A cold storage evaporator frosting detection method and system based on image processing

Abstract

The present application belongs to the technical field of energy-saving control of cold storage refrigeration system, and particularly relates to a cold storage evaporator frosting detection method and system based on image processing, which comprises the following steps: building a hardware light path and obtaining an original gray image of the cold storage evaporator, and then performing pretreatment and partitioning; extracting micro-texture uniformity features and macro-structure complexity features of the surface texture area, and a physical gap filling rate of the gap filling area; constructing a light-heat coupling thermal resistance model, calculating a phase state evolution weight factor, and combining the equivalent heat transfer blockage index calculated based on the dominant features in different stages; and performing energy-saving defrosting control according to the blockage index and its change rate, and cooperating with the phase state weight. By introducing the phase state evolution weight, the present application realizes adaptive monitoring from frost phase diffuse reflection to ice phase specular reflection, effectively prevents defrosting delay, and improves the refrigeration energy efficiency of the cold storage refrigeration system.

Description

Technical Field

[0001] This invention relates to the field of energy-saving control technology for cold storage refrigeration systems. More specifically, this invention relates to a method and system for detecting frost buildup on cold storage evaporators based on image processing. Background Technology

[0002] During the operation of refrigeration facilities such as cold storage rooms, frost formation on the evaporator fins due to contact with warm, humid air is an unavoidable physical phenomenon. The continuous accumulation of frost significantly increases the heat exchange resistance of the fins and hinders the normal flow of cold air, leading to a decrease in evaporation temperature, an increase in compressor power consumption, and an overall reduction in the energy efficiency of the refrigeration system. Therefore, to achieve energy conservation and consumption reduction in cold storage rooms, it is necessary to monitor the frost formation on the evaporator fins in real time and perform defrosting operations as needed.

[0003] Currently, frost detection technology mainly relies on gradient-based edge detection or gray-level histogram statistics. Specifically, image gradient-based edge detection methods, such as the Sober operator, indirectly assess frost thickness by detecting changes in the sharpness of frost edges, or image gray-level statistical analysis methods directly reflect the degree of frost coverage by calculating the gray-level mean, variance, or binarized pixel ratio of the entire image or a specific region.

[0004] However, in the complex and ever-changing working conditions of actual cold storage, the frosting process is not a simple linear thickening, but a physical phase evolution process that includes condensation, frost crystal growth, frost layer compaction, and finally the formation of an ice shell. In the early stage of frost crystal growth, the rough surface leads to enhanced diffuse reflection and obvious gradient characteristics. However, in the later stage of frosting, after the frost layer is compacted into a dense ice shell, its surface becomes smooth, causing specular reflection. This change in physical properties from rough to smooth can cause feature reversal in gradient-based edge detection algorithms, that is, the gradient characteristics weaken when the ice layer forms, which can easily be misjudged as a state without frost or with thin frost, thus causing serious delays in defrosting operations.

[0005] Secondly, relying solely on pixel ratio to assess the degree of frosting is essentially a linear or quasi-linear evaluation model. According to the basic principles of fluid mechanics, when airflow passes through the gaps between evaporator fins, the air resistance increases exponentially with the blockage rate. Therefore, using linear indicators to assess the impact of frosting on duct blockage will fail to accurately reflect the actual rate of deterioration in heat exchange efficiency, leading to inaccurate control of defrosting timing or duration. Either defrosting too early or too long will result in energy waste, or defrosting too late or insufficiently will lead to persistently low cooling efficiency.

[0006] In summary, existing image processing-based frost detection technologies often face challenges such as poor detection stability and insufficient control precision in practical applications because they fail to effectively overcome the feature reversal problem caused by the phase transition between frost and ice in the early and late stages of frost formation, and lack modeling of the nonlinear growth characteristics of wind resistance. This restricts the further improvement of energy efficiency in cold storage refrigeration systems. Summary of the Invention

[0007] To address the technical problems in existing technologies, such as misjudgment of feature reversal due to the smoothness of the ice surface and the inability of linear indices to accurately reflect nonlinear wind resistance changes, resulting in inaccurate defrosting control, this invention provides solutions in the following aspects.

[0008] In a first aspect, the present invention provides a method for detecting frost on a cold storage evaporator based on image processing, comprising: constructing a hardware optical path and acquiring an original grayscale image of the cold storage evaporator; performing contrast-limited adaptive histogram equalization on the original grayscale image; and segmenting the image into a surface texture region and a gap-filling region using a preset mask; analyzing the surface texture region to extract micro-texture uniformity features and macro-structural complexity features, and simultaneously calculating the physical gap-filling rate of the gap-filling region; constructing a photothermal coupling thermal resistance model; calculating a phase evolution weighting factor based on the mean high light intensity, mean grayscale, and gradient standard deviation; using the phase evolution weighting factor to weight and fuse the micro-texture uniformity features, the macro-structural complexity features, and the physical gap-filling rate to obtain an equivalent heat transfer blockage index; monitoring the change of the equivalent heat transfer blockage index in real time; determining the current frost or ice state in conjunction with the phase evolution weighting factor; and generating a defrost start command or a defrost termination command accordingly.

[0009] This invention constructs a photothermal coupling thermal resistance model and calculates a phase evolution weighting factor based on the mean high light intensity, mean gray level, and gradient standard deviation. This phase evolution weighting factor can keenly capture the optical phase transition from diffuse reflection to specular reflection in ice phases. Logically, it forces the algorithm to adaptively adjust feature dependencies. Using this phase evolution weighting factor, the micro-texture uniformity features, macro-structural complexity features, and physical gap filling rate are weighted and fused to derive an equivalent heat transfer blockage index. This derivation process achieves an adaptive switch from frost-dependent surface texture to ice-dependent physical blockage rate, overcoming the feature inversion blind zone. The final index can monotonically increase to map the nonlinear surge of thermal resistance. Combined with the phase evolution weighting factor, start-stop operations are executed, thereby effectively preventing defrosting delays and making defrosting control more precise and energy-efficient.

[0010] Preferably, the step of setting up the hardware optical path and acquiring the original grayscale image of the cold storage evaporator includes: installing a camera behind the return air inlet of the evaporator, making the optical axis of the camera lens form an angle with the fin plane, and configuring a side light source to provide grazing light, so that during the frost-free period, the light undergoes specular reflection on the metal surface and moves away from the lens, during the frosting period, it undergoes diffuse reflection, and during the ice-forming period, it undergoes irregular specular reflection.

[0011] This invention utilizes the different reflection characteristics of side-grazing light on the surfaces of metal, frost, and ice to provide a clear characteristic basis for distinguishing dark metal, bright frost, and highly reflective ice at the physical optical path level, effectively suppressing the interference of ambient light and cold fog in cold storage.

[0012] Preferably, the microtexture uniformity feature is obtained by calculating the inverse difference moment of the gray-level co-occurrence matrix, and the calculation of the inverse difference moment satisfies the following relationship: ;

[0013] in, Indicates the inverse moment. The number of elements in the normalized gray-level co-occurrence matrix represents the number of elements in the normalized gray-level co- Line number Column elements, This indicates a row-level grayscale index. This indicates a column grayscale index.

[0014] Preferably, the extraction process of the macroscopic structural complexity feature includes: performing local binarization processing on the surface texture area to extract the texture skeleton; using box counting, covering the texture skeleton with boxes of different sizes and counting the number of boxes; fitting the slope between the logarithm of the number of boxes and the logarithm of the inverse of the box size through linear regression to obtain the fractal dimension; and using the fractal dimension as the macroscopic structural complexity feature.

[0015] Preferably, the calculation of the physical gap fill rate includes: calculating the ratio of the number of pixels with gray values ​​greater than the adaptive binarization threshold in the gap fill area to the total number of pixels in the gap fill area, and using the ratio as the physical gap fill rate.

[0016] Preferably, the phase evolution weighting factor satisfies the following relationship:

[0017] ;

[0018] in, This represents the phase evolution weighting factor. Represents the sensitivity coefficient. This represents the average highlight intensity of the surface texture area. This represents the average grayscale value of the surface texture area. This represents the standard deviation of the gradient in the surface texture region. To prevent division by zero constant, This represents the phase threshold.

[0019] This invention, by calculating the ratio of the difference between the specular highlight and the mean to the standard deviation of the gradient, can keenly capture the physical phase transition characteristics of a material from a rough diffuse reflection stage to a smooth specular reflection stage. Furthermore, it utilizes the Sigmoid function to map these physical phase transition characteristics to weights ranging from 0 to 1, providing an absolutely reliable mathematical basis for the logical fusion of multidimensional physical features.

[0020] Preferably, the calculation of the equivalent heat transfer blockage index satisfies the following relationship:

[0021] ;

[0022] in, Indicates the equivalent heat blockage index. This indicates the complexity characteristics of the macroscopic structure. Indicates the uniformity characteristics of microtexture. Indicates the physical gap filling rate. This represents the wind resistance growth index. Indicates the dimensional alignment factor. To prevent the removal of the zero constant.

[0023] This invention seamlessly integrates phase weights into the computational flow. During the frost period, it primarily utilizes fractal dimension and texture features to characterize thermal resistance, while during the ice period, it switches to an exponential gap-filling model that conforms to the laws of fluid mechanics. This accurately reflects the objective physical fact that the pressure drop increases exponentially due to the reduction in cross-sectional area, significantly improving the stability of defrosting under extreme conditions.

[0024] Preferably, the conditions for generating the defrost start command include any one of the following:

[0025] When the equivalent heat transfer blockage index is greater than the preset high-level blockage threshold and continues for a preset time;

[0026] When the time change rate of the equivalent heat transfer blockage index is greater than the preset growth rate threshold, and the phase evolution weight factor is greater than the first preset phase threshold.

[0027] Preferably, the defrost termination command is generated under the condition that the following conditions are met simultaneously:

[0028] The equivalent heat transfer blockage index is less than the preset low-level recovery threshold.

[0029] The phase evolution weighting factor is less than the second preset phase threshold.

[0030] This invention comprehensively judges the rate of change of the absolute value of the equivalent heat transfer blockage index and the phase evolution weighting factor, which not only prevents false start-ups caused by occasional data fluctuations, but also ensures rapid intervention under severe icing conditions. At the same time, it terminates defrosting by verifying the fading of the high-reflectivity phase, effectively avoiding secondary icing and energy waste caused by incomplete ice melting.

[0031] Secondly, the present invention provides a cold storage evaporator frost detection system based on image processing, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned cold storage evaporator frost detection method based on image processing is implemented.

[0032] By adopting the above technical solution, a computer program is generated from the above-mentioned image processing-based cold storage evaporator frost detection method and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0033] The beneficial effects of this invention are as follows:

[0034] This invention effectively solves the technical challenge of feature inversion in traditional visual inspection by deeply integrating the lateral grazing optical path with the photothermal coupling thermal resistance model. It utilizes the difference in optical reflection from the rough frost phase to the smooth ice phase to construct a phase evolution weight factor, enabling the algorithm to automatically reduce the weight of ineffective texture features when ice layers are generated, and instead rely on the physical fill rate, thus ensuring extremely high sensitivity and accuracy of state recognition throughout the entire phase cycle.

[0035] Furthermore, it breaks through the limitations of traditional methods that rely on single linear pixel statistics to evaluate the degree of frost. It constructs an exponential growth model that conforms to the laws of fluid mechanics to characterize the change in thermal resistance. By smoothly and adaptively switching the equivalent heat transfer blockage index at different frost stages, it truly restores the phenomenon of nonlinear wind resistance surge, thereby guiding the defrosting equipment to start and stop precisely at the optimal time. This avoids defrosting delays and eliminates redundant energy consumption, significantly improving the operational stability and overall energy efficiency of the cold storage refrigeration system. Attached Figure Description

[0036] Figure 1 This is a flowchart of a cold storage evaporator frost detection method based on image processing according to the present invention;

[0037] Figure 2 This is a comparison curve of the detection effects of the present invention and the prior art throughout the entire frosting cycle;

[0038] Figure 3 This is a logic curve diagram showing the change of each parameter over time in the phase evolution weighted model according to the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0041] This invention discloses an image processing-based method for detecting frost buildup on cold storage evaporators, referring to... Figure 1 This includes steps S1-S4:

[0042] S1. Build the hardware optical path and obtain the original grayscale image of the cold storage evaporator. Perform contrast-limited adaptive histogram equalization on the original grayscale image and use a preset mask to segment the image into surface texture area and gap filling area.

[0043] In an optional embodiment, to establish a physical optical path environment that reflects the phase properties of matter and to preprocess the data, a camera is mounted on the side and rear of the evaporator return air inlet. The bracket is adjusted so that the optical axis of the camera lens forms an angle of 30 to 45 degrees with the fin plane. At the same time, a side-facing low-angle strip light source is configured to provide grazing light with a color temperature of 6500K, so that the light illuminates the fin surface at a grazing angle. During the frost-free period, due to the smooth surface of the metal fins, the grazing light undergoes specular reflection on the metal surface. The reflected light moves away from the lens along the reflection angle, and the imaging is mainly diffuse reflection, resulting in a relatively dark overall image. During the frosting period, the growth of frost crystals causes the surface to become rough, resulting in strong diffuse reflection. Some light enters the lens, increasing the image brightness and enriching the texture. During the freezing period, the frost layer compacts into ice, and the surface becomes smooth but irregular again, forming a bright specular reflection spot at a specific angle, creating a strong contrast with the surrounding dark areas.

[0044] After acquiring the original grayscale image of the cold storage evaporator, a contrast-limited adaptive histogram equalization process is applied to the original grayscale image to suppress the overall low-frequency grayscale drift caused by cold fog in the cold storage and enhance local contrast. Subsequently, a preset mask is used to segment the image into a surface texture area and a gap filling area. The surface texture area corresponds to the metal plane area of ​​the fins and is used to analyze the changes in texture details; the gap filling area corresponds to the airflow channels between the fins and is used to calculate the degree of physical blockage.

[0045] Thus, through specific optical path design and image partitioning processing, environmental interference can be effectively suppressed, and a physical optical feature basis can be provided for distinguishing frost and ice, effectively solving the problem that ice layers are difficult to identify under ordinary optical paths.

[0046] S2. Analyze the surface texture area, extract the micro-texture uniformity features and macro-structure complexity features, and calculate the physical gap filling rate of the gap filling area.

[0047] In an optional embodiment, the microtexture uniformity feature is obtained by calculating the inverse moment of the gray-level co-occurrence matrix, and the formula for calculating the inverse moment is: ;

[0048] in, Indicates the inverse moment. The number of elements in the normalized gray-level co-occurrence matrix represents the number of elements in the normalized gray-level co- Line number Column elements, This indicates a row-level grayscale index. This indicates a column grayscale index.

[0049] For example, suppose in a simplified 3×3 local image patch, when the grayscale changes drastically, the matrix elements are scattered far from the diagonal. ,at this time The denominator is This contribution is The calculated inverse moment value is low, indicating that the frost layer is thicker and rougher; the grayscale change is gradual, and the matrix elements are concentrated near the diagonal. ,at this time The denominator is This contribution is The calculated inverse moment value is high, indicating that there is no frost or that the ice has formed and the surface is smooth.

[0050] Secondly, macroscopic structural complexity features are extracted. Local binarization is performed on the surface texture area to extract the texture skeleton. Box counting is used, covering the texture skeleton with boxes of different sizes, and the number of boxes is counted. The fractal dimension is obtained by fitting the slope between the logarithm of the number of boxes and the logarithm of the inverse of the box size through linear regression. This fractal dimension is then used as the macroscopic structural complexity feature; a larger value indicates a more complex surface morphology.

[0051] Finally, the ratio of the number of pixels with gray values ​​greater than the adaptive binarization threshold in the gap filling area to the total number of pixels in the gap filling area is calculated, and this ratio is used as the physical gap fill rate. For example, if there are 10,000 pixels in the gap filling area, the adaptive binarization threshold is set to 150, and the number of pixels with gray values ​​greater than the adaptive binarization threshold in the gap filling area is 6,000, the calculated physical gap fill rate is 0.6. The value of this physical gap fill rate ranges from 0 to 1, directly reflecting the blockage ratio of the physical channel.

[0052] In this way, through multi-dimensional feature extraction, we can comprehensively capture the texture changes and physical blockages during the frosting process, providing rich data support for subsequent model calculations.

[0053] S3. Construct a photothermal coupling thermal resistance model. Calculate the phase evolution weighting factor based on the mean high light intensity, mean gray level, and gradient standard deviation. Use the phase evolution weighting factor to weight and fuse the micro-texture uniformity characteristics, macro-structure complexity characteristics, and physical gap filling rate to obtain the equivalent heat transfer blockage index.

[0054] In an optional embodiment, the phase evolution weighting factor is calculated using the following formula:

[0055] ;

[0056] in, This represents the phase evolution weighting factor. Represents the sensitivity coefficient. This represents the average highlight intensity of the top 5% of grayscale values ​​in the surface texture area. This represents the average grayscale value of the surface texture area. This represents the standard deviation of the gradient in the surface texture region. To prevent division by zero constant, This represents the phase threshold.

[0057] For example, setting the sensitivity coefficient The value is 2, and the zero constant is excluded. The phase threshold is 1. With a value of 10, and all grayscale values ​​ranging from 0 to 255, when in the diffuse reflection stage, the image is generally bright but uniform, with no obvious highlights and a rough texture. The average highlight intensity is measured. The grayscale average is 140. 120, gradient standard deviation The value is 24. Substituting this into the core term, we get... Calculate the exponential term: Calculate the phase evolution weighting factor: It was determined to be a complete frost phase.

[0058] During the specular reflection stage, the image exhibits locally bright spots, while the rest of the image is darkened due to transmission, and the surface is smooth. The average intensity of the highlights is measured. The grayscale average is 250. 100, gradient standard deviation The value is 4. Substituting this into the core term, we get... Calculate the exponential term: Calculate the phase evolution weighting factor: It was determined to be a complete glacial phase.

[0059] Next, the equivalent heat transfer blockage index is calculated, and the formula is:

[0060] ;

[0061] in, Indicates the equivalent heat blockage index. This indicates the complexity characteristics of the macroscopic structure. Indicates the uniformity characteristics of microtexture. Indicates the physical gap filling rate. This represents the wind resistance growth index. Indicates the dimensional alignment factor. To prevent the removal of the zero constant.

[0062] For example, setting the drag growth index The dimension alignment factor is 2.5. The value is 1.5, excluding zero constant. When the phase evolution weighting factor is 0, the macroscopic structural complexity characteristics are measured. The microtexture uniformity characteristic is 1.5. The value is 0.3. Substituting this into the formula, we get... At this point, the equivalent heat transfer blockage index mainly reflects the texture characteristics.

[0063] When the phase evolution weighting factor is 1, the physical gap filling rate is measured. The value is 0.8. Substituting this into the relational expression, we get... At this point, the equivalent heat transfer blockage index increases exponentially, truly reflecting the surge in wind resistance caused by severe blockage.

[0064] Thus, by using the photothermal coupling thermal resistance model, texture features and physical filling features were successfully combined, and a smooth switching was achieved by using the phase evolution weighting factor, ensuring the accuracy of the evaluation index.

[0065] Reference Figure 2 The curves visually demonstrate that the equivalent heat transfer blockage index of the present invention continues to increase monotonically after the characteristic inversion point, thus solving the misjudgment problem caused by the smoothness of the ice layer. The attached figure is a schematic curve after normalization.

[0066] S4. Monitor the changes in the equivalent heat transfer blockage index in real time, combine the phase evolution weighting factor to determine the current frosting or icing state, and generate a defrosting start command or a defrosting termination command accordingly.

[0067] In an optional embodiment, if the equivalent heat transfer blockage index is detected to be greater than a preset high-level blockage threshold and lasts for a preset time, a defrost start command is generated. For example, the high-level blockage threshold is set to 5. If the equivalent heat transfer blockage index is detected to reach 5.2 and lasts for more than 10 minutes, a defrost start command is generated. Alternatively, when the time change rate of the equivalent heat transfer blockage index is greater than a preset growth rate threshold and the phase evolution weight factor is greater than a first preset phase threshold, for example, the growth rate threshold is set to 0.5 / min and the first preset phase threshold is 0.8. If the equivalent heat transfer blockage index suddenly increases from 3 to 4 within 1 minute and the phase evolution weight factor is 0.9, it is determined that rapid freezing has occurred, and a defrost start command is generated immediately.

[0068] The defrost termination command is generated only when the equivalent heat transfer blockage index is less than a preset low-level recovery threshold and the phase evolution weight factor is less than a second preset phase threshold. For example, the low-level recovery threshold is set to 0.5 and the second preset phase threshold is set to 0.2. During the defrosting process, the defrost termination command is generated only when the equivalent heat transfer blockage index is less than 0.5 and the phase evolution weight factor is less than 0.2, so as to prevent the defrost from stopping prematurely due to the remaining ice shell on the surface.

[0069] In this way, by comprehensively judging the blockage index and phase weight, the problem of stopping defrosting before it is completely defrosted is avoided, and the energy waste caused by premature defrosting is also avoided, thus achieving precise energy-saving control.

[0070] Reference Figure 3 This demonstrates the logic of the weight switching of the phase evolution weight factor as the frost phase evolves, reflecting the smooth transition of the refrigeration system between the frost and ice periods.

[0071] This invention also discloses an image processing-based frost detection system for cold storage evaporators, comprising a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an image processing-based frost detection method for cold storage evaporators according to the present invention.

[0072] The aforementioned image processing-based frost detection system for cold storage evaporators also includes other components well-known to those skilled in the art, such as communication buses and communication interfaces. Their setup and functions are known in the art and will not be described in detail here.

[0073] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A method for detecting frost formation on a cold storage evaporator based on image processing, characterized in that, include: A hardware optical path is constructed and the original grayscale image of the cold storage evaporator is obtained. The original grayscale image is subjected to contrast-limited adaptive histogram equalization processing, and the image is divided into surface texture area and gap filling area using a preset mask. The surface texture area is analyzed to extract micro-texture uniformity features and macro-structural complexity features, and the physical gap filling rate of the gap filling area is calculated. A photothermal coupling thermal resistance model is constructed. A phase evolution weighting factor is calculated based on the mean high light intensity, mean grayscale, and gradient standard deviation to determine the current frosting or icing state. This phase evolution weighting factor is then used to weight and fuse the micro-texture uniformity feature, the macro-structural complexity feature, and the physical gap filling rate to obtain the equivalent heat transfer blockage index, satisfying the following relationship: ； in, Indicates the equivalent heat blockage index. This indicates the complexity characteristics of the macroscopic structure. Indicates the uniformity characteristics of microtexture. Indicates the physical gap filling rate. This represents the wind resistance growth index. Indicates the dimensional alignment factor. To prevent division by zero constant; The system monitors the changes in the equivalent heat transfer blockage index in real time, combines the phase evolution weighting factor to determine the current frosting or icing state, and generates a defrosting start command or a defrosting termination command accordingly.

2. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The process of establishing the hardware optical path and acquiring the original grayscale image of the cold storage evaporator includes: The camera is installed on the side and rear of the evaporator return air inlet, with the camera lens optical axis at an angle to the fin plane. A side light source is configured to provide grazing light, so that during the frost-free period, the light is specularly reflected off the metal surface and away from the lens, during the frosting period, it is diffusely reflected, and during the ice-forming period, it is irregularly specularly reflected.

3. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The microtexture uniformity feature is obtained by calculating the inverse difference moment of the gray-level co-occurrence matrix, and the calculation of the inverse difference moment satisfies the following relationship: ； in, Indicates the inverse moment. The number of elements in the normalized gray-level co-occurrence matrix represents the number of elements in the normalized gray-level co- Line number Column elements, This indicates a row-level grayscale index. This indicates a column grayscale index.

4. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The extraction process of the macroscopic structural complexity feature includes: performing local binarization on the surface texture area to extract the texture skeleton; using box counting, covering the texture skeleton with boxes of different sizes and counting the number of boxes; fitting the slope between the logarithm of the number of boxes and the logarithm of the inverse of the box size through linear regression to obtain the fractal dimension; and using the fractal dimension as the macroscopic structural complexity feature.

5. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The calculation of the physical gap fill rate includes: calculating the ratio of the number of pixels with gray values ​​greater than the adaptive binarization threshold in the gap fill area to the total number of pixels in the gap fill area, and using the ratio as the physical gap fill rate.

6. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The phase evolution weighting factor satisfies the following relationship: ； in, This represents the phase evolution weighting factor. Represents the sensitivity coefficient. This represents the average highlight intensity of the surface texture area. This represents the average grayscale value of the surface texture area. This represents the standard deviation of the gradient in the surface texture region. To prevent division by zero constant, This represents the phase threshold.

7. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The conditions for generating the defrost start command include any of the following: When the equivalent heat transfer blockage index is greater than the preset high-level blockage threshold and continues for a preset time; When the time change rate of the equivalent heat transfer blockage index is greater than the preset growth rate threshold, and the phase evolution weight factor is greater than the first preset phase threshold.

8. The image processing-based method for detecting frost buildup on a cold storage evaporator according to claim 1, characterized in that, The defrost termination command is generated when the following conditions are met simultaneously: The equivalent heat transfer blockage index is less than the preset low-level recovery threshold. The phase evolution weighting factor is less than the second preset phase threshold.

9. A cold storage evaporator frost detection system based on image processing, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for detecting frost on a cold storage evaporator based on image processing, as described in any one of claims 1-8.

Images