Fruit and vegetable freshness preservation method based on dry fog of slightly acidic electrolyzed water

Abstract

The present invention discloses a method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water, which relates to the technical field of fruit and vegetable preservation, and includes the following steps: atomizing the slightly acidic electrolyzed water to form dry fog within a set starting particle size range and acting on the surface of fruits and vegetables during the transmission on a high-speed assembly line to form a wide-area bactericidal protection fog layer; collecting multi-spectral image information of the outer surface of the fruits and vegetables through a multi-spectral image acquisition device during the passage of the fruits and vegetables through the assembly line; performing processing operations on the collected original images, and at the same time matching the image frames with the positions of the fruits and vegetables on the assembly line according to the assembly line speed and imaging frequency. The present invention analyzes the severity of fruit and vegetable cracks through multi-spectral image recognition and convolutional neural network, and adjusts the particle size of the dry fog in real time to achieve differential preservation treatment, improves the adhesion of fog droplets in the crack area to the bactericidal effect, effectively makes up for the problem of uneven coverage of traditional fixed particle size dry fog, and significantly improves the preservation accuracy and anti-pollution ability.

Description

Technical Field

[0001] The present invention relates to the technical field of fruit and vegetable preservation, and particularly to a method for preserving fruits and vegetables based on microacidic electrolyzed water dry fog. Background Art

[0002] Using microacidic electrolyzed water dry fog for fruit and vegetable preservation means atomizing electrolyzed water with bactericidal and disinfectant effects and weak acidity (usually containing low-concentration hypochlorous acid) into extremely fine dry fog particles, and spraying or soaking fruits and vegetables in a specific environment. Since the dry fog particles are extremely small and evenly distributed, they can quickly penetrate into the gaps or folds on the surface of fruits and vegetables without increasing the moisture burden on the surface of fruits and vegetables, effectively inhibiting the growth of microorganisms such as bacteria and fungi, thereby slowing down spoilage and extending the preservation time. In addition, microacidic electrolyzed water is usually relatively safe and mild, which can reduce the impact on the quality and taste of fruits and vegetables, and achieve a more environmentally friendly and efficient preservation effect.

[0003] During the processes of sorting, grading, weighing, etc. on a high-speed fruit and vegetable production line, microacidic electrolyzed water dry fog is often synchronously used for preservation to form a bactericidal protective layer on the surface of fruits and vegetables, inhibiting microbial contamination introduced by mechanical sorting, collisions, manual contact and other links. However, the prior art generally uses ultrafine dry fog with a fixed particle size (such as 2 - 5 μm) for spraying, lacking the ability to adaptively regulate the surface state of fruits and vegetables. Especially when fruits and vegetables have serious fine cracks due to factors such as moisture fluctuations and mechanical extrusion, the ultrafine dry fog is difficult to settle and penetrate the crack gaps due to its small particle size, poor inertia and large disturbance, resulting in insufficient adhesion in the crack area and incomplete formation of the liquid film, and a significant decline in the bactericidal effect. There are problems such as uneven preservation coverage, weak protection ability for minimally invasive wounds, and easy cross-infection, seriously restricting the stability and adaptability of this preservation method in a complex dynamic environment.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to provide a method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water. By introducing multi-spectral image acquisition and intelligent image recognition technologies, combined with a convolutional neural network model, the severity of micro-cracks on the surface of fruits and vegetables is accurately identified and quantitatively scored. Based on the scoring results, the particle size of the dry fog of slightly acidic electrolyzed water is adjusted in real time, realizing differential and dynamic preservation treatment of fruits and vegetables in different crack states, significantly improving the sedimentation and adhesion ability of the dry fog in the crack area and the liquid film formation efficiency, strengthening the local sterilization effect, effectively making up for the technical defects of uneven coverage and insufficient preservation of traditional dry fog with a fixed particle size in the minimally invasive wound area, and significantly improving the preservation accuracy and anti-pollution ability of fruits and vegetables under a high-speed processing production line, so as to solve the problems in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: A method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water, comprising the following steps:

[0007] Atomize slightly acidic electrolyzed water into dry fog according to a set initial particle size range through an atomization control unit, and act on the surface of fruits and vegetables during the transmission on a high-speed production line to form a wide-area sterilization and protection fog layer, realizing preliminary sterilization, disinfection and preservation treatment of the overall surface of fruits and vegetables;

[0008] Collect multi-spectral image information of the outer surface of fruits and vegetables in a non-contact manner through a multi-spectral image acquisition device during the passage of fruits and vegetables through the production line;

[0009] The original image data obtained from the multi-spectral imaging acquisition device usually contains interference factors such as noise and illumination changes. In order to obtain stable recognition accuracy, it is necessary to perform preprocessing operations such as denoising, normalization, brightness correction, and background removal on the image data. At the same time, match the image frames with the positions of fruits and vegetables on the production line according to the production line speed and imaging frequency to ensure that the preprocessing results accurately correspond to the actual spatial positioning of fruits and vegetables;

[0010] Identify and locate the micro-crack area based on the preprocessed image data. For the micro-crack area, extract the crack morphological features reflecting the severity of micro-cracks on the surface of fruits and vegetables. After deeply analyzing the extracted features through feature engineering technology, use the analyzed features as multi-dimensional feature vectors to characterize the severity of micro-cracks on the surface of fruits and vegetables;

[0011] Input the feature vector into a convolutional neural network model that has been pre-trained offline and deployed on-site, and score the severity of the cracks according to the morphological features of the cracks;

[0012] According to the severity of the cracks output by the convolutional neural network, the working parameters of the atomizing nozzle are adjusted in real time to dynamically change the size of the dry fog particles. If severe cracks are detected on the surface of the fruits and vegetables, the droplet size is automatically increased to enhance the sedimentation ability and adhesion of the droplets, ensuring that they can effectively enter the crack gaps and adhere to the inside of the cracks to form a liquid film, thereby strengthening the local sterilization and preservation effects. In the case of minor or non-existent cracks, a smaller particle size (such as 2 - 5 μm) is maintained to increase the coverage area and reduce water consumption.

[0013] Preferably, the multi-spectral image acquisition device is used to collect the multi-spectral image information of the outer surface of the fruits and vegetables, which specifically includes the following steps:

[0014] First, install the multi-spectral image acquisition device at an appropriate position on the fruit and vegetable processing production line to ensure that the viewing angle covers the area where the fruits and vegetables pass through;

[0015] Second, when the fruits and vegetables pass through this acquisition area, trigger the sensor to synchronously start the multi-spectral imaging device to continuously scan the surface of the fruits and vegetables;

[0016] Then, use the multi-channel light source to irradiate the surface of the fruits and vegetables frame by frame according to the preset wavelength bands (such as visible light, near-infrared, short-wave infrared, etc.), and at the same time, the imaging device synchronously collects the reflected images of each wavelength band;

[0017] Subsequently, send the collected multi-spectral image data to the data processing unit to provide high-quality and multi-dimensional visual information support for subsequent analysis steps such as image preprocessing, feature extraction, and crack recognition.

[0018] Preferably, the image frames are matched with the positions of the fruits and vegetables on the production line according to the production line speed and imaging frequency, which generally includes the following steps:

[0019] First, install a high-precision encoder or position sensor at a key position on the production line to detect the running speed of the conveyor belt and the time nodes when the fruits and vegetables pass through in real time;

[0020] Second, synchronize the sampling frequency of the imaging system with the speed data of the sensor so that each frame of image corresponds to the position of the fruits and vegetables at a specific time point;

[0021] Then, calculate the theoretical spatial position of the fruits and vegetables corresponding to each frame of image on the conveyor belt based on the linear speed of the conveyor belt and the image acquisition time interval;

[0022] Then, combine the fruit and vegetable entry trigger signal or the fruit and vegetable contour information detected in the image to fine-tune and calibrate the actual position;

[0023] Finally, bind the image frame number with the fruit and vegetable identification information to establish a one-to-one correspondence, ensuring that each image can be accurately corresponded to the actual fruit and vegetable individual in the preprocessing, feature extraction, and subsequent control processes, and realizing the synchronous registration of image processing and physical position.

[0024] Preferably, on the basis of the preprocessed image data, identify and locate the microcrack region, which can be carried out by combining multi-spectral image enhancement and morphological analysis. First, enhance the contrast in spectral response between the crack region and the normal fruit and vegetable epidermis through band fusion or band difference, making the microcracks more prominent in a specific band combination; then, use edge detection algorithms (such as Canny or Laplacian operators) to extract the suspected crack edges, and then combine image segmentation techniques (such as region growing, threshold segmentation, or superpixel-based clustering) to locate the preliminary crack region; then, optimize the crack shape through morphological operations (such as erosion, dilation, thinning), remove artifacts and redundant noise; finally, further screen out the target regions that truly conform to the microcrack characteristics by combining geometric features and spatial positions, realizing high-precision identification and location.

[0025] Preferably, for the microcrack region, extract the crack morphological features that reflect the severity of the microcracks on the fruit and vegetable surface. Among them, the extracted crack morphological features include the dispersion degree of the cracks in the spatial space of the fruit and vegetable surface and the reflectance change of the crack region in different spectral bands. After deeply analyzing the dispersion degree of the cracks in the spatial space of the fruit and vegetable surface and the reflectance change of the crack region in different spectral bands through feature engineering techniques, generate the crack distribution dispersion index and the crack penetration index respectively. Take the crack distribution dispersion index and the crack penetration index as multi-dimensional feature vectors to characterize the severity of the microcracks on the fruit and vegetable surface.

[0026] Preferably, input the feature vector constructed by the crack distribution dispersion index and the crack penetration index into the convolutional neural network model that has been pre-trained offline and deployed on-site, generate the crack structure abnormality degree based on the convolutional neural network model, and score the severity of the cracks through the crack structure abnormality degree.

[0027] Preferably, the specific steps for deeply analyzing the dispersion degree of the cracks in the spatial space of the fruit and vegetable surface through feature engineering techniques to generate the crack distribution dispersion index are as follows:

[0028] First, for all crack regions, mark the center point coordinates of each crack to construct a crack spatial position set , where N represents the number of cracks detected in the image, is the geometric center coordinate of the i-th crack region. On this basis, construct a crack distance map , where, , which represents the Euclidean distance between the geometric centers of crack i and crack j, and is used to measure the spatial dispersion degree between cracks;

[0029] According to the crack spatial position set and the crack distance diagram, the maximum connection dissipation ratio is introduced to calculate the crack distribution dispersion index, and the calculation expression is:

[0030]

[0031] where is the crack distribution dispersion index, is a very small positive number to prevent the denominator from being zero; is the farthest distance between crack i and other cracks, representing its most extreme dispersion trend; is the reciprocal sum of the distances between crack i and other cracks, measuring its overall connection tightness; the crack distribution dispersion index combines the dual characteristics of local extreme value diffusion and global sparsity. The larger it is, the more dispersed the crack distribution is, the more extensive the damage to the fruit and vegetable surface is, and the higher the severity of the cracks. It is applicable to the serious damage recognition scenarios with complex and extensive micro-cracks.

[0032] Damage recognition scenario.

[0033] Preferably, the specific steps to generate the crack penetration index by deeply analyzing the reflectivity change of the crack area in different spectral bands through feature engineering technology are as follows:

[0034] First, in the preprocessed multi-spectral image, the pixel-level reflectivity values of the crack area in multiple specific bands are extracted and denoted as , where represents the pixel position, is the k-th band. Next, the local reflectivity gradient vector of the crack area in the band dimension is defined as:

[0035]

[0036] where is the local reflectivity gradient vector at the pixel position , which reflects the reflectivity change trend of this point in multiple spectral bands. This vector describes the reflectivity change rate of the crack area between different bands (similar to the slope feature of the spectral curve), and n represents the total number of bands;

[0037] After obtaining the local reflectivity gradient vector of the crack area, the crack penetration index is calculated based on the tensor projection intensity of the local reflectivity gradient vector in the main extension direction of the crack, which is used to measure the consistency degree between the local spectral change trend and the main direction of the crack structure, and characterize the penetration depth and development severity of the crack in the tissue structure. Let the main extension direction vector of the crack be , the calculation expression of the crack penetration index in the crack area is as follows:

[0038]

[0039] Where: is the crack penetration index, represents the spatial range of the crack area, is the projection value of the local reflectivity gradient vector in the main crack direction, is the power factor for enhancing the change in crack penetration, is the crack edge sharpness factor (such as the weight map with enhanced second-order derivative of the edge) for emphasizing the area with more severe rupture.

[0040] Preferably, according to the crack severity output by the convolutional neural network, the working parameters of the atomizing nozzle are adjusted in real time to dynamically change the dry fog particle size. The specific steps are as follows:

[0041] After obtaining the crack structure abnormality degree, first judge whether the difference between it and the crack structure abnormality reference threshold is positive, that is, compare whether the current crack severity on the fruit and vegetable surface exceeds the preset benchmark. For this purpose, a difference function is defined, and the expression of the difference function is:

[0042]

[0043] Where, is the crack structure abnormality degree output by the convolutional neural network, and the larger the value, the more severe the crack; is the crack structure abnormality reference threshold for measuring whether the crack reaches or exceeds the acceptable limit; is the difference between the crack structure abnormality degree and the crack structure abnormality reference threshold, which is a temporary judgment quantity for determining whether to increase the particle size. If , it means that the current crack condition is still within the safe range; if , it indicates that the crack abnormality degree exceeds the benchmark and the fog droplet size needs to be dynamically increased in the subsequent steps;

[0044] If , the default particle size within the initial particle size range of the atomizing nozzle is maintained;

[0045] If , it is necessary to amplify on the basis of this initial particle size range, make the fog droplet size increase moderately and penetrate the crack gap. The expression for amplifying the initial particle size range is:

[0046]

[0047] Where is the target fog droplet size, and The minimum and maximum allowable droplet sizes are designed for the system to limit the extreme values of the final particle size, ensuring the safety of the nozzle hardware and the stability of the spraying effect. The initial particle size range (standard atomization parameters used when cracks are not severe or non-existent) typically ranges from 2 to 5 μm. The droplet size regulation sensitivity coefficient is used to amplify or reduce the adjustment range brought about.

[0048] After calculating the target droplet size the system sends the target droplet size to the spray control unit, and in real time updates parameters such as the working voltage, frequency, or nozzle hole geometry channel of the atomizing nozzle, so that the actual dry fog particle size sprayed is adjusted to the target droplet size. When a high degree of crack abnormality is detected, the nozzle automatically adjusts to the large particle size range, enhancing the droplet sedimentation ability and adhesion, and thus enabling the droplets to penetrate into the crack gaps and form a relatively thick liquid film, strengthening sterilization and preservation; if the crack degree is slight, the default small particle size is maintained to achieve a balance between wide-area coverage and water conservation. Through this closed-loop mechanism, the spray system can quickly respond on the production line and implement targeted preservation strategies, improving the protection effect of fruits and vegetables in the micro-crack scenario and reducing the risks of cross-contamination and spoilage.

[0049] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0050] By introducing multi-spectral image acquisition and intelligent image recognition technology, the present invention combines a convolutional neural network model to accurately identify and quantitatively score the severity of micro-cracks on the surface of fruits and vegetables, and based on the scoring results, it real-time regulates the particle size of the slightly acidic electrolyzed water dry fog, achieving differential and dynamic preservation treatment of fruits and vegetables in different crack states. It can significantly enhance the sedimentation and adhesion ability of the dry fog in the crack area and the liquid film formation efficiency, strengthening the local sterilization effect, effectively making up for the technical defects of uneven coverage and insufficient preservation of traditional fixed-particle-size dry fog in the minimally invasive wound area, and significantly improving the preservation accuracy and anti-pollution ability of fruits and vegetables on the high-speed processing production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0052] Figure 1 It is a flowchart of the method for preserving fruits and vegetables based on slightly acidic electrolyzed water dry fog of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be more complete and thorough, and will fully convey the concept of the example embodiments to those skilled in the art.

[0054] The present invention provides a Figure 1 method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water as shown below, including the following steps:

[0055] Atomize slightly acidic electrolyzed water through an atomization control unit to form dry fog within a set starting particle size range (e.g., 2 - 5 μm), and act on the surface of fruits and vegetables during high-speed pipeline transportation to form a wide-area sterilization protection fog layer, achieving preliminary sterilization, disinfection, and preservation treatment of the overall surface of fruits and vegetables;

[0056] The setting of this particle size aims to obtain rapid droplet diffusion, high coverage, and gasification efficiency at the initial stage of treatment, and serve as the initial reference standard for subsequent adaptive particle size adjustment, providing basic preservation guarantee and continuous processing ability for the system.

[0057] Collect multi-spectral image information of the outer surface of fruits and vegetables in a non-contact manner through a multi-spectral image acquisition device during the passage of fruits and vegetables through the pipeline;

[0058] This image acquisition system covers multiple bands such as visible light, near-infrared, and short-wave infrared, and can comprehensively capture information such as the color texture, spectral reflectance change, and microstructural perturbation on the surface of fruits and vegetables. In particular, it can identify micro-crack areas invisible to the naked eye caused by factors such as mechanical impact, water swelling, or epidermal tissue rupture, providing high-resolution original image data input for subsequent data analysis and judgment.

[0059] Collect multi-spectral image information of the outer surface of fruits and vegetables through a multi-spectral image acquisition device, specifically including the following steps:

[0060] First, install the multi-spectral image acquisition device at an appropriate position on the fruit and vegetable processing pipeline to ensure that the viewing angle covers the area where the fruits and vegetables pass through;

[0061] Second, when the fruits and vegetables pass through this acquisition area, trigger the sensor to synchronously start the multi-spectral imaging device to continuously scan the surface of the fruits and vegetables;

[0062] Next, use a multi-channel light source to irradiate the surface of the fruits and vegetables frame by frame according to preset bands (such as visible light, near-infrared, short-wave infrared, etc.), and at the same time, the imaging device synchronously acquires reflection images of each band;

[0063] Subsequently, the collected multi-spectral image data is sent to the data processing unit to provide high-quality and multi-dimensional visual information support for subsequent analysis steps such as image preprocessing, feature extraction, and crack recognition.

[0064] The original image data obtained from the multi-spectral imaging acquisition device usually contains interference factors such as noise and illumination changes. To obtain stable recognition accuracy, preprocessing operations such as denoising, normalization, brightness correction, and background removal need to be performed on the image data. At the same time, the image frames are matched with the positions of the fruits and vegetables on the conveyor belt according to the conveyor belt speed and imaging frequency to ensure that the preprocessing results accurately correspond to the actual spatial positioning of the fruits and vegetables.

[0065] Matching the image frames with the positions of the fruits and vegetables on the conveyor belt according to the conveyor belt speed and imaging frequency generally includes the following steps:

[0066] First, a high-precision encoder or position sensor is set at key positions on the conveyor belt to detect the running speed of the conveyor belt and the time nodes when the fruits and vegetables pass through in real time.

[0067] Secondly, the sampling frequency of the imaging system is synchronized with the speed data of the sensor so that each frame of image corresponds to the position of the fruits and vegetables at a specific time point.

[0068] Then, according to the linear speed of the conveyor belt and the image acquisition time interval, the theoretical spatial position of the fruits and vegetables corresponding to each frame of image is calculated.

[0069] Then, in combination with the fruit and vegetable entry trigger signal or the fruit and vegetable contour information detected in the image, the actual position is finely adjusted and calibrated.

[0070] Finally, the image frame number is bound to the fruit and vegetable identification information to establish a one-to-one correspondence, ensuring that each image can be accurately corresponded to the actual fruit and vegetable individual in the preprocessing, feature extraction, and subsequent control links, and realizing the synchronous registration of image processing and physical position.

[0071] Based on the preprocessed image data, the micro-crack area is identified and located. For the micro-crack area, the crack morphological features reflecting the severity of the micro-cracks on the fruit and vegetable surface are extracted. After deeply analyzing the extracted features through feature engineering techniques, the analyzed features are used as multi-dimensional feature vectors to characterize the severity of the micro-cracks on the fruit and vegetable surface.

[0072] Based on the preprocessed image data, the microcrack region can be identified and located by combining multi-spectral image enhancement and morphological analysis. First, the contrast in spectral response between the crack region and the normal fruit and vegetable epidermis is enhanced through band fusion or band difference, making the microcracks more prominent in a specific band combination. Then, edge detection algorithms (such as Canny or Laplacian operators) are used to extract the suspected crack edges, and combined with image segmentation techniques (such as region growing, threshold segmentation, or superpixel-based clustering) to locate the preliminary crack region. Next, morphological operations (such as erosion, dilation, and thinning) are used to optimize the crack shape and remove artifacts and redundant noise. Finally, the target region that truly conforms to the microcrack characteristics is further screened by combining geometric features and spatial positions to achieve high-precision identification and location.

[0073] For the microcrack region, the crack morphological features reflecting the severity of the microcracks on the fruit and vegetable surface are extracted. Among them, the extracted crack morphological features include the degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface and the change in reflectance of the crack region in different spectral bands. After in-depth analysis of the degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface and the change in reflectance of the crack region in different spectral bands through feature engineering techniques, a crack distribution dispersion index and a crack penetration index are respectively generated. The crack distribution dispersion index and the crack penetration index are used as multi-dimensional feature vectors to characterize the severity of the microcracks on the fruit and vegetable surface.

[0074] A higher degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface usually indicates that there are multiple independent and discontinuous microcrack regions on the fruit and vegetable surface. This feature reflects a wide range of damage to the overall epidermal structure of the fruit and vegetable, uneven local stress, or a relatively high degree of tissue fragility. Compared with cracks with a concentrated distribution, a discrete distribution often means that the fruit and vegetable have been affected by stress or mechanical disturbances at multiple points, and micro-damages appear in a multi-source and multi-point manner, making it more difficult to cover and repair through local preservation interventions. At the same time, it also increases the risk of multi-point microbial invasion and cross-contamination. Therefore, from the perspective of preservation and quality stability, a high-discrepancy crack spatial distribution usually represents a more severe crack and a more fragile overall state of the fruit and vegetable, which should attract key attention and differential treatment.

[0075] The specific steps for generating the crack distribution dispersion index through in-depth analysis of the degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface by feature engineering techniques are as follows:

[0076] First, for all crack regions, the central point coordinates of each crack are marked to construct a crack spatial position set , where N represents the number of cracks detected in the image, is the geometric center coordinate of the i-th crack region. On this basis, a crack distance map is constructed, where , which represents the Euclidean distance between the geometric centers of crack i and crack j, is used to measure the spatial discreteness degree between cracks;

[0077] The function of this step is to structure the discrete distribution relationship in space into a quantifiable and computable distance network, laying a foundation for the construction of the crack distribution discreteness index.

[0078] According to the crack spatial position set and the crack distance map, the maximum connection dissipation ratio is introduced to calculate the crack distribution discreteness index, and the calculation expression is:

[0079]

[0080] Among them, is the crack distribution discreteness index, is a very small positive number to prevent the denominator from being zero; is the farthest distance between crack i and other cracks, representing its most extreme discrete trend; is the reciprocal sum of the distances between crack i and other cracks, measuring its overall connection tightness; the crack distribution discreteness index integrates the dual characteristics of local extreme diffusion and global sparsity. The larger it is, the more dispersed the crack distribution is, the more extensive the damage to the fruit and vegetable surface is, and the higher the severity of the cracks. It is applicable to the identification scenario of severe damage with complex and widespread microcracks.

[0081] From the crack distribution discreteness index, it can be seen that the larger the value of the crack distribution discreteness index generated by deeply analyzing the discreteness degree of cracks in the space of the fruit and vegetable surface through feature engineering technology, the more severe the microcracks on the fruit and vegetable surface are. The crack distribution discreteness index comprehensively analyzes the distribution characteristics of cracks in the space of the fruit and vegetable surface through feature engineering technology, and measures the diffusivity and dispersion degree between crack regions. When the cracks are widely distributed, discrete in position, and have large mutual spacings, the crack distribution discreteness index will increase significantly, reflecting that the fruit and vegetable epidermis is stressed and damaged at multiple points, the structural integrity is reduced, and the microcracks are more severe; while when the cracks are relatively concentrated or the number is small and the space is compact, the crack distribution discreteness index is low, indicating that the damaged range of the fruit and vegetable is limited and the overall epidermis remains in a good state. Therefore, the level of the crack distribution discreteness index can effectively characterize the severity of the cracks, providing a reliable quantitative basis for the intelligent judgment of the fruit and vegetable preservation strategy.

[0082] The more significant the change in the reflectivity of the crack area in different spectral bands, the more severe the microcracks on the surface of fruits and vegetables usually are. The reason is that the generation of microcracks will cause local changes in the surface structure, material and moisture distribution of fruits and vegetables. In particular, phenomena such as depressions, tissue damage, moisture seepage or cell sap leakage often occur inside the crack area. These changes will significantly affect the light reflection characteristics of this area in multiple bands such as visible light, near-infrared or short-wave infrared. For example, water has a strong absorption ability in the near-infrared band. If there is water seepage in the crack area, its reflectivity in the near-infrared band will decrease significantly; while the change in surface roughness caused by structural damage will also produce a strong scattering effect in the visible band. Therefore, when there are obvious differences in the reflectivity changes of multiple spectral channels in the crack area, it often means that the crack is not only a surface scratch, but more likely involves deep tissue damage below the fruit and vegetable epidermis, reflecting a higher crack severity.

[0083] The specific steps to generate the crack penetration index by deeply analyzing the change in the reflectivity of the crack area in different spectral bands through feature engineering techniques are as follows:

[0084] First, in the preprocessed multi-spectral image, extract the pixel-level reflectivity values of the crack area in multiple specific bands, denoted as , where represents the pixel position, is the k-th band. Then, define the local reflectivity gradient vector of this crack area in the band dimension as:

[0085]

[0086] where, is the local reflectivity gradient vector at the pixel position , which reflects the change trend of the reflectivity at this point in multiple spectral bands. This vector describes the change rate of the reflectivity between different bands in the crack area (similar to the slope feature of the spectral curve), and n represents the total number of bands;

[0087] The function of this step is to establish a directional spectral change feature to capture the cross-band reflectivity mutation behavior caused by structural fracture, moisture penetration, internal tissue exposure, etc., and provide a basis for subsequent in-depth characterization.

[0088] After obtaining the local reflectivity gradient vector of the crack area, calculate the crack penetration index based on the tensor projection intensity of the local reflectivity gradient vector in the main extension direction of the crack, which is used to measure the degree of consistency between the local spectral change trend and the main direction of the crack structure, and characterize the penetration depth and development severity of the crack in the tissue structure. Let the main extension direction vector of the crack be , then the calculation expression of the crack penetration index of the crack area is:

[0089]

[0090] Wherein: is the crack penetration index, represents the spatial range of the crack area, is the projection value of the local reflectivity gradient vector in the main crack direction, is the power factor for enhancing the change in crack penetration, is the crack edge sharpness factor (such as the weight map for enhancing the second derivative of the edge), which is used to emphasize the areas with more severe rupture degree;

[0091] The more the area shows a drastic spectral response change along the crack direction, the more likely it is the location with deep - level structural damage, thus it can be used to accurately quantify the penetration and severity of micro - cracks.

[0092] From the crack penetration index, it can be seen that the larger the performance value of the crack penetration index generated by in - depth analysis of the reflectivity change of the crack area in different spectral bands through feature engineering technology, generally means that the micro - cracks on the fruit and vegetable surface are more severe. The crack penetration index is constructed based on the reflectivity change trend of the crack area under different spectral bands, and it comprehensively considers the intensity of the spectral response, the directionality, and the degree of consistency with the main crack axis direction. When the crack is deeper, more complex, or has penetrated through the fruit peel to the internal tissue, it will cause the area to show a more drastic and direction - specific reflectivity change under multi - band spectra, and then form a higher - consistency tensor projection between the spectral gradient vector and the main crack direction, thereby increasing the crack penetration index. On the contrary, if the crack is shallower or is a surface scratch, its influence on the spectral response is weak and has no obvious directionality, then the projection intensity is low, and the performance value of the crack penetration index also decreases accordingly. Therefore, the crack penetration index can effectively quantify the structural severity of micro - cracks and is an important reference basis for the intelligent assessment of crack risks.

[0093] Input the feature vector into a convolutional neural network model that has been pre - trained offline and deployed on - site, and score the severity of the crack according to the morphological characteristics of the crack;

[0094] Input the feature vector constructed by the crack distribution discrete index and the crack penetration index into a convolutional neural network model that has been pre - trained offline and deployed on - site, generate the crack structure abnormality degree based on the convolutional neural network model, and score the severity of the crack through the crack structure abnormality degree.

[0095] A convolutional neural network model that is pre-trained offline and deployed on-site refers to the process of finely training a convolutional neural network in a laboratory or computing platform in advance using supervised learning for the specific application requirements of fruit and vegetable surface crack recognition. Specifically, first in the offline stage, a large number of multi-spectral image sample data of fruit and vegetable surfaces are collected, and these image data are pre-processed and manually annotated, including marking the location, type, and severity of cracks, to form a dataset with high-quality annotation information. Then, multiple morphological features of cracks (such as crack distribution dispersion index, crack depth penetration index, etc.) are used as input features to construct feature vectors, which are associated with the corresponding annotation information (such as crack severity level or specific value) to generate standard training sample pairs. Next, a suitable convolutional neural network structure (such as LeNet, ResNet, or Inception series networks) is selected, and based on the crack feature data and corresponding annotations, a powerful server with GPU or CPU computing power is used for long-term, multi-round iterative training and parameter adjustment to achieve the optimal configuration of the network structure, convolutional kernel weights, and parameters. When the model performance reaches the ideal recognition accuracy and generalization ability on an independent test set, the offline training process of the convolutional neural network model is completed, and efficient crack recognition network parameters suitable for on-site use are obtained.

[0096] Deploying on-site means transferring the above-mentioned convolutional neural network model completed in offline training to the fruit and vegetable processing pipeline site and performing actual operations with the help of embedded computing devices, edge computing devices, or industrial-grade computing units (such as edge servers, industrial computers, etc.). In actual on-site applications, real-time multi-spectral image acquisition devices are installed on the fruit and vegetable production line. When fruits and vegetables pass by, image acquisition and pre-processing operations can be quickly completed, and the crack distribution dispersion index and crack depth penetration index feature parameters are extracted in real-time from the image data collected on-site. Then, the pre-deployed convolutional neural network model is used to process and analyze these feature vectors in real-time to quickly obtain the crack structure abnormality degree. At this time, the convolutional neural network model no longer needs to be trained on a large scale again, but only needs to complete efficient forward propagation operations to quickly output the crack structure abnormality degree of fruits and vegetables, and quickly and accurately evaluate the severity of cracks through the crack structure abnormality degree. This offline training and online deployment strategy not only ensures that the model has high recognition accuracy and generalization performance, but also effectively avoids the computing power burden and response delay brought by re-training on-site, thus greatly improving the real-time performance and stability of crack recognition and intelligent decision-making in actual industrial applications, and realizing dynamic crack recognition and real-time assessment of crack severity.

[0097] The convolutional neural network is not specifically defined here, and it can achieve comprehensive analysis of the crack distribution dispersion index and the crack depth penetration index to generate the crack structure abnormality degree Any convolutional neural network can be used. To implement the technical solution of the present invention, the present invention provides a specific implementation; crack structure abnormality The generated expression is:

[0098] , where in the formula, 、 are respectively the preset proportionality coefficients of the crack distribution dispersion index and the crack penetration index , and 、 are both greater than 0. The preset proportionality coefficient refers to when constructing the expression of the crack structure abnormality, in order to reasonably weight and fuse two indexes of different dimensions or dimensions (that is, the crack distribution dispersion index and the crack penetration index ), the weight factors that need to be artificially preset in advance are and . The role of these two coefficients is to adjust and balance the weights of the crack distribution dispersion index and the crack penetration index in the final abnormality score. Since and may have differences in the numerical range, change trend and influence mechanism, it is easy to be distorted if directly added, so it is necessary to perform normalization or difference compensation through the preset proportionality coefficient, so that the contribution degrees of the two in the output of the convolutional neural network conform to the physical or statistical significance of the actual crack severity. Usually, these coefficients are obtained through experiments or optimization during the training stage, or set by empirical values, and 、 are both greater than 0 to ensure the interpretability and stability of the model.

[0099] It can be seen from the crack structure abnormality that the larger the value of the crack distribution dispersion index generated by deeply analyzing the dispersion degree of cracks in the spatial space on the fruit and vegetable surface through feature engineering technology, and the larger the value of the crack penetration index generated by deeply analyzing the reflectance change of different spectral bands on the crack area through feature engineering technology, that is, when predicting the severity of cracks through the convolutional neural network model pre-trained offline and deployed on-site, the larger the value of the crack structure abnormality generated, it indicates that the micro-cracks on the fruit and vegetable surface are more serious, and vice versa, it indicates that the micro-cracks on the fruit and vegetable surface are less serious.

[0100] According to the severity of cracks output by the convolutional neural network, the working parameters of the atomizing nozzle are adjusted in real time, and the size of the dry fog droplets is dynamically changed. If serious cracks are detected on the surface of fruits and vegetables, the droplet size is automatically increased to enhance the sedimentation ability and adhesion of the droplets, ensuring that they can effectively enter the crack gaps and adhere to the inside of the cracks to form a liquid film, thereby enhancing the local sterilization and preservation effects. In the case of lighter or non-existent cracks, a smaller droplet size (such as 2 - 5 μm) is maintained to increase the coverage area and reduce water consumption.

[0101] Through the above adaptive regulation mechanism, directional enhanced preservation of the crack areas on the surface of fruits and vegetables can be achieved during high-speed pipeline operation, significantly reducing the risks of minimally invasive wound infection and cross-contamination, and extending the storage and transportation life of fruits and vegetables.

[0102] According to the severity of cracks output by the convolutional neural network, the working parameters of the atomizing nozzle are adjusted in real time, and the size of the dry fog droplets is dynamically changed. The specific steps are as follows:

[0103] After obtaining the crack structure abnormality degree, first judge whether the difference between it and the crack structure abnormality degree reference threshold is positive, that is, compare whether the current severity of cracks on the surface of fruits and vegetables exceeds the preset benchmark. For this purpose, a difference function is defined, and the expression of the difference function is:

[0104]

[0105] Among them, is the crack structure abnormality degree output by the convolutional neural network. The larger the value, the more serious the crack. is the crack structure abnormality degree reference threshold, which is used to measure whether the crack reaches or exceeds the acceptable limit. is the difference between the crack structure abnormality degree and the crack structure abnormality degree reference threshold, which is a temporary judgment quantity for determining whether to increase the particle size. If , it means that the current crack condition is still within the safe range; if , it indicates that the crack abnormality degree exceeds the benchmark, and the droplet size needs to be dynamically increased in the subsequent steps.

[0106] The function of this step is to quickly identify the cracks on the surface of fruits and vegetables, avoid unnecessary adjustment of the droplet size, and effectively improve the system response efficiency.

[0107] If , the default particle size within the initial particle size range of the atomizing nozzle is maintained.

[0108] If , it is necessary to expand on the basis of this initial particle size range, making the droplet size increase appropriately and penetrate the crack gaps. The expression for expanding the initial particle size range is:

[0109]

[0110] Among them, is the target droplet size, and are the minimum and maximum droplet sizes allowed by the system design, which are used to limit the extreme values of the final particle size, ensuring the safety of the nozzle hardware and the stability of the spraying effect. is the initial particle size range (the standard atomization parameters used when the crack is not serious or does not exist), and its value is usually in the range of 2 - 5μm. is the sensitivity coefficient of droplet size regulation, which is used to amplify or reduce the adjustment range brought by

[0111] Through this step, when is relatively large, an exponential increase in the droplet size can be achieved, but it is still limited within the range to ensure that the particle size will not be overly enlarged or fall below the available value.

[0112] After calculating the target droplet size , the system sends this target droplet size to the spray control unit, and in real time updates parameters such as the working voltage, frequency, or the geometric channel of the spray hole of the atomizing nozzle, so that the size of the actually sprayed dry fog droplets is adjusted to the target droplet size. When a high degree of crack abnormality is detected, the nozzle will automatically adjust to the large particle size range, improving the sedimentation ability and adhesion of the droplets, so that the droplets can penetrate into the crack gaps and form a relatively thick liquid film, strengthening sterilization and preservation; if the crack degree is slight, the default small particle size will be maintained to achieve a balance between wide - area coverage and water conservation. Through this closed - loop mechanism, the spray system can quickly respond on the production line and implement targeted preservation strategies, improving the protection effect of fruits and vegetables in the micro - crack scenario, and reducing the risks of cross - contamination and spoilage.

[0113] By introducing multi - spectral image acquisition and intelligent image recognition technology, this invention combines a convolutional neural network model to accurately identify and quantitatively score the severity of micro - cracks on the surface of fruits and vegetables, and based on the scoring results, it real - time regulates the size of the micro - acidic electrolyzed water dry fog, achieving differential and dynamic fresh - keeping treatment of fruits and vegetables in different crack states. It can significantly improve the sedimentation and adhesion ability of dry fog in the crack area and the liquid film formation efficiency, strengthen the local sterilization effect, effectively make up for the technical defects of uneven coverage and insufficient preservation of traditional fixed - size dry fog in the minimally invasive wound area, thus significantly improving the fresh - keeping accuracy, anti - pollution ability and commodity stability of fruits and vegetables on the high - speed processing production line, and having good intelligent, flexible and application promotion value.

[0114] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0115] Only some exemplary embodiments of the present invention have been described above by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

[0116] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for preserving fruits and vegetables based on dry mist of slightly acidic electrolyzed water, characterized in that, It includes the following steps: Atomize slightly acidic electrolyzed water into dry fog within a set initial particle size range and act on the surface of fruits and vegetables during the transmission process on a high-speed assembly line to form a wide-area sterilizing and protective fog layer; Collect multispectral image information of the outer surface of fruits and vegetables through a multispectral image acquisition device during the passage of fruits and vegetables through the assembly line; Perform preprocessing operations on the collected original images, and at the same time match the image frames with the positions of fruits and vegetables on the assembly line according to the assembly line speed and imaging frequency; Based on the preprocessed image data, identify and locate the microcrack regions. For the microcrack regions, extract the crack morphological features that reflect the severity of the microcracks on the surface of fruits and vegetables. After deeply analyzing the extracted features through feature engineering techniques, use the analyzed features as multi-dimensional feature vectors to characterize the severity of the microcracks on the surface of fruits and vegetables; Input the feature vectors into a convolutional neural network model that has been pre-trained offline and deployed on-site, and score the severity of the cracks according to the morphological features of the cracks; According to the severity of the cracks output by the convolutional neural network, adjust the working parameters of the atomizing nozzle in real time to dynamically change the particle size of the dry fog.

2. The method for preserving fruits and vegetables based on dry mist of slightly acidic electrolyzed water according to claim 1, wherein, Collect multispectral image information of the outer surface of fruits and vegetables through a multispectral image acquisition device, which specifically includes the following steps: Install a multispectral image acquisition device on the fruit and vegetable processing assembly line to ensure that the viewing angle covers the area where the fruits and vegetables pass; When the fruits and vegetables pass through the acquisition area, trigger the sensor to synchronously start the multispectral imaging device to continuously scan the surface of the fruits and vegetables; Use a multi-channel light source to irradiate the surface of the fruits and vegetables frame by frame according to the preset wavelength bands, and synchronously collect the reflected images of each wavelength band through the imaging device; Send the collected multispectral image data to the data processing unit.

3. The method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water according to claim 1, wherein, Match the image frames with the positions of fruits and vegetables on the assembly line according to the assembly line speed and imaging frequency, including the following steps: Set a high-precision encoder at a key position on the assembly line to detect the running speed of the conveyor belt and the time nodes when the fruits and vegetables pass in real time; Synchronize the sampling frequency of the imaging system with the speed data of the sensor so that each frame of image corresponds to the position of the fruits and vegetables at a certain time point; Calculate the theoretical spatial position of the fruits and vegetables corresponding to each frame of image on the conveyor belt according to the linear speed of the conveyor belt and the image acquisition time interval; Combined with the fruit and vegetable contour information detected in the image, fine-tune and calibrate the actual position; Bind the image frame numbers with the fruit and vegetable identification information to establish a one-to-one correspondence.

4. The method for preserving fruits and vegetables based on dry mist of slightly acidic electrolyzed water according to claim 1, wherein, Adopt a method combining multispectral image enhancement and morphological analysis to identify and locate the microcrack regions based on the preprocessed image data. The specific steps are as follows: Enhance the contrast of the spectral response between the crack regions and the normal fruit and vegetable epidermis through band fusion, making the microcracks more prominent in a specific band combination; Adopt an edge detection algorithm to extract the suspected crack edges, and then combine image segmentation technology to locate the preliminary crack regions; Optimize the crack shape through morphological operations to remove artifacts and redundant noise; Further screen out the target regions that meet the microcrack characteristics by combining geometric features and spatial positions to achieve high-precision identification and location.

5. The method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water according to claim 1, characterized in that, For the microcrack region, morphological features of the cracks that reflect the severity of the microcracks on the fruit and vegetable surface are extracted. Among them, the extracted morphological features of the cracks include the degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface and the change in the reflectance of the crack region in different spectral bands. After deeply analyzing the degree of dispersion of the cracks in the spatial domain on the fruit and vegetable surface and the change in the reflectance of the crack region in different spectral bands through feature engineering techniques, a crack distribution dispersion index and a crack penetration index are respectively generated. The crack distribution dispersion index and the crack penetration index are used as multi-dimensional feature vectors to characterize the severity of the microcracks on the fruit and vegetable surface.

6. The method for preserving fruits and vegetables based on dry mist of slightly acidic electrolyzed water according to claim 5, wherein, The feature vector constructed by the crack distribution dispersion index and the crack penetration index is input into a convolutional neural network model that has been pre-trained offline and deployed on-site. Based on the convolutional neural network model, a crack structure abnormality degree is generated, and the severity of the cracks is scored through the crack structure abnormality degree.

7. The method for preserving fruits and vegetables based on dry fog of slightly acidic electrolyzed water according to claim 5, characterized in that The specific steps for generating the crack distribution dispersion index by deeply analyzing the dispersion degree of cracks in the spatial domain on the fruit and vegetable surface through feature engineering techniques are as follows: For all crack regions, mark the central point coordinates of each crack to construct a crack spatial position set , where N represents the number of cracks detected in the image, is the geometric center coordinate of the i -th crack region; Construct a crack distance map represents the Euclidean distance between the geometric centers of crack i and crack j , which is used to measure the spatial dispersion degree between cracks; According to the crack spatial position set and the crack distance map, the maximum connection dissipation ratio is introduced to calculate the crack distribution dispersion index, and the calculation is expressed as:

8. The fresh-keeping method for fruits and vegetables based on dry mist of slightly acidic electrolyzed water according to claim 5, wherein, The specific steps for generating the crack penetration index by deeply analyzing the reflectance changes of the crack area in different spectral bands through feature engineering techniques are as follows: In the preprocessed multi-spectral image, extract the pixel-level reflectance values of the crack area in multiple specific bands, denoted as Define the local reflectance gradient vector of the crack area in the band dimension as: , reflecting the reflectance change trend of this point in multiple spectral bands, n indicating the total number of bands; After obtaining the local reflectivity gradient vector of the crack region, the crack penetration index is calculated based on the tensor projection intensity of the local reflectivity gradient vector in the main crack extension direction, which is used to measure the degree of consistency between the local spectral change trend and the main direction of the crack structure, and to characterize the penetration depth and development severity of the crack in the tissue structure. Let the main extension direction vector of the crack be , then the calculation expression of the crack penetration index in the crack region is: , The projection value of the refractive index gradient vector in the main crack direction, is the power factor for enhancing the change in crack deep penetration, is the crack edge sharpness factor, which is used to emphasize the regions with a more severe degree of rupture.

9. The method for preserving fruits and vegetables based on dry mist of slightly acidic electrolyzed water according to claim 5, characterized in that According to the severity of the cracks output by the convolutional neural network, the working parameters of the atomizing nozzle are adjusted in real time to dynamically change the size of the dry fog particles. The specific steps are as follows: Define a difference function to judge whether the difference between the crack structure abnormality degree and the reference threshold of the crack structure abnormality degree is positive, that is, to compare whether the severity of the cracks on the current fruit and vegetable surface exceeds the preset benchmark. The expression of the difference function is: , where is the crack structure abnormality degree output by the convolutional neural network, is the reference threshold of the crack structure abnormality degree, is the difference between the crack structure abnormality degree and the reference threshold of the crack structure abnormality degree, which is used to determine whether the particle size needs to be increased; Then, on the basis of the initial particle size range, it is amplified, and the droplet particle size is moderately increased. The expression for amplifying the initial particle size range is: , is used to limit the extreme value of the final particle size, is the droplet particle size regulation sensitivity coefficient, which is used to amplify or reduce the adjustment range brought by After calculating the target droplet size , send the target droplet size to the spray control unit to adjust the dry mist droplet size of the actual spraying to the target droplet size.

Images