Cake rolling machine cake thickness control system based on visual inspection

Abstract

The invention provides a cake rolling machine cake thickness control system based on visual inspection, and relates to the technical field of machine vision, and the system comprises a machine vision image obtaining module which is used for obtaining a machine vision image of a cake product when the cake rolling machine rolls a new cake product; the current cake thickness condition determination module is used for detecting the current cake thickness condition based on the machine vision image; the future rolling control strategy correction module is used for correcting the current future rolling control strategy of the cake rolling machine based on the current cake thickness condition; and the relay rolling control module is used for performing corresponding relay rolling control on the cake rolling machine based on the corrected future rolling control strategy. The current cake thickness condition is detected on the basis of the machine learning technology, the current future rolling control strategy of the cake rolling machine is corrected on the basis of the current cake thickness condition, corresponding relay rolling control is conducted on the cake rolling machine on the basis of the corrected future rolling control strategy, real-time accurate control over the cake product thickness is achieved, and the cake product quality is improved.

Description

Technical Field

[0001] The invention relates to the technical field of machine vision, and in particular to a pancake thickness control system of a pancake rolling machine based on visual detection. Background Art

[0002] In traditional pancake rolling machines, the thickness of the pancakes usually relies on manual adjustment or simple mechanical settings, lacking real-time and precise control. However, the thickness of the pancakes is directly related to their taste, quality, and production efficiency, and traditional pancake rolling machines are often unable to dynamically adjust the actual thickness during each rolling process, which may lead to uneven thickness, thus affecting product quality and even causing waste. Therefore, achieving precise pancake thickness control in the automated production process has become a core problem that needs to be solved in the field of food machinery.

[0003] With the continuous development of science and technology, machine vision technology has provided a new solution to this problem. Through cameras and image processing algorithms, machine vision can obtain and analyze the appearance characteristics of the pancake in real time and accurately measure its thickness. Therefore, the thickness control system of the pancake rolling machine based on machine vision has become an urgent need to improve production efficiency and ensure product consistency and quality. Summary of the invention

[0004] One of the purposes of the present invention is to provide a pancake thickness control system for a pancake rolling machine based on visual detection, which detects the current pancake thickness based on machine learning technology, and based on the current pancake thickness, corrects the future rolling control strategy of the pancake rolling machine. Based on the corrected future rolling control strategy, the pancake rolling machine is correspondingly relayed for rolling control. The thickness of the pancake product can be dynamically adjusted during each rolling process, thereby achieving real-time and precise control of the thickness of the pancake product and improving the quality of the pancake product.

[0005] An embodiment of the present invention provides a pancake thickness control system for a pancake rolling machine based on visual detection, comprising:

[0006] A machine vision image acquisition module, used to acquire a machine vision image of a cake product when the cake rolling machine rolls out a new cake product;

[0007] A current cake thickness determination module is used to detect the current cake thickness based on machine vision images;

[0008] The future rolling control strategy correction module is used to correct the future rolling control strategy of the pancake rolling machine based on the current pancake thickness;

[0009] The relay rolling control module is used to perform corresponding relay rolling control on the pancake rolling machine based on the revised future rolling control strategy.

[0010] Optionally, the machine vision image acquisition module acquires a machine vision image of the pie product, including:

[0011] The machine vision image of the pie product is obtained by a machine vision camera.

[0012] Optionally, based on machine vision images, detect the current cake thickness, including:

[0013] Based on image recognition technology, the current cake thickness is detected according to the machine vision image.

[0014] Optionally, the future rolling control strategy correction module corrects the future rolling control strategy of the pancake rolling machine based on the current pancake thickness, including:

[0015] Perform feature description processing on the current pancake thickness and future rolling control strategy to obtain a feature description vector;

[0016] Based on the feature description vector, query the correction rule library to determine the target correction rule;

[0017] Based on the target correction rules, the future rolling control strategy is corrected accordingly.

[0018] Optionally, obtaining a machine vision image of a cake product includes:

[0019] Before the next rolling stroke of the pancake rolling machine begins, dynamically simulate the lighting environment of the pancake product when the pancake rolling machine performs the next rolling stroke to obtain a dynamic simulated lighting environment;

[0020] Based on the target correction rule determined last time, determine the focus of image acquisition;

[0021] Based on the dynamic simulation of the lighting environment and the focus of image acquisition, determine the machine vision camera array positioning control strategy, the fill light device array positioning control strategy and the image acquisition timing;

[0022] Based on the positioning control strategy of the machine vision camera array, the machine vision camera array in the pancake rolling machine is controlled to be positioned accordingly;

[0023] Based on the positioning control strategy of the fill light device array, the positioning control strategy of the fill light device array in the pancake rolling machine is controlled to be positioned accordingly;

[0024] When the rolling process begins, if the image acquisition timing is reached, the machine vision camera array is controlled to acquire the machine vision image of the cake product.

[0025] Optionally, determining the focus of image acquisition based on the target correction rule determined last time includes:

[0026] Analyze the correction object type of the target correction rule determined last time;

[0027] determining the affected part of the pie product corresponding to the type of correction object;

[0028] Focus image acquisition on the affected area.

[0029] Optionally, the determining of the machine vision camera array position control strategy, the fill light device array position control strategy and the image acquisition timing based on the dynamic simulation lighting environment and the image acquisition focus includes:

[0030] Predict the rolling status of the pancake product after the last rolling stroke of the pancake rolling machine is completed;

[0031] Based on the rolling situation, the simulated cake model in the dynamic simulated lighting environment is simulated and configured accordingly;

[0032] Determine a first model region corresponding to the focus of image acquisition and a second model region other than the first model region from the simulation pie model after the situation simulation configuration;

[0033] generating a first image acquisition constraint based on the multimodal parameters of the first model region;

[0034] Control the machine vision camera array model and the fill light device array model in the dynamic simulated lighting environment to perform dynamic simulation of image acquisition, and control to pause the dynamic simulation of image acquisition after reaching the optimal condition; wherein the optimal condition includes: the first image of the first model area captured by the machine vision camera array model under the fill light effect of the fill light device array model meets the image acquisition constraint and the second image of the second model area captured meets the standard second image acquisition constraint;

[0035] Determining a machine vision camera array in-position control strategy based on a first current in-position parameter of a machine vision camera array model in a dynamic simulated lighting environment after controlling to pause the dynamic simulation of image acquisition;

[0036] Determine the fill light device array position control strategy based on a second current position parameter of the fill light device array model in the dynamic simulated lighting environment after controlling the pause to perform the dynamic simulation of image acquisition;

[0037] Based on the control pause for dynamic simulation of image acquisition, the progress of the next rolling stroke of the pancake rolling machine in the dynamic simulation lighting environment is simulated to determine the timing of image acquisition.

[0038] Optional, visual inspection-based cake thickness control system for cake rolling machine, also includes:

[0039] The expected pancake thickness information acquisition module is used to acquire the user's expected pancake thickness information when the user inputs a personalized pancake thickness customization instruction;

[0040] The expected pancake thickness information adding module is used to add the expected pancake thickness information into the future rolling control strategy.

[0041] Optionally, the expected pie thickness information acquisition module acquires the user's expected pie thickness information, including:

[0042] Guide users to approach the pie-thick interactive screen;

[0043] When the user approaches the pie thickness interactive screen, the pie thickness interactive screen is controlled to display the pie thickness control axis and pie thickness control prompt information to the user; wherein the pie thickness control axis is a vertical numerical axis that gradually increases from 0 scale to the maximum pie thickness scale from bottom to top;

[0044] Obtain the landing point trajectory formed by the user's sight point on the pie-thick interactive screen moving as the user's sight line changes;

[0045] Determine the last first target landing point that meets the first landing point condition from the landing point trajectory;

[0046] Determine a first local trajectory after the first target landing point from the landing point trajectory;

[0047] Determine a second target landing point that meets a second landing point condition from the first local trajectory;

[0048] Determine the first target cake thickness scale on the cake thickness control axis corresponding to the perpendicular point of the second target landing point, and associate it with the corresponding second target landing point;

[0049] When the second target landing point is unique, the expected pie thickness information is determined based on the first target pie thickness scale associated with the second target landing point; otherwise, if the second target landing points meet the landing point association condition, the selection auxiliary information corresponding to the first target pie thickness scale associated with the two second target landing points with the farthest distance between them is obtained;

[0050] Control the interactive screen to display selection auxiliary information to the user;

[0051] When the landing point trajectory continues to generate a new second partial trajectory, the last third target landing point that meets the second landing point condition is determined from the second partial trajectory;

[0052] Determine the second target cake thickness scale on the cake thickness control axis corresponding to the position of the perpendicular point of the third target landing point to the cake thickness control axis;

[0053] Based on the second target pie thickness scale, desired pie thickness information is determined.

[0054] Optionally, the first landing point condition includes: a straight line connecting the landing point and the 0 scale position on the pancake thickness control axis is perpendicular to the pancake thickness control axis;

[0055] The second landing point condition includes: the landing point stay time exceeds the time threshold;

[0056] The landing point association condition includes: the straight-line distance between at least two landing points among the landing points is less than or equal to a first distance threshold and the straight-line distance between at least two landing points among the landing points is greater than or equal to a second distance threshold; the first distance threshold is less than the second distance threshold.

[0057] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0058] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0060] Figure 1 It is a schematic diagram of a pancake thickness control system of a pancake rolling machine based on visual detection in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0062] The embodiment of the present invention provides a pancake thickness control system for a pancake rolling machine based on visual detection, such as Figure 1 As shown, including:

[0063] The machine vision image acquisition module 1 is used to acquire the machine vision image of the cake product when the cake rolling machine rolls out the new cake product;

[0064] The current cake thickness condition determination module 2 is used to detect the current cake thickness condition based on the machine vision image;

[0065] The future rolling control strategy correction module 3 is used to correct the future rolling control strategy of the rolling machine based on the current pancake thickness;

[0066] The relay rolling control module 4 is used to perform corresponding relay rolling control on the pancake rolling machine based on the revised future rolling control strategy.

[0067] The working principle and beneficial effects of the above technical solution are:

[0068] A new pancake product refers to a new single / single batch of pancakes that need to undergo a rolling process at the same time; the current pancake thickness situation at least includes: the pancake thickness at different positions on the pancake product; before the pancake rolling machine works, the technicians will set the dough type, rolling target thickness, etc. of the pancake product, and will also formulate a rolling control strategy for the pancake rolling machine based on this, in order to control the pancake rolling machine during rolling. The future rolling control strategy of the pancake rolling machine at that time is the remaining unimplemented part of the rolling control strategy; the current pancake thickness situation will reflect how the pancake product needs to be further rolled to evenly reach the rolling target thickness, and therefore, based on this, the future rolling control strategy of the pancake rolling machine at that time can be corrected, so that the corrected future rolling control strategy is suitable for relaying the two pancake rolling machines to continue the rolling work; finally, based on the corrected future rolling control strategy, the pancake rolling machine is correspondingly relayed for rolling control.

[0069] The present invention detects the current cake thickness based on machine learning technology, and based on the current cake thickness, corrects the future rolling control strategy of the cake rolling machine. Based on the corrected future rolling control strategy, the cake rolling machine is correspondingly relayed for rolling control. The thickness of the cake product can be dynamically adjusted during each rolling process, thereby achieving real-time and precise control of the thickness of the cake product and improving the quality of the cake product.

[0070] In one embodiment, the machine vision image acquisition module acquires a machine vision image of a pie product, including:

[0071] The machine vision image of the pie product is obtained by a machine vision camera.

[0072] Machine vision cameras (such as high-resolution industrial cameras) are installed above and on the side of the production line of the pancake rolling machine to ensure that the surface and sides of the pancake being rolled can be fully captured. By controlling the machine vision camera to shoot during the pancake rolling process, machine vision images of the pancake products can be obtained.

[0073] In one embodiment, based on the machine vision image, detecting the current cake thickness includes:

[0074] Based on image recognition technology, the current cake thickness is detected according to the machine vision image.

[0075] First, the machine vision image is preprocessed, including grayscale, denoising, and contrast enhancement. Next, an edge detection algorithm (such as Canny or Sobel) is used to extract the contour of the preprocessed machine vision image. Finally, the cake thickness at different locations on the cake product is determined based on the contour, that is, the current cake thickness.

[0076] In one embodiment, the future rolling control strategy correction module corrects the future rolling control strategy of the pancake rolling machine based on the current pancake thickness, including:

[0077] Perform feature description processing on the current pancake thickness and future rolling control strategy to obtain a feature description vector;

[0078] Based on the feature description vector, query the correction rule library to determine the target correction rule;

[0079] Based on the target correction rules, the future rolling control strategy is corrected accordingly.

[0080] When performing feature description processing, the features of the current cake thickness and the future rolling control strategy are extracted, including at least: the cake thickness at different positions on the cake product, the cake product dough type, the rolling target thickness, the rolling pressure, the rolling temperature, etc., and then these features are expressed in vector form as feature description vectors; there are correction rules corresponding to different feature description vectors in the correction rule library, and the correction rule library is queried based on the feature description vector to determine the most appropriate target correction rule. For example: the feature expressed as the feature description vector is that the thickness of a certain area on the current cake is 5mm and the rolling target thickness is 3mm, then the corresponding target correction rule is to increase the rolling pressure by 20%. Specifically, the correction rules corresponding to different feature description vectors can be set in advance by technical experts based on the control experience of the cake rolling machine.

[0081] In one embodiment, the step of acquiring a machine vision image of a pie product includes:

[0082] Before the next rolling stroke of the pancake rolling machine begins, dynamically simulate the lighting environment of the pancake product when the pancake rolling machine performs the next rolling stroke to obtain a dynamic simulated lighting environment;

[0083] Based on the target correction rule determined last time, determine the focus of image acquisition;

[0084] Based on the dynamic simulation of the lighting environment and the focus of image acquisition, determine the machine vision camera array positioning control strategy, the fill light device array positioning control strategy and the image acquisition timing;

[0085] Based on the positioning control strategy of the machine vision camera array, the machine vision camera array in the pancake rolling machine is controlled to be positioned accordingly;

[0086] Based on the positioning control strategy of the fill light device array, the positioning control strategy of the fill light device array in the pancake rolling machine is controlled to be positioned accordingly;

[0087] When the rolling process begins, if the image acquisition timing is reached, the machine vision camera array is controlled to acquire the machine vision image of the cake product.

[0088] The working principle and beneficial effects of the above technical solution are:

[0089] Each rolling stroke of the pancake rolling machine includes: the rolling roller leaves the pancake product, the rolling roller is lifted, the rolling roller falls to contact the pancake product, and the rolling roller contacts to roll the pancake. The rolling roller leaves the pancake product and contacts to roll the pancake. It is impossible to collect a comprehensive image of the pancake product. Therefore, it is necessary to obtain the machine vision image of the pancake product in the two processes of the rolling roller lifting and the rolling roller falling to contact the pancake product. However, during the rolling process of the pancake product, there are multiple influences on the surrounding light, such as: occlusion by other mechanical actions, etc. Therefore, it is necessary to perform optimal fill light to avoid errors when the machine vision image is used to determine the current pancake thickness. Before the next rolling stroke of the pancake rolling machine begins, the lighting environment of the pancake product when the pancake rolling machine performs the next rolling stroke is dynamically simulated. The lighting environment refers to the dynamic environment in which the light irradiating the pancake product changes due to the movement of the rolling roller in the next rolling stroke. After simulation, a dynamic simulated lighting environment is obtained. The target correction rule determined last time refers to the target correction rule determined last time based on the feature description vector, by querying the correction rule library. The target correction rule determined last time determines the focus of image acquisition, and the focus of image acquisition is the area on the cake product that needs to be photographed. Based on the dynamic simulation of the lighting environment and the focus of image acquisition, the machine vision camera array in-position control strategy, the fill light device array in-position control strategy and the image acquisition timing are determined. The machine vision camera array in-position control strategy is the strategy for controlling the machine vision camera array in position. Correspondingly, the fill light device array in-position control strategy is the strategy for controlling the fill light device array in position. The image acquisition timing is the timing for controlling the machine vision camera array to acquire the machine vision image of the cake product. The machine vision camera array is an array composed of multiple movable machine vision cameras, and the fill light device array is an array composed of multiple movable supplementary devices. The two arrays are arranged next to the rolling table where the cake product is located.

[0090] The embodiment of the present invention optimizes the machine vision image acquisition process by dynamically simulating the lighting environment and combining the last target correction rule, ensuring that clear and accurate images of the pancake product can be obtained in different rolling strokes of the pancake rolling machine. By reasonably controlling the placement of the machine vision camera array and the fill light device array, and accurately controlling the image acquisition timing, image errors caused by lighting changes or mechanical occlusions can be avoided.

[0091] In one embodiment, determining the focus of image acquisition based on the target correction rule determined last time includes:

[0092] Analyze the correction object type of the target correction rule determined last time;

[0093] determining the affected part of the pie product corresponding to the type of correction object;

[0094] Focus image acquisition on the affected area.

[0095] The correction object type refers to the type of correction object of the target correction rule, which reflects the focus of the correction. Therefore, it can be based on the corresponding affected part of the cake product. For example, if the correction object type is to increase the rolling pressure of a certain rolling area, the affected part is the cake area corresponding to the rolling area. Finally, the affected part is used as the focus of image acquisition.

[0096] In one embodiment, the method of determining the machine vision camera array position control strategy, the fill light device array position control strategy and the image acquisition timing based on the dynamic simulation lighting environment and the image acquisition focus includes:

[0097] Predict the rolling status of the pancake product after the last rolling stroke of the pancake rolling machine is completed;

[0098] Based on the rolling situation, the simulated cake model in the dynamic simulated lighting environment is simulated and configured accordingly;

[0099] Determine a first model region corresponding to the focus of image acquisition and a second model region other than the first model region from the simulation pie model after the situation simulation configuration;

[0100] generating a first image acquisition constraint based on the multimodal parameters of the first model region;

[0101] Control the machine vision camera array model and the fill light device array model in the dynamic simulated lighting environment to perform dynamic simulation of image acquisition, and control to pause the dynamic simulation of image acquisition after reaching the optimal condition; wherein the optimal condition includes: the first image of the first model area captured by the machine vision camera array model under the fill light effect of the fill light device array model meets the image acquisition constraint and the second image of the second model area captured meets the standard second image acquisition constraint;

[0102] Determining a machine vision camera array in-position control strategy based on a first current in-position parameter of a machine vision camera array model in a dynamic simulated lighting environment after controlling to pause the dynamic simulation of image acquisition;

[0103] Determine the fill light device array position control strategy based on a second current position parameter of the fill light device array model in the dynamic simulated lighting environment after controlling the pause to perform the dynamic simulation of image acquisition;

[0104] Based on the control pause for dynamic simulation of image acquisition, the progress of the next rolling stroke of the pancake rolling machine in the dynamic simulation lighting environment is simulated to determine the timing of image acquisition.

[0105] The rolling conditions at least include: the thickness of the cake at different positions on the cake product; when predicting the rolling conditions, the original condition of the cake product is determined based on the machine vision image of the cake product obtained last time, and then based on the future rolling control strategy corrected last time, the rolling condition of the cake product in the last rolling stroke is determined, and the rolling condition is predicted based on the original condition and the rolling condition. The prediction can be based on a large number of historical records of cake products in the rolling process. There will be a simulated cake model in the dynamic simulated lighting environment, and the other cake products will be simulated and configured based on the rolling condition, and the simulated cake model will be configured to be consistent with the current cake product condition. The multimodal parameters of the first model area include: parameters such as the shape and position of the first model area; the first image acquisition constraint is to constrain the provision of appropriate fill light to the first model area (ensuring that the illumination of the area is sufficient and uniform, etc.) and to shoot the first model area at an appropriate position (ensuring that the shooting range covers the entire model area, etc.); the standard second image acquisition constraint is to constrain the provision of basic appropriate fill light to the second model area (basically sufficient illumination, etc.) and to shoot the second model area at a basically appropriate position (the shooting range basically covers the entire model area, etc.); the machine vision camera array model and the fill light device array model in the dynamic simulated lighting environment are controlled to perform dynamic simulation of image acquisition until the optimal condition is reached and the control is suspended to perform dynamic simulation of image acquisition. At this time, the machine vision camera array in-place control strategy can be determined based on the first current in-place parameter, and the fill light device array in-place control strategy can be determined based on the second current in-place parameter, and the travel progress can be used as the image acquisition timing; the first current in-place parameter refers to the current position distribution, respective shooting angles, etc. of the machine vision camera array model in the dynamic simulated lighting environment; correspondingly, the second current in-place parameter refers to the current position distribution, respective fill light angles, fill light intensity, etc. of the fill light device array model in the dynamic simulated lighting environment.

[0106] The embodiment of the present invention can accurately determine the positioning control strategy and image acquisition timing of the machine vision camera array and the fill light device array based on dynamic simulation of the lighting environment and the focus of image acquisition, thereby optimizing the image acquisition conditions during the rolling process. By predicting the rolling situation and simulating the lighting environment, the optimization of the lighting and shooting angle of the image shooting area is ensured, the imaging quality is improved, and the monitoring and control accuracy of the rolling quality is further improved. In addition, real-time adjustment based on dynamic simulation can effectively adapt to the needs of different rolling stages, and realize an automated, efficient, and accurate image acquisition and analysis process.

[0107] In one embodiment, the pancake thickness control system of the pancake rolling machine based on visual detection further includes:

[0108] The expected pancake thickness information acquisition module is used to acquire the user's expected pancake thickness information when the user inputs a personalized pancake thickness customization instruction;

[0109] The expected pancake thickness information adding module is used to add the expected pancake thickness information into the future rolling control strategy.

[0110] When the user inputs a personalized pancake thickness customization instruction, it means that the user wants to customize the pancake thickness, and obtains the expected pancake thickness information, which is the thickness of the pancake product that the user expects to customize the personalized pancake thickness. The expected pancake thickness information is added to the future rolling control strategy to replace the current rolling target thickness, so that the pancake rolling machine can roll the pancake according to the user's expectations, and finally output the pancake product with the personalized customized pancake thickness that the user expects.

[0111] In one embodiment, the expected cake thickness information acquisition module acquires the user's expected cake thickness information, including:

[0112] Guide users to approach the pie-thick interactive screen;

[0113] When the user approaches the pie thickness interactive screen, the pie thickness interactive screen is controlled to display the pie thickness control axis and pie thickness control prompt information to the user; wherein the pie thickness control axis is a vertical numerical axis that gradually increases from 0 scale to the maximum pie thickness scale from bottom to top;

[0114] Obtain the landing point trajectory formed by the user's sight point on the pie-thick interactive screen moving as the user's sight line changes;

[0115] Determine the last first target landing point that meets the first landing point condition from the landing point trajectory;

[0116] Determine a first local trajectory after the first target landing point from the landing point trajectory;

[0117] Determine a second target landing point that meets a second landing point condition from the first local trajectory;

[0118] Determine the first target cake thickness scale on the cake thickness control axis corresponding to the perpendicular point of the second target landing point, and associate it with the corresponding second target landing point;

[0119] When the second target landing point is unique, the expected pie thickness information is determined based on the first target pie thickness scale associated with the second target landing point; otherwise, if the second target landing points meet the landing point association condition, the selection auxiliary information corresponding to the first target pie thickness scale associated with the two second target landing points with the farthest distance between them is obtained;

[0120] Control the interactive screen to display selection auxiliary information to the user;

[0121] When the landing point trajectory continues to generate a new second partial trajectory, the last third target landing point that meets the second landing point condition is determined from the second partial trajectory;

[0122] Determine the second target cake thickness scale on the cake thickness control axis corresponding to the position of the perpendicular point of the third target landing point to the cake thickness control axis;

[0123] Determining expected cake thickness information based on the second target cake thickness scale;

[0124] The first landing point condition includes: the straight line connecting the landing point and the 0 scale position on the pancake thickness control axis is perpendicular to the pancake thickness control axis;

[0125] The second landing point condition includes: the landing point stay time exceeds the time threshold;

[0126] The landing point association condition includes: the straight-line distance between at least two landing points among the landing points is less than or equal to a first distance threshold and the straight-line distance between at least two landing points among the landing points is greater than or equal to a second distance threshold; the first distance threshold is less than the second distance threshold.

[0127] The working principle and beneficial effects of the above technical solution are:

[0128] The pancake thickness interactive screen is a display screen set up for interacting with users. It is installed next to the pancake rolling machine to guide users to approach the pancake thickness interactive screen. Guidance can be provided through voice broadcast. When the user approaches the pancake thickness interactive screen, the pancake thickness interactive screen is controlled to display the pancake thickness control axis and pancake thickness control prompt information to the user. The pancake thickness control prompt information prompts the user how to interact with the pancake thickness control axis through changes in line of sight to determine the desired pancake thickness, such as: moving the line of sight from the 0 scale to the scale of the desired pancake thickness; because during the rolling process, the pancake thickness will only continue to decrease, and the maximum pancake thickness will not exceed the thickness of the current pancake product. The first target landing point meets the first landing point condition, which means that the first local trajectory after it can be used to determine the pie thickness actually expected by the user. The second target landing point meets the second landing point condition, which means that it can be further used to determine the pie thickness actually expected by the user. When the second target landing point is unique, it means that the first target pie thickness scale associated with the second target landing point is the pie thickness actually expected by the user, and the first target pie thickness scale associated with the second target landing point is used as the expected pie thickness information; otherwise (the second target landing point is not unique), if the second target landing points meet the landing point association condition, it means that the user is hesitating to choose his actual expected thickness, and the selection auxiliary information corresponding to the first target pie thickness scale associated with the two second target landing points with the farthest distance between them is obtained. The selection auxiliary information is used for the user to overcome hesitation and finally choose his actual The expected thickness, such as the taste difference of the first target pancake thickness scale associated with the two second target landing points with the farthest spacing, the graphical information of the pancake size difference, etc. The purpose of selecting the farthest spacing is to provide the user with the greatest difference prompt effect, which is most convenient for them to intuitively understand the taste and pancake size of different pancake thicknesses, and improve the auxiliary effect. After the user has viewed the auxiliary selection information, he will continue to select the expected thickness through changes in his line of sight. The landing point trajectory will continue to generate a new second local trajectory, and the last third target landing point that meets the second landing point conditions is determined from the second local trajectory. The position of the vertical point of the third target landing point to the pancake thickness control axis is determined. The second target pancake thickness scale corresponding to the vertical point of the vertical line from the third target landing point to the pancake thickness control axis is determined, which is the expected thickness finally selected by the user. Finally, based on the second target pancake thickness scale, the expected pancake thickness information is determined.

[0129] In the first landing point condition, the straight line connecting the landing point and the 0 scale position on the pancake thickness control axis is perpendicular to the pancake thickness control axis, indicating that the user has officially started to make a selection based on the pancake thickness control prompt information, so that the first local trajectory thereafter can be used to determine the pancake thickness actually expected by the user, avoiding unnecessary occupation of processing resources due to excessive processing of the trajectory. In the second landing point condition, the landing point dwell time refers to the length of time that the landing point stops moving, that is, the length of time that the user continues to stare at it. The duration threshold can be 10 seconds. When the second landing point condition is met, it means that the user wants the thickness scale corresponding to the landing point as his or her expected pancake thickness. In the landing point association condition, the first distance threshold can be 2 cm, and the second distance threshold can be 6 cm. There are at least two landing points whose straight-line distance is less than or equal to the first distance threshold and there are at least two landing points whose straight-line distance is greater than or equal to the second distance threshold. This means that the user has continuously stared at two different thickness scales back and forth at least once, indicating that the user is hesitant in making a choice.

[0130] The embodiment of the present invention guides the user to interact with the pancake thickness interactive screen to accurately obtain the pancake thickness information expected by the user, thereby improving the user experience and optimizing the pancake rolling process. By capturing and analyzing the landing point trajectory formed by the change of sight, the system can accurately identify the user's choice on the pancake thickness control axis, avoiding the cumbersome process of traditional manual adjustment. The first landing point condition ensures the validity of the trajectory, avoids unnecessary data processing, and reduces resource occupation. The second landing point condition confirms the user's true intention by the length of stay, ensuring the accuracy of the selection. In addition, the design of the landing point association condition effectively identifies the user's hesitation and helps the user make decisions by providing auxiliary information, such as intuitive illustrations of different thicknesses of taste, pancake size differences, etc. It can provide effective guidance when the user hesitates and improve the decision-making efficiency of the choice. Overall, this technology not only improves the accuracy of the operation, but also enhances the intelligence of the user's interaction with the device, making the pancake thickness adjustment more humane and intelligent.

[0131] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A cake thickness control system for a cake rolling machine based on visual detection, characterized in that: include: A machine vision image acquisition module, used to acquire a machine vision image of a cake product when the cake rolling machine rolls out a new cake product; A current cake thickness determination module is used to detect the current cake thickness based on machine vision images; The future rolling control strategy correction module is used to correct the future rolling control strategy of the pancake rolling machine based on the current pancake thickness; The relay rolling control module is used to perform corresponding relay rolling control on the pancake rolling machine based on the revised future rolling control strategy.

2. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 1, characterized in that: The machine vision image acquisition module acquires the machine vision image of the pie product, including: The machine vision image of the pie product is obtained by a machine vision camera.

3. The pancake thickness control system of the pancake rolling machine based on visual detection as claimed in claim 1, characterized in that: Based on machine vision images, the current cake thickness is detected, including: Based on image recognition technology, the current cake thickness is detected according to the machine vision image.

4. The pancake thickness control system of the pancake rolling machine based on visual detection as claimed in claim 1, characterized in that: The future rolling control strategy correction module corrects the future rolling control strategy of the pancake rolling machine based on the current pancake thickness, including: Perform feature description processing on the current pancake thickness and future rolling control strategy to obtain a feature description vector; Based on the feature description vector, query the correction rule library to determine the target correction rule; Based on the target correction rules, the future rolling control strategy is corrected accordingly.

5. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 1, characterized in that: The method of obtaining a machine vision image of a cake product comprises: Before the next rolling stroke of the pancake rolling machine begins, dynamically simulate the lighting environment of the pancake product when the pancake rolling machine performs the next rolling stroke to obtain a dynamic simulated lighting environment; Based on the target correction rule determined last time, determine the focus of image acquisition; Based on the dynamic simulation of the lighting environment and the focus of image acquisition, determine the machine vision camera array positioning control strategy, the fill light device array positioning control strategy and the image acquisition timing; Based on the positioning control strategy of the machine vision camera array, the machine vision camera array in the pancake rolling machine is controlled to be positioned accordingly; Based on the positioning control strategy of the fill light device array, the positioning control strategy of the fill light device array in the pancake rolling machine is controlled to be positioned accordingly; When the rolling process begins, if the image acquisition timing is reached, the machine vision camera array is controlled to acquire the machine vision image of the cake product.

6. The pancake thickness control system of the pancake rolling machine based on visual detection as claimed in claim 5, characterized in that: The step of determining the focus of image acquisition based on the target correction rule determined last time includes: Analyze the correction object type of the target correction rule determined last time; determining the affected part of the pie product corresponding to the type of correction object; Focus image acquisition on the affected area.

7. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 5, characterized in that: The method of determining the positioning control strategy of the machine vision camera array, the positioning control strategy of the fill light device array and the image acquisition timing based on the dynamic simulation lighting environment and the image acquisition focus includes: Predict the rolling status of the pancake product after the last rolling stroke of the pancake rolling machine is completed; Based on the rolling situation, the simulated cake model in the dynamic simulated lighting environment is simulated and configured accordingly; Determine a first model region corresponding to the focus of image acquisition and a second model region other than the first model region from the simulation pie model after the situation simulation configuration; generating a first image acquisition constraint based on the multimodal parameters of the first model region; Control the machine vision camera array model and the fill light device array model in the dynamic simulated lighting environment to perform dynamic simulation of image acquisition, and control to pause the dynamic simulation of image acquisition after reaching the optimal condition; wherein the optimal condition includes: the first image of the first model area captured by the machine vision camera array model under the fill light effect of the fill light device array model meets the image acquisition constraint and the second image of the second model area captured meets the standard second image acquisition constraint; Determining a machine vision camera array in-position control strategy based on a first current in-position parameter of a machine vision camera array model in a dynamic simulated lighting environment after controlling to pause the dynamic simulation of image acquisition; Determine the fill light device array position control strategy based on a second current position parameter of the fill light device array model in the dynamic simulated lighting environment after controlling the pause to perform the dynamic simulation of image acquisition; Based on the control pause for dynamic simulation of image acquisition, the progress of the next rolling stroke of the pancake rolling machine in the dynamic simulation lighting environment is simulated to determine the timing of image acquisition.

8. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 1, characterized in that: Also includes: The expected pancake thickness information acquisition module is used to acquire the user's expected pancake thickness information when the user inputs a personalized pancake thickness customization instruction; The expected pancake thickness information adding module is used to add the expected pancake thickness information into the future rolling control strategy.

9. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 8, characterized in that: The expected pie thickness information acquisition module acquires the user's expected pie thickness information, including: Guide users to approach the pie-thick interactive screen; When the user approaches the pie thickness interactive screen, the pie thickness interactive screen is controlled to display the pie thickness control axis and pie thickness control prompt information to the user; wherein the pie thickness control axis is a vertical numerical axis that gradually increases from 0 scale to the maximum pie thickness scale from bottom to top; Obtain the landing point trajectory formed by the user's sight point on the pie-thick interactive screen moving as the user's sight line changes; Determine the last first target landing point that meets the first landing point condition from the landing point trajectory; Determine a first local trajectory after the first target landing point from the landing point trajectory; Determine a second target landing point that meets a second landing point condition from the first local trajectory; Determine the first target cake thickness scale on the cake thickness control axis corresponding to the perpendicular point of the second target landing point, and associate it with the corresponding second target landing point; When the second target landing point is unique, the expected pie thickness information is determined based on the first target pie thickness scale associated with the second target landing point; otherwise, if the second target landing points meet the landing point association condition, the selection auxiliary information corresponding to the first target pie thickness scale associated with the two second target landing points with the farthest distance between them is obtained; Control the interactive screen to display selection auxiliary information to the user; When the landing point trajectory continues to generate a new second partial trajectory, the last third target landing point that meets the second landing point condition is determined from the second partial trajectory; Determine the second target cake thickness scale on the cake thickness control axis corresponding to the position of the perpendicular point of the third target landing point to the cake thickness control axis; Based on the second target pie thickness scale, desired pie thickness information is determined.

10. The pancake thickness control system of a pancake rolling machine based on visual detection as claimed in claim 9, characterized in that: The first landing point condition includes: the straight line connecting the landing point and the 0 scale position on the pancake thickness control axis is perpendicular to the pancake thickness control axis; The second landing point condition includes: the landing point stay time exceeds the time threshold; The landing point association condition includes: the straight-line distance between at least two landing points among the landing points is less than or equal to a first distance threshold and the straight-line distance between at least two landing points among the landing points is greater than or equal to a second distance threshold; the first distance threshold is less than the second distance threshold.

Images