlled convection induction includes heating the bottled liquid to be tested using the heating component, cooling the bottled liquid to be tested using the cooling component, and rotating the bottled liquid to be tested using the rotating component. S2. During the controlled convection induction process, thermal imaging images of the bottled liquid to be detected at multiple moments are acquired by the thermal imaging camera. S3. Obtain convection velocity field information corresponding to each of the thermal imaging images based on the thermal imaging images; S4. Predict the parameter indicators of the bottled liquid to be tested based on all the convection velocity field information.

Images