Spicy chicken product seasoning optimization method and system based on different materials

Abstract

The invention relates to the technical field of food processing, and discloses a spicy chicken product seasoning optimization method and system based on different materials, and the method comprises the steps: obtaining an infrared image sequence of a spicy chicken stir-frying process, carrying out the dense optical flow calculation of the infrared image sequence, and constructing a thermal texture optical flow field; decomposing the thermal texture optical flow field to obtain a spin-free potential flow component and a scattering-free vortex component, and constructing an evaporation potential energy field; a surface temperature field of the infrared image sequence is extracted, a two-dimensional phase space is constructed based on the evaporation potential energy field and the surface temperature field, a dynamic point cloud cluster is formed, and the phase volume of the dynamic point cloud cluster is calculated; tracking a time evolution curve of the phase volume, determining a thermal synchronization candidate moment, extracting an isotherm, carrying out hole identification on a connected domain defined by the isotherm, obtaining the number of holes, and determining seasoning feeding time. According to the method, the thermal synchronization time when the state difference of materials at different parts is minimized is locked, bubble noise and oil film defects are effectively distinguished, and uneven adsorption caused by misjudgment is reduced.

Description

Technical Field

[0001] This invention relates to the field of food processing technology, specifically to a method and system for optimizing the seasoning of spicy chicken products based on different ingredients. Background Technology

[0002] Spicy chicken, a Sichuan-Chongqing style cooked food, relies heavily on the precise coordination of heat and timing of ingredient addition during the high-temperature frying and re-stir-frying processes in its industrial production. With the evolution of food engineering technology, automated planetary fryers and continuous frying equipment have been widely used in the large-scale preparation of spicy chicken. The processing is usually controlled by preset time and temperature curves or single-point temperature feedback based on thermocouples. In order to pursue the consistency of product taste, some advanced production lines have begun to try to introduce non-contact infrared imaging or machine vision technology. By monitoring the overall color change or average radiation temperature of the material surface, they can help determine the timing of seasoning addition, in order to improve the quality stability of automated production.

[0003] However, the industrial stir-frying process involves high-intensity mechanical stirring. The strong turbulent motion driven by the stirring blades often masks the subtle characteristics of moisture evaporation and diffusion on the surface of the chicken. Existing visual technologies struggle to eliminate the interference of mechanical motion, leading to distorted judgments of dehydration status. Secondly, the raw materials for spicy chicken naturally include different tissues such as chicken skin, lean meat, and bones. The specific heat capacity, thermal conductivity, and moisture content of each part vary greatly, resulting in inconsistent thermal response rates during the heating process. Existing technologies often rely on global average temperature for judgment, which can easily lead to asynchronous phenomena such as some raw materials charring while others are under-dehydrated. Furthermore, during the high-temperature re-frying stage, the boiling bubbles in the oil film are visually easily confused with adsorption defects. Existing technologies lack the ability to effectively distinguish between bubbles and adsorption defects, resulting in the seasonings failing to form effective adsorption or detaching. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for optimizing the seasoning of spicy chicken products based on different ingredients, comprising:

[0006] Infrared image sequences of the spicy chicken stir-frying process are obtained, and dense optical flow calculations are performed on the infrared image sequences to construct a thermal texture optical flow field;

[0007] The thermal texture optical flow field is decomposed to obtain an irrotational potential flow component and a divergent eddy current component. An evaporation potential energy field is constructed based on the irrotational potential flow component and the divergent eddy current component.

[0008] The surface temperature field of the infrared image sequence is extracted, and a two-dimensional phase space is constructed based on the evaporation potential energy field and the surface temperature field. The pixels on the material surface are mapped to the two-dimensional phase space to form a dynamic point cloud. The phase volume of the dynamic point cloud is calculated.

[0009] The time evolution curve of the phase volume is tracked, and the local minimum value of the phase volume is used as the optimization target to determine the candidate time of thermal synchronization. The isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization are extracted, and the connected domains enclosed by the isotherms are identified to obtain the number of holes. The seasoning addition time is determined according to the number of holes and the candidate time of thermal synchronization to achieve the optimization of the seasoning of spicy chicken products.

[0010] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients described in this invention, the method for constructing an evaporation potential energy field based on the irrotational potential flow component and the non-dispersion eddy current component includes performing Helmholtz decomposition on the thermal texture optical flow field and decomposing the thermal texture optical flow field into an irrotational potential flow component and a non-dispersion eddy current component by solving the Poisson equation.

[0011] Calculate the curl intensity distribution of the non-dispersion eddy component, and construct a suppression mask based on the curl intensity distribution, wherein the curl intensity distribution represents the spatial influence range of the mechanical stirring action in the wok;

[0012] The background flow field component is obtained by spatially weighting the non-dispersion eddy component using the suppression mask.

[0013] A phase change diffusion component is constructed based on the irrotational potential flow component, and the phase change diffusion component is fused with the background flow field component to obtain the evaporation potential energy field. The feature fusion includes retaining the diffusion-type motion characteristics related to the outward evaporation of water in the phase change diffusion component, while using the background flow field component to correct the global displacement deviation caused by mechanical stirring.

[0014] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients described in this invention, the calculation of the phase volume of the dynamic point cloud includes normalizing the evaporation potential energy field and the surface temperature field to obtain normalized evaporation potential energy value and normalized temperature value; and constructing a two-dimensional phase space with the normalized evaporation potential energy value as the first dimension and the normalized temperature value as the second dimension.

[0015] The pixels on the surface of the material are traversed, the normalized evaporation potential energy value and normalized temperature value corresponding to each pixel are extracted and projected into the two-dimensional phase space to generate corresponding mapping points, and a dynamic point cloud is formed based on the set of the mapping points.

[0016] The convex hull algorithm is used to extract the outer contour vertices of the dynamic point cloud, and the area of ​​the region enclosed by the outer contour vertices is calculated as the phase volume; the phase volume represents the degree of dispersion of the material in the thermodynamic state.

[0017] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients described in this invention, the step of determining the candidate time of thermal synchronization includes collecting the phase volume on a continuous time series and constructing the time evolution curve of the phase volume.

[0018] The time derivative of the time evolution curve is calculated, and the inflection point where the time derivative changes from a positive value to a negative value is identified to determine that the phase volume has entered the shrinkage stage.

[0019] After the phase volume enters the shrinkage stage, the convergence rate of the phase volume is monitored, and the local minimum point of the phase volume is determined according to the convergence rate, and the corresponding time is recorded as the candidate time of thermal synchronization. The candidate time of thermal synchronization represents the moment when the overall distribution difference between the surface temperature and moisture evaporation state of different materials in the two-dimensional phase space is minimized.

[0020] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients according to the present invention, the method for identifying holes in the connected regions enclosed by the isotherms includes, at the candidate thermal synchronization time, extracting the connected regions in the surface temperature field that are greater than the preset oil film-forming temperature threshold as potential oil film regions.

[0021] Topological analysis was performed on the potential oil film region to identify closed low-value regions within the connected regions as topological holes.

[0022] Construct a survival state map of the topological pores over time to distinguish between transient pores and structural pores; count the number of structural pores as the total number of pores; wherein, transient pores represent random noise generated on the surface of the high-temperature oil film due to the rupture of boiling bubbles, and structural pores represent oil film defects caused by insufficient surface tension or obstruction of water evaporation airflow.

[0023] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients as described in this invention, the construction of the survival status map of the topological holes in the time series includes tracking the survival status of topological holes at the same spatial location in multiple consecutive frames of images and calculating the life cycle duration of each topological hole.

[0024] The lifecycle duration distribution of all topological holes at the current moment is statistically analyzed, and a lifecycle benchmark is calculated based on the statistical characteristics of the lifecycle duration distribution. Holes with a lifecycle duration less than or equal to the lifecycle benchmark are marked as transient holes, and holes with a lifecycle duration greater than the lifecycle benchmark are marked as structural holes. The lifecycle benchmark is adjusted according to the degree of oil temperature fluctuation during the frying process to eliminate the interference of boiling bubbles on the determination of oil film integrity.

[0025] As a preferred embodiment of the method for optimizing the seasoning of spicy chicken products based on different ingredients according to the present invention, the method for determining the seasoning addition time based on the number of holes and the candidate time of thermal synchronization includes real-time monitoring of the volume-time evolution curve and a verification stage based on the candidate time of thermal synchronization.

[0026] During the verification phase, the number of structural pores is monitored. When the number of structural pores is less than a preset pore threshold and the phase volume remains within the neighborhood of a local minimum, a seasoning dispensing instruction is generated. The seasoning dispensing instruction is configured to trigger a pulsed spray action, and the duration of the pulsed spray action matches the effective window period of the thermal synchronization candidate time.

[0027] A seasoning optimization system for spicy chicken products based on different ingredients, wherein:

[0028] The stir-frying data module acquires an infrared image sequence of the stir-frying process of spicy chicken, performs dense optical flow calculation on the infrared image sequence, and constructs a thermal texture optical flow field.

[0029] The potential energy field construction module decomposes the thermal texture optical flow field to obtain an irrotational potential flow component and a divergent eddy current component, and constructs an evaporation potential energy field based on the irrotational potential flow component and the divergent eddy current component.

[0030] The phase volume calculation module extracts the surface temperature field of the infrared image sequence, constructs a two-dimensional phase space based on the evaporation potential energy field and the surface temperature field, maps the pixels on the material surface to the two-dimensional phase space to form a dynamic point cloud, and calculates the phase volume of the dynamic point cloud.

[0031] The seasoning optimization module tracks the time evolution curve of the phase volume and uses the local minimum value of the phase volume as the optimization target to determine the candidate time of thermal synchronization. It extracts the isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization, identifies holes in the connected domain enclosed by the isotherms, obtains the number of holes, and determines the seasoning addition time based on the number of holes and the candidate time of thermal synchronization, thereby realizing the seasoning optimization of spicy chicken products.

[0032] A computer device includes: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0033] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in any one of the present invention.

[0034] The beneficial effects of this invention are as follows: This invention decouples the stir-frying flow field into an irrotational potential flow representing water evaporation and a non-dispersion vortex representing mechanical stirring. By constructing a vortex suppression mask to filter out displacement noise, it accurately captures subtle water phase change characteristics under the strong interference conditions of industrial stir-frying. Addressing the issue of large differences in thermal inertia in different parts of spicy chicken, a two-dimensional phase space of evaporation potential energy and surface temperature is constructed. Utilizing the local minimum of phase volume contraction in dynamic point cloud clusters, the thermal synchronization moment with the minimum material state difference in different parts is locked, solving the problem of inconsistent feeding timing caused by differences in thermal inertia in different parts. Temporal topology analysis is introduced. Through pore survival state maps and life cycle statistics, transient boiling bubble noise and structural oil film defects are effectively distinguished, ensuring that feeding is triggered only when the physical structure of the oil film is relatively intact, reducing misjudgment-induced uneven adsorption and lowering the seasoning loss rate. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 The overall flowchart of the method for optimizing the seasoning of spicy chicken products based on different ingredients provided in the embodiments of the present invention is shown.

[0037] Figure 2 This is a flowchart for determining the timing of seasoning addition, provided in an embodiment of the present invention.

[0038] Figure 3 A computer device diagram illustrating the method for optimizing the seasoning of spicy chicken products based on different ingredients, provided in an embodiment of the present invention. Detailed Implementation

[0039] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0040] Example 1, referring to Figures 1-3 As an embodiment of the present invention, a method for optimizing the seasoning of spicy chicken products based on different ingredients is provided, including:

[0041] S1: Obtain an infrared image sequence of the spicy chicken stir-frying process, perform dense optical flow calculation on the infrared image sequence, and construct a thermal texture optical flow field.

[0042] During the stir-frying process of spicy chicken, the surface of the material undergoes multiple thermodynamic state evolutions, including the coupling of various physical processes such as moisture evaporation, oil penetration, and temperature distribution changes. These processes directly affect the adhesion of seasonings and the final flavor. This invention obtains the temperature distribution of the material surface through infrared image sequences and uses a dense optical flow algorithm to construct a thermal texture optical flow field representing the micro-texture motion of the material surface. The thermal texture optical flow field contains composite information of two motion modes: mechanical stirring and moisture evaporation, laying the foundation for subsequent signal purification through vector field decomposition.

[0043] Specifically, the process of acquiring an infrared image sequence of the spicy chicken stir-frying process includes the following steps: an infrared thermal imaging camera is installed above the stir-frying equipment in the spicy chicken stir-frying process, and the temperature distribution images of the material surface are continuously acquired according to a preset sampling frame rate to form an infrared image sequence; the field of view of the infrared thermal imaging camera covers the entire surface of the stir-frying equipment, while avoiding interference from high-temperature components such as the edge of the pot and the stove; the preset sampling frequency is determined according to the stir-frying cycle to ensure that the change process of the temperature field on the material surface can be captured.

[0044] In one optional embodiment, the resolution of the infrared thermal imaging camera can be 640 pixels by 480 pixels, with a sampling frame rate of 20 frames per second, which can reduce the amount of data processing while ensuring temporal resolution. Secondly, in order to improve spatial resolution, the resolution of the infrared thermal imaging camera can be increased to 1024 pixels by 768 pixels, at which point the sampling frame rate can be appropriately reduced to 10 frames per second. In this case, the infrared thermal imaging camera detects the infrared energy radiated from the surface of the material and converts the temperature distribution into a grayscale image. The grayscale value of each pixel reflects the surface temperature at the corresponding spatial location. Compared with visible light cameras, infrared thermal imaging cameras are not affected by ambient light and oil fume obstruction, and can stably acquire the temperature field information of the material surface, providing a reliable data source for subsequent analysis based on the temperature field motion characteristics.

[0045] In this embodiment, the dense optical flow calculation of the infrared image sequence includes the following steps: preprocessing the infrared images of consecutive frames by removing image noise through Gaussian filtering; calculating the displacement vector of each pixel between adjacent frames using a dense optical flow algorithm to obtain the thermal texture optical flow field, wherein the dense optical flow algorithm is based on the assumption of constant image brightness and spatial smoothness constraints, and obtains the dense optical flow field by solving an optimization problem; specifically, the dense optical flow algorithm uses the Farneback algorithm, which constructs a quadratic polynomial approximation of the local region of the image based on the idea of ​​polynomial expansion, assuming that the brightness distribution of a pixel in the neighborhood of the previous frame image can be approximated as a quadratic polynomial. In a polynomial form, the brightness distribution at the corresponding position in the next frame image can also be approximated as a quadratic polynomial. By comparing the differences in the coefficients of the two polynomials, the displacement vector at the corresponding position can be derived, thereby obtaining a denser and more accurate optical flow field while maintaining computational efficiency. In an optional embodiment, the dense optical flow algorithm can also employ a pyramid algorithm. First, a multi-scale pyramid is constructed for the image. The optical flow is calculated starting from the coarsest scale and then progressively transferred to the finer scale. The optical flow calculated at the coarsest scale is used as the initial estimate for the finer scale. The optical flow field is gradually refined through iterative optimization, thereby obtaining a denser and more accurate optical flow field while maintaining computational efficiency.

[0046] Furthermore, the displacement vectors of each pixel calculated between adjacent frames are combined into a two-dimensional vector field. Each element of the vector field contains a horizontal component and a vertical component, which represent the movement speed of the pixel along the horizontal and vertical directions in the image coordinate system, respectively, thus obtaining the thermal texture optical flow field. Among them, under the action of mechanical stirring, the optical flow vector near the stirring blade area shows obvious rotational characteristics; while during the water evaporation process, the optical flow vector in the area with higher surface temperature shows a divergent characteristic from the inside to the outside.

[0047] It should be noted that, in response to the problem that traditional temperature detection methods can only obtain static temperature distribution and lack motion information, a thermal texture optical flow field is constructed by using infrared image sequences and dense optical flow algorithms. This combines temperature information with motion information and can represent the change characteristics of the material surface at the pixel level, providing data support for subsequent differentiation between mechanical disturbances and real evaporation signals by kinematic feature differences.

[0048] S2: Decompose the thermal texture optical flow field to obtain the irrotational potential flow component and the non-dispersion eddy current component, and construct the evaporation potential energy field based on the irrotational potential flow component and the non-dispersion eddy current component.

[0049] In this embodiment, since the thermal texture optical flow field obtained from step S1 contains mixed motion information of multiple physical processes such as mechanical stirring and water evaporation, it is difficult to accurately represent the evaporation state of the material by directly using the original optical flow field. Therefore, by performing vector field decomposition on the thermal texture optical flow field, the two motion modes are mathematically orthogonally decoupled by utilizing the difference between the high-rotational motion generated by mechanical stirring and the low-rotational motion generated by water evaporation. Then, by suppressing the high-rotational component and enhancing the low-rotational component, an evaporation potential energy field with a high signal-to-noise ratio is constructed.

[0050] Specifically, the process of decomposing the thermal texture optical flow field to obtain the irrotational potential flow component and the divergence-free eddy current component is as follows: The thermal texture optical flow field is decomposed by Helmholtz decomposition, and the thermal texture optical flow field is decomposed into an irrotational potential flow component and a divergence-free eddy current component by solving the Poisson equation; wherein, the Helmholtz decomposition is based on the vector field decomposition theorem, which orthogonally decouples any vector field into two mutually independent components. The irrotational potential flow component corresponds to the gradient field of the potential function, which mainly represents the divergent motion generated by water evaporation. The divergent motion diffuses from the inside to the outside and has zero curl. The divergence-free eddy current component corresponds to the orthogonal gradient field of the stream function, which mainly represents the rotational motion generated by mechanical stirring. The rotational motion has obvious rotational characteristics and zero divergence.

[0051] The specific calculation steps of the Helmholtz decomposition are as follows: First, the thermal texture optical flow field is analyzed. For each vector point in the vector, calculate the divergence and curl. Spatial location coordinates, The time variable is used, where the divergence is calculated using the central difference method, by calculating the thermal texture optical flow field. The curl is obtained by summing the partial derivatives of the horizontal component along the horizontal direction and the partial derivatives of the vertical component along the vertical direction; the curl is also calculated using the central difference method, by calculating the thermal texture optical flow field. The difference between the partial derivatives of the vertical component along the horizontal direction and the partial derivatives of the horizontal component along the vertical direction is obtained, thus generating the divergence and curl fields of the entire optical flow field. Next, two Poisson equations are constructed based on the divergence and curl fields respectively. For the potential flow component, the divergence field is used as the source term of the Poisson equation, and the potential function is obtained through numerical solution; the gradient of the potential function is the irrotational potential flow component. For the eddy component, the curl field is used as the source term of the Poisson equation, and the stream function is obtained through numerical solution; the curl of the stream function is the divergence-free eddy component. In this embodiment, the Poisson equation is discretized using the finite difference method, transforming the continuous partial differential equations into a system of linear equations. The linear equations are solved using an iterative algorithm to obtain the numerical solutions of the potential function and stream function, and then the irrotational potential flow component is obtained through numerical differentiation. and non-dispersion eddy component .

[0052] In one optional embodiment, the iterative algorithm can employ the Gauss-Seidel iteration method, which updates the values ​​of the potential function and the stream function point by point. The iteration process continues until the maximum difference between two adjacent iterations is less than a preset convergence threshold, which can control the computation time while ensuring computational accuracy. Alternatively, the iterative algorithm can employ the conjugate gradient method, which has a faster convergence speed than the Gauss-Seidel iteration method and is suitable for real-time processing of large-scale image sequences.

[0053] It should be noted that, in order to address the problem that the thermal texture optical flow field, which combines two motion modes of mechanical stirring and water evaporation, makes it difficult to directly represent the evaporation state, the composite motion field is deconstructed into two orthogonal components. The non-zero divergence of the undulation potential flow component corresponds to the divergence process of water evaporating from the inside of the material to the outside, while the non-zero curl of the undulation eddy current component corresponds to the rotational motion generated by the stirring blades. This avoids the limitation of traditional image processing methods in distinguishing different motion modes and achieves the decoupling of the evaporation signal and the mechanical disturbance signal at the physical level.

[0054] Furthermore, constructing the evaporation potential energy field based on the irrotational potential flow component and the divergent eddy current component includes the following steps: based on the divergent eddy current component The curl value is calculated for each vector point, where the divergence-free vortex component is calculated using the central difference method. The curl value at each position is obtained by taking the difference between the partial derivative of the vertical component along the horizontal direction and the partial derivative of the horizontal component along the vertical direction. Take the absolute value of the curl. As the curl intensity, the curl intensity distribution is obtained by traversing the entire flow field region. The rotation intensity distribution indicates the spatial influence range of the mechanical stirring action in the wok. The area near the stirring blades is directly driven by the machine and has a higher rotation intensity, exhibiting strong rotational motion. The area far from the stirring center is less affected by the mechanical action and has a lower rotation intensity. The movement of the material surface reflects more the natural diffusion caused by moisture evaporation.

[0055] Curl intensity distribution Normalization is performed to map the curl intensity values ​​at each location to a standard range of zero to one, resulting in the normalized curl intensity field. A suppression mask is constructed based on the normalized curl intensity field, and the specific formula is as follows:

[0056]

[0057] in, To suppress masking, the value range is [0, 1]. In regions with high curl intensity, the mask value is close to 0, and in regions with low curl intensity, the mask value is close to 1. Spatial location coordinates; The time variable is used; the suppression mask is used to spatially weight the suppression of areas with strong mechanical stirring effects. Subsequently, the suppression mask is used to process the non-dispersion eddy current component, which can weaken the interference of the strong rotational motion of the stirring center on the evaporation state determination and highlight the area that truly reflects the evaporation of water.

[0058] It should be noted that, in response to the problem that mechanical stirring can cause strong overall rotational motion on the surface of materials, which, if left untreated, will mask the local diffusion characteristics of moisture evaporation, the spatial suppression of the stirring-affected area is achieved by calculating the curl intensity distribution of the non-dispersion eddy component and constructing a reverse mask. This allows the subsequent construction of the evaporation potential field to focus on the area that truly reflects the moisture evaporation state, avoiding being dominated by the rigid displacement caused by stirring, and improving the accuracy of the evaporation state characterization.

[0059] Furthermore, the suppression mask and the diffuse eddy component are calculated pixel-by-pixel. That is, for each position in the flow field, the product of the suppression mask value and the vector value of the diffuse eddy component at that position is calculated to obtain the weighted background flow field component. The background flow field component retains the eddy information in the region where the stirring effect is weak. The eddies in these regions mainly reflect the local disturbances and boundary effects on the material surface, which are of reference value for correcting global displacement deviations. At the same time, by suppressing the strong rotational motion of the stirring center, the influence of strong interference signals on the subsequent construction of the evaporation potential energy field is avoided.

[0060] Furthermore, constructing the phase change diffusion component based on the irrotational potential flow component includes the following steps: First, calculating the irrotational potential flow component. The divergence field was used to calculate the irrotational potential flow component using the central difference method. The divergence value at each location is obtained by summing the partial derivatives of the horizontal component along the horizontal direction and the partial derivatives of the vertical component along the vertical direction. The magnitude of the divergence field reflects the divergence intensity at each location on the material surface. A positive divergence indicates outward divergent flow at the corresponding location, corresponding to the evaporation of moisture to the outside. A negative divergence indicates convergent flow.

[0061] Secondly, spatial smoothing filtering is performed on the divergence field to suppress high-frequency noise, resulting in a smoothed divergence field. In this embodiment, Gaussian filtering is used for spatial smoothing filtering. By setting appropriate filter kernel size and standard deviation parameters, random noise introduced by optical flow calculation errors is removed while preserving the main distribution characteristics of the divergence field.

[0062] Finally, the partial derivatives of the divergence field along the horizontal and vertical directions are calculated using the central difference method to obtain the gradient of the smoothed divergence field. The direction of the gradient points to the direction of the fastest increase in divergence, i.e., the region where water evaporation is most intense. The gradient vector is normalized to obtain the unit direction vector. The product of the smoothed divergence field value and the unit vector of the gradient direction is calculated, and a diffusion intensity adjustment coefficient is introduced to scale the amplitude to obtain the phase change diffusion component. The phase change diffusion component emphasizes the diffusion-type motion characteristics of water evaporation from the material surface, and its direction points to the region where evaporation is most intense, while its amplitude reflects the evaporation intensity.

[0063] In one optional embodiment, the diffusion intensity adjustment coefficient can be set to a fixed value, and a suitable numerical range can be determined through experimental calibration. The value is between 0.1 and 1, which can maintain the diffusion characteristics while avoiding numerical overflow. In addition, the diffusion intensity adjustment coefficient can also be determined according to the average temperature of the material surface. The higher the temperature, the more intense the evaporation, and the larger the corresponding adjustment coefficient value. By establishing a mapping relationship between temperature and adjustment coefficient, the response of the evaporation potential energy field to temperature changes can be realized.

[0064] Furthermore, the feature fusion of the phase change diffusion component and the background flow field component to obtain the evaporation potential energy field includes: calculating the amplitudes of the phase change diffusion component and the background flow field component at various spatial locations, i.e., the vector magnitudes; and weighting and summing the amplitudes of the phase change diffusion component and the background flow field component to obtain the numerical distribution of the evaporation potential energy field. This feature fusion retains the diffusion-type motion characteristics related to the outward evaporation of water in the phase change diffusion component while using the background flow field component to correct for global displacement deviations caused by mechanical stirring. Specifically, the amplitude of the phase change diffusion component represents the water evaporation state and is the main component of the evaporation potential energy field, with a relatively large weighting coefficient; the amplitude of the background flow field component is used to correct for global displacement deviations caused by stirring, with a relatively small weighting coefficient.

[0065] In this embodiment, the weighting coefficients of the phase change diffusion component and the background flow field component can be adjusted according to the intensity of the turbulence. The intensity of the turbulence is determined by monitoring the average value of the overall curl intensity of the non-dispersion eddy component. When the average curl intensity is large, it indicates that the turbulence is more intense. At this time, the weight of the background flow field component is increased to better correct the global displacement deviation. When the average curl intensity is small, it indicates that the turbulence is weak. At this time, the weight of the background flow field component is decreased to make the phase change diffusion component dominate.

[0066] It should be noted that, in the preparation process of spicy chicken products, traditional temperature field analysis methods can only reflect the static temperature distribution on the surface of the material and cannot quantify the process of moisture evaporation. Furthermore, these methods are easily affected by mechanical stirring. Therefore, Helmholtz decomposition is used to deconstruct the thermal texture optical flow field into two orthogonal components: potential flow and eddy current. Targeted processing is applied to each component. For the potential flow component, the divergence field is calculated and the diffusion characteristics are enhanced by combining the gradient direction to highlight evaporation information. For the eddy current component, the strong interference region at the center of stirring is suppressed by using curl intensity. Finally, a weighted fusion is used to obtain an evaporation potential energy field that accurately reflects the evaporation state and corrects global displacement deviations. This avoids the limitations of traditional methods in distinguishing between evaporation signals and mechanical disturbances, providing high-quality feature input for subsequent phase space analysis to determine candidate moments for thermal synchronization, thereby improving the accuracy of seasoning addition timing.

[0067] S3: Extract the surface temperature field of the infrared image sequence, construct a two-dimensional phase space based on the evaporation potential energy field and the surface temperature field, map the pixels on the material surface to the two-dimensional phase space to form a dynamic point cloud, and calculate the phase volume of the dynamic point cloud.

[0068] In this embodiment, the evaporation potential energy field constructed in step S2 describes the process of moisture evaporation on the material surface, while the surface temperature field in the infrared image sequence reflects the thermodynamic state of the material surface. Both describe the physical characteristics of the stir-frying process of spicy chicken products from kinematic and thermodynamic perspectives, respectively. This invention maps the state of each position on the material surface to points in a two-dimensional phase space by using the evaporation potential energy field and the surface temperature field as two coordinate dimensions of the phase space, forming a dynamic point cloud. The spatial distribution of the dynamic point cloud reflects the discreteness of the material surface state, while the change in phase volume quantifies the state convergence process, providing a geometric feature basis for subsequently determining the candidate time of thermal synchronization.

[0069] Specifically, extracting the surface temperature field of the infrared image sequence includes: for each frame of the infrared image sequence, converting the infrared radiation intensity value of each pixel into the corresponding surface temperature value according to the radiation-temperature conversion relationship calibrated by the infrared thermal imaging camera, to obtain the surface temperature field at each moment; the surface temperature field is a two-dimensional spatial distribution, and the temperature value of each pixel reflects the thermodynamic state of the corresponding material surface position.

[0070] Further, constructing a two-dimensional phase space based on the evaporation potential energy field and the surface temperature field, and mapping the pixels on the material surface to the two-dimensional phase space to form a dynamic point cloud includes the following steps: Normalizing the evaporation potential energy field and the surface temperature field respectively; traversing all pixels in the evaporation potential energy field at the current moment and calculating the maximum and minimum values ​​of the evaporation potential energy field; based on the maximum and minimum values ​​of the evaporation potential energy field at the current moment, mapping the evaporation potential energy value of each pixel to the first dimension interval using the minimum-maximum normalization method to obtain the normalized evaporation potential energy value; traversing all pixels in the surface temperature field at the current moment and calculating the maximum and minimum values ​​of the surface temperature field; based on the maximum and minimum values ​​of the surface temperature field at the current moment, mapping the temperature value of each pixel to the second dimension interval using the minimum-maximum normalization method to obtain the normalized temperature value.

[0071] Both the first and second dimension intervals are standard intervals from zero to one. Through normalization, the two physical quantities, evaporation potential energy and temperature, have the same numerical scale in phase space, which avoids one dimension from dominating the geometric structure of phase space due to an excessively large numerical range, and ensures the rationality of the distribution of point clouds in phase space.

[0072] Based on the evaporation potential energy field and the surface temperature field, an orthogonal coordinate system is established as a two-dimensional phase space with the normalized evaporation potential energy value as the first dimension and the normalized temperature value as the second dimension, where the horizontal axis represents the water evaporation state and the vertical axis represents the surface thermodynamic temperature.

[0073] The process iterates through the pixels on the material surface. For each pixel, the corresponding normalized evaporation potential energy value is extracted as the first-dimensional coordinate, and the corresponding normalized temperature value is extracted as the second-dimensional coordinate. Mapping points are then generated in the two-dimensional phase space. The position of each mapping point reflects the thermodynamic state of the corresponding location on the material surface. Locations with high evaporation potential energy and high temperature are mapped to the upper right region of the phase space, while locations with low evaporation potential energy and low temperature are mapped to the lower left region of the phase space. The mapping points of all pixels on the material surface are then collected to form a dynamic point cloud.

[0074] It should be noted that, in the process of stir-frying spicy chicken, the different sizes of chicken pieces result in different heat transfer rates and moisture evaporation rates, leading to differences in their thermodynamic states. Traditional single-feature analysis is insufficient to quantify these state differences. By using evaporation potential energy and temperature as two orthogonal dimensions of the phase space, the states at different locations on the material surface form dynamic point clouds in the phase space. The spatial distribution range of the dynamic point clouds reflects the non-uniformity of the material surface state. When the thermodynamic states of different materials differ significantly, the dynamic point clouds exhibit a large spatial distribution. When the thermodynamic states of different materials tend to be consistent, the dynamic point clouds shrink to a smaller area, providing a geometrical basis for subsequent quantification of the degree of state synchronization through phase volume.

[0075] Further, calculating the phase volume of the dynamic point cloud cluster includes: extracting the outer contour vertices of the dynamic point cloud cluster using a convex hull algorithm. The convex hull algorithm is based on computational geometry principles and finds the smallest convex polygon on a two-dimensional plane that can contain all the point clouds. The vertices of the smallest convex polygon are the outer contour vertices. Specifically, the convex hull algorithm first identifies the points with the smallest and largest x-coordinates in the point cloud cluster as initial vertices. Then, using these two points as references, it gradually expands the convex hull boundary by calculating the angle or distance of the line connecting other points relative to the reference points, until all point cloud points are located on the convex hull boundary or inside the convex hull.

[0076] The smallest convex polygon is connected to adjacent vertices in sequence to form a closed polygon, and the area of ​​the closed polygon is calculated. The area is used to represent the range occupied by the dynamic point cloud in the two-dimensional phase space, that is, the phase volume. The larger the phase volume, the more discrete the thermodynamic state of each position on the material surface is, and the more obvious the state difference between different materials. The smaller the phase volume, the more uniform the material surface state is, and the higher the degree of thermal synchronization.

[0077] It should be noted that, in the preparation of spicy chicken products, traditional methods struggle to quantify the dispersion and synchronization of the thermodynamic state of the material surface. By constructing a two-dimensional phase space of evaporation potential energy and temperature and mapping the pixel points on the material surface as dynamic point clouds, the composite state of the material surface is transformed into a geometric distribution in the phase space. The phase volume, as a measure of the spatial occupancy of the point cloud, directly reflects the non-uniformity of the material surface state. The temporal evolution of the phase volume quantifies the dynamic process of the material gradually converging from an initial highly non-uniform state to a thermally synchronized state during the frying process, providing quantifiable geometric characteristic indicators for subsequently determining the optimal timing for seasoning addition.

[0078] S4: Track the time evolution curve of the phase volume, and use the local minimum value of the phase volume as the optimization target to determine the candidate time of thermal synchronization. Extract the isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization, identify holes in the connected domain enclosed by the isotherms, obtain the number of holes, and determine the seasoning addition time based on the number of holes and the candidate time of thermal synchronization to achieve seasoning optimization of spicy chicken products.

[0079] Furthermore, as the frying process proceeds, the temperature and evaporation state of different ingredients gradually become consistent under the continuous heat transfer and moisture evaporation, and the phase volume shows an evolutionary trend of first increasing and then decreasing. The local minimum of the phase volume corresponds to the moment when the thermodynamic state of the material surface is more uniform, that is, the candidate moment of thermal synchronization. At this time, adding seasonings can make the seasonings adhere more uniformly to the material surface. However, relying solely on the local minimum of the phase volume is not enough to guarantee the effective adsorption of seasonings, because the integrity of the oil film on the material surface directly affects the adhesion effect of seasonings. When there are pores or defects in the oil film, the seasonings cannot be uniformly covered. Therefore, after determining the candidate moment of thermal synchronization, this invention further performs topological structure analysis on the surface temperature field, judges the integrity of the oil film by identifying pores, and determines the final seasoning addition time by combining the two conditions of thermal synchronization state and oil film topological structure.

[0080] Specifically, the process of tracking the time evolution curve of the phase volume and determining the candidate moment for thermal synchronization with the local minimum value of the phase volume as the optimization objective is as follows: According to the sampling frame rate of the infrared image sequence, the phase volume at the corresponding moment is calculated for each frame image. The phase volume values ​​at each moment are arranged in chronological order, and a time evolution curve is plotted with time as the horizontal axis and phase volume as the vertical axis. The time evolution curve reflects the evolution of the thermodynamic state of the material surface from initial non-uniformity to thermal synchronization. In the early stage of frying, there are differences in temperature and evaporation state between different materials, resulting in a larger phase volume. As frying progresses, heat transfer and moisture evaporation gradually make the surface state of the material more uniform, and the phase volume gradually decreases and approaches a local minimum value.

[0081] The time derivative of the time evolution curve is calculated using the central difference method. The inflection point where the time derivative changes from a positive value to a negative value is identified. The inflection point corresponds to the turning point when the phase volume changes from the growth stage to the contraction stage, indicating that the surface state of the material has entered the synchronization stage from the initial discretization stage.

[0082] It should be noted that the phase volume may show a brief increasing trend in the early stage of frying. This is because the temperature distribution of the material is relatively uniform when it is initially fed. With the irregular action of heating and frying, the temperature and evaporation state of different parts of the material surface become differentiated, resulting in an increase in phase volume. When the homogenization effect of heat transfer and moisture evaporation gradually becomes dominant, the phase volume turns to decrease. By identifying the inflection point, it can be determined that the phase volume has entered the shrinkage stage, which can accurately capture the key moment when the material state begins to evolve in the direction of thermal synchronization.

[0083] Furthermore, after the phase volume enters the shrinkage stage, the convergence rate of the phase volume is monitored. Based on the convergence rate, the local minimum point of the phase volume is determined, and the corresponding time is recorded as a candidate time for thermal synchronization. Specifically, by continuously calculating the time derivative of the time evolution curve, the absolute value of the time derivative represents the convergence rate of the phase volume. When the convergence rate gradually approaches zero, that is, when the absolute value of the time derivative is less than a preset threshold, it indicates that the phase volume is approaching a local minimum. The local minimum point of the time evolution curve is identified, that is, the moment when the time derivative changes from a negative value to zero or is close to zero. The corresponding time is recorded as a candidate time for thermal synchronization. The candidate time for thermal synchronization represents the moment when the overall distribution difference of the surface temperature and moisture evaporation state of different materials in the two-dimensional phase space is minimized. At this moment, the thermodynamic state of the material surface reaches the maximum uniformity.

[0084] The preset threshold can be set according to the stability requirements of the frying process. For example, for high-end products that require high uniformity, the threshold can be set to one percent of the initial value of the phase volume to ensure that the phase volume converges to a local minimum. For general products, the threshold can be appropriately relaxed to five percent to improve production efficiency while ensuring thermal synchronization.

[0085] It should be noted that, in response to the problem that traditional methods rely on human experience to determine the timing of seasoning addition or add seasoning at fixed times, resulting in poor flavor stability among different batches of products, this paper addresses the issue by tracking the time evolution curve of phase volume and identifying local minimum points. This transforms the thermal synchronization state of the material into a quantifiable geometric characteristic index. The local minimum point of phase volume corresponds to the moment when the difference between the material surface temperature and the evaporation state is small. Adding seasoning at this time allows for a more uniform distribution and adsorption of the seasoning on the material surface, improving the objectivity and repeatability of determining the timing of seasoning addition.

[0086] Further, extracting isotherms from the surface temperature field corresponding to the candidate thermal synchronization time, and identifying holes in the connected regions enclosed by the isotherms to obtain the number of holes includes the following steps: at the candidate thermal synchronization time, extracting connected regions in the surface temperature field that are greater than a preset oil film-forming temperature threshold as potential oil film regions, wherein the surface temperature field corresponding to the candidate thermal synchronization time is segmented by a threshold, and pixels with temperature values ​​greater than the preset oil film-forming temperature threshold are marked as high-temperature regions, and pixels with temperature values ​​less than or equal to the threshold are marked as low-temperature regions; performing connectivity analysis on the high-temperature regions, identifying spatially adjacent sets of high-temperature pixels as connected regions, and the connected regions correspond to potential regions on the material surface where an oil film has already formed or where film-forming conditions are present. In this embodiment, the preset oil film-forming temperature threshold is determined based on the physical properties of edible oil during the cooking of spicy chicken. When the surface temperature of the material is higher than the flow threshold temperature of the oil, the oil can form a continuous liquid film under the action of surface tension. For example, the oil film-forming temperature threshold is set between 120 and 150 degrees Celsius, and the specific value can be adjusted according to the type of oil used.

[0087] A topological analysis is performed on the potential oil film region to identify closed low-value regions within the connected regions as topological holes. Specifically, this includes: within each connected region, searching for a set of low-temperature pixels surrounded by high-temperature pixels, and determining whether the set of low-temperature pixels forms a closed region, i.e., the boundary of the set of low-temperature pixels is completely composed of high-temperature pixels; marking the set of low-temperature pixels that meets the closure condition as a topological hole; the topological hole corresponds to a temperature depression within the connected region of the oil film, which may be a defective region not completely covered by the oil film, or it may be an instantaneous temperature disturbance formed by the bursting of boiling bubbles.

[0088] A time-series survival status map of the topological holes is constructed to distinguish between transient and structural holes. Specifically, by tracking the survival status of topological holes at the same spatial location in multiple consecutive frames, for each topological hole, the time of its first appearance and the time of its last appearance are recorded, and the time difference between the two is calculated as the lifetime of the topological hole. The lifetime distribution of all topological holes at the current moment is statistically analyzed, and the mean of the distribution is calculated as the lifetime baseline. Holes with a lifetime duration less than or equal to the lifetime baseline are marked as transient holes, and holes with a lifetime duration greater than the lifetime baseline are marked as structural holes.

[0089] The transient holes refer to random noise generated on the surface of the high-temperature oil film due to the rupture of boiling bubbles. They have a short lifespan and usually disappear within a few frames of images. The structural holes refer to oil film defects caused by insufficient surface tension or obstruction of water evaporation airflow. They have a longer lifespan and persist in multiple consecutive frames of images. The number of structural holes is counted as the number of holes.

[0090] In this embodiment, the life cycle benchmark is adjusted according to the degree of oil temperature fluctuation during the frying process to eliminate the interference of boiling bubbles on the determination of oil film integrity: by monitoring the temperature variance of the high-temperature region in the surface temperature field, the degree of oil temperature fluctuation is quantified. When the temperature variance is large, it indicates that the boiling disturbance is severe and the number of transient bubbles increases, and the life cycle benchmark is increased accordingly to avoid misjudging short-lived bubbles as structural holes. When the temperature variance is small, the life cycle benchmark is reduced to improve the sensitivity of structural hole identification.

[0091] It should be noted that, in response to the issue that the integrity of the oil film on the material surface directly affects the adhesion of seasonings, a topological structure analysis of the surface temperature field is performed. Based on the connected domains enclosed by isotherms, topological voids are identified. Furthermore, lifecycle analysis is used to distinguish between transient boiling bubbles and structural oil film defects, thereby achieving a quantitative assessment of oil film integrity. This avoids uneven adhesion caused by adding seasonings when oil film defects exist, and provides topological constraints for determining the final seasoning addition time.

[0092] Furthermore, such as Figure 2 The flowchart shown is for determining the timing of seasoning addition. Determining the seasoning addition time based on the number of pores and the candidate thermal synchronization time includes: real-time monitoring of the time evolution curve of the phase volume; when the phase volume reaches a local minimum and a candidate thermal synchronization time is determined, the process enters the verification stage; in the verification stage, the number of structural pores is continuously monitored; when the number of structural pores is less than a preset pore threshold, it indicates that the oil film topology is relatively complete; when the phase volume remains within the neighborhood of the local minimum, it indicates that the thermodynamic state of the material is maintained in a highly synchronized state. When the above two conditions are met, a seasoning addition command is generated.

[0093] The local minimum neighborhood can be defined as a range of 10% above and below the local minimum of the phase volume. When the phase volume exceeds this range, the thermal synchronization state is no longer satisfied, and seasoning addition is stopped. Regarding the preset pore threshold, during the process debugging phase, data on the seasoning adsorption effect under different numbers of structural pores are collected through multiple experiments. A correlation curve between the number of structural pores and the flavor uniformity score of the finished product is established. The critical value of the number of pores corresponding to the flavor uniformity score dropping to the qualified standard line is selected as the preset pore threshold. Therefore, while ensuring the basic integrity of the oil film, the judgment conditions are appropriately relaxed to improve production efficiency.

[0094] In this embodiment, the seasoning delivery command is configured to trigger a pulse spraying action. The seasoning spraying system is controlled by a solenoid valve or a pneumatic valve to perform short-term spraying. The duration of the pulse spraying action matches the effective window period of the thermal synchronization candidate moment. Specifically, the effective window period is the length of time that the phase volume is maintained within the neighborhood of a local minimum, which is determined by monitoring the rate of change of the phase volume. When the phase volume begins to deviate from the neighborhood of the local minimum, the effective window period ends. The duration of the pulse spraying is set to 50% to 80% of the effective window period to ensure that the seasoning adsorption is completed when the thermodynamic state of the spicy chicken material is synchronized and the oil film topology is relatively intact.

[0095] It should be noted that, in response to the problem that traditional seasoning addition methods only consider the single factor of material temperature or cooking time, ignoring the synergistic effect of material thermal synchronization and oil film integrity, resulting in uneven seasoning adhesion and poor flavor stability, a seasoning addition decision mechanism under dual constraints is established by comprehensively considering the thermal synchronization state represented by the local minimum of phase volume and the topological integrity of the oil film represented by the number of structural pores. This ensures that the seasoning is added at the optimal time when the material surface temperature and evaporation state are maximally uniform and the oil film is relatively intact, so that the seasoning is evenly distributed and fully adsorbed on the material surface, improving the flavor stability and batch consistency of spicy chicken products, and optimizing the seasoning for spicy chicken products with different ingredients.

[0096] On the other hand, this embodiment also provides a seasoning optimization system for spicy chicken products based on different ingredients, which includes:

[0097] The stir-frying data module acquires an infrared image sequence of the stir-frying process of spicy chicken, performs dense optical flow calculation on the infrared image sequence, and constructs a thermal texture optical flow field.

[0098] The potential energy field construction module decomposes the thermal texture optical flow field to obtain an irrotational potential flow component and a divergent eddy current component, and constructs an evaporation potential energy field based on the irrotational potential flow component and the divergent eddy current component.

[0099] The phase volume calculation module extracts the surface temperature field of the infrared image sequence, constructs a two-dimensional phase space based on the evaporation potential energy field and the surface temperature field, maps the pixels on the material surface to the two-dimensional phase space to form a dynamic point cloud, and calculates the phase volume of the dynamic point cloud.

[0100] The seasoning optimization module tracks the time evolution curve of the phase volume and uses the local minimum value of the phase volume as the optimization target to determine the candidate time of thermal synchronization. It extracts the isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization, identifies holes in the connected domain enclosed by the isotherms, obtains the number of holes, and determines the seasoning addition time based on the number of holes and the candidate time of thermal synchronization, thereby realizing the seasoning optimization of spicy chicken products.

[0101] like Figure 3 As shown, if the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0102] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0103] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0104] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0105] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing the seasoning of spicy chicken products based on different ingredients, characterized in that, include: Infrared image sequences of the spicy chicken stir-frying process are obtained, and dense optical flow calculations are performed on the infrared image sequences to construct a thermal texture optical flow field; The thermal texture optical flow field is decomposed to obtain an irrotational potential flow component and a divergent eddy current component. An evaporation potential energy field is constructed based on the irrotational potential flow component and the divergent eddy current component. The surface temperature field of the infrared image sequence is extracted, and a two-dimensional phase space is constructed based on the evaporation potential energy field and the surface temperature field. The pixels on the material surface are mapped to the two-dimensional phase space to form a dynamic point cloud. The phase volume of the dynamic point cloud is calculated. The time evolution curve of the phase volume is tracked, and the local minimum value of the phase volume is used as the optimization target to determine the candidate time of thermal synchronization. The isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization are extracted, and the connected domains enclosed by the isotherms are identified to obtain the number of holes. The seasoning addition time is determined according to the number of holes and the candidate time of thermal synchronization to achieve the optimization of the seasoning of spicy chicken products.

2. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 1, characterized in that: Constructing the evaporation potential energy field based on the irrotational potential flow component and the divergent eddy current component includes performing Helmholtz decomposition on the thermal texture optical flow field and solving the Poisson equation to decompose the thermal texture optical flow field into an irrotational potential flow component and a divergent eddy current component. Calculate the curl intensity distribution of the non-dispersion eddy component, and construct a suppression mask based on the curl intensity distribution, wherein the curl intensity distribution represents the spatial influence range of the mechanical stirring action in the wok; The background flow field component is obtained by spatially weighting the non-dispersion eddy component using the suppression mask. A phase change diffusion component is constructed based on the irrotational potential flow component, and the phase change diffusion component is fused with the background flow field component to obtain the evaporation potential energy field. The feature fusion includes retaining the diffusion-type motion characteristics related to the outward evaporation of water in the phase change diffusion component, while using the background flow field component to correct the global displacement deviation caused by mechanical stirring.

3. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 2, characterized in that: Calculating the phase volume of the dynamic point cloud includes normalizing the evaporation potential energy field and the surface temperature field to obtain normalized evaporation potential energy value and normalized temperature value; and constructing a two-dimensional phase space with the normalized evaporation potential energy value as the first dimension and the normalized temperature value as the second dimension. The pixels on the surface of the material are traversed, the normalized evaporation potential energy value and normalized temperature value corresponding to each pixel are extracted and projected into the two-dimensional phase space to generate corresponding mapping points, and a dynamic point cloud is formed based on the set of the mapping points. The convex hull algorithm is used to extract the outer contour vertices of the dynamic point cloud, and the area of ​​the region enclosed by the outer contour vertices is calculated as the phase volume; the phase volume represents the degree of dispersion of the material in the thermodynamic state.

4. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 3, characterized in that: The determination of the candidate thermal synchronization moment includes collecting the phase volume on a continuous time series and constructing a time evolution curve of the phase volume; The time derivative of the time evolution curve is calculated, and the inflection point where the time derivative changes from a positive value to a negative value is identified to determine that the phase volume has entered the shrinkage stage. After the phase volume enters the shrinkage stage, the convergence rate of the phase volume is monitored, and the local minimum point of the phase volume is determined according to the convergence rate, and the corresponding time is recorded as the candidate time of thermal synchronization. The candidate time of thermal synchronization represents the moment when the overall distribution difference between the surface temperature and moisture evaporation state of different materials in the two-dimensional phase space is minimized.

5. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 4, characterized in that: Hole identification in the connected region enclosed by the isotherms includes, at the candidate thermal synchronization time, extracting the connected regions in the surface temperature field that are greater than a preset oil film formation temperature threshold as potential oil film regions. Topological analysis was performed on the potential oil film region to identify closed low-value regions within the connected regions as topological holes. Construct a survival state map of the topological pores over time to distinguish between transient pores and structural pores; count the number of structural pores as the total number of pores; wherein, transient pores represent random noise generated on the surface of the high-temperature oil film due to the rupture of boiling bubbles, and structural pores represent oil film defects caused by insufficient surface tension or obstruction of water evaporation airflow.

6. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 5, characterized in that: Constructing the survival status map of the topological holes in the time series includes tracking the survival status of topological holes at the same spatial location in multiple consecutive frames of images and calculating the life cycle duration of each topological hole. The lifecycle duration distribution of all topological holes at the current moment is statistically analyzed, and a lifecycle benchmark is calculated based on the statistical characteristics of the lifecycle duration distribution. Holes with a lifecycle duration less than or equal to the lifecycle benchmark are marked as transient holes, and holes with a lifecycle duration greater than the lifecycle benchmark are marked as structural holes. The lifecycle benchmark is adjusted according to the degree of oil temperature fluctuation during the frying process to eliminate the interference of boiling bubbles on the determination of oil film integrity.

7. The method for optimizing the seasoning of spicy chicken products based on different ingredients as described in claim 6, characterized in that: Determining the seasoning addition time based on the number of holes and the candidate thermal synchronization time includes real-time monitoring of the volume-time evolution curve and a verification phase based on the candidate thermal synchronization time. During the verification phase, the number of structural pores is monitored. When the number of structural pores is less than a preset pore threshold and the phase volume remains within the neighborhood of a local minimum, a seasoning dispensing instruction is generated. The seasoning dispensing instruction is configured to trigger a pulsed spray action, and the duration of the pulsed spray action matches the effective window period of the thermal synchronization candidate time.

8. A seasoning optimization system for spicy chicken products based on different ingredients, employing the method described in any one of claims 1-7, characterized in that: The stir-frying data module acquires an infrared image sequence of the stir-frying process of spicy chicken, performs dense optical flow calculation on the infrared image sequence, and constructs a thermal texture optical flow field. The potential energy field construction module decomposes the thermal texture optical flow field to obtain an irrotational potential flow component and a divergent eddy current component, and constructs an evaporation potential energy field based on the irrotational potential flow component and the divergent eddy current component. The phase volume calculation module extracts the surface temperature field of the infrared image sequence, constructs a two-dimensional phase space based on the evaporation potential energy field and the surface temperature field, maps the pixels on the material surface to the two-dimensional phase space to form a dynamic point cloud, and calculates the phase volume of the dynamic point cloud. The seasoning optimization module tracks the time evolution curve of the phase volume and uses the local minimum value of the phase volume as the optimization target to determine the candidate time of thermal synchronization. It extracts the isotherms in the surface temperature field corresponding to the candidate time of thermal synchronization, identifies holes in the connected domain enclosed by the isotherms, obtains the number of holes, and determines the seasoning addition time based on the number of holes and the candidate time of thermal synchronization, thereby realizing the seasoning optimization of spicy chicken products.

9. A computer device, comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.

Images