Tea wall breaking rate detection method, equipment and system

Abstract

The invention relates to the technical field of image area detection, in particular to a tea wall breaking rate detection method, equipment and system. The method comprises the following steps: acquiring gray images of leaves at different angles after each batch of rolling, and initially quantifying the wall breaking condition through a pixel value difference area to obtain an initial wall breaking index; determining a suspected crack edge according to the edge direction consistency degree and the width smoothness condition, obtaining a wall breaking sufficient evaluation index according to the length position distribution uniformity condition of the suspected crack, and analyzing the boundary complexity to adjust the initial index to obtain an adjusted wall breaking index; and obtaining a confidence coefficient by adjusting index consistency and a historical batch similarity degree through different angles, and determining a final wall breaking rate based on the confidence coefficient. According to the method, wall breaking evaluation is adjusted by considering the influence of leaf veins and tearing characteristics, abnormal conditions caused by process drift are found in time through real-time detection and correlation historical process reference analysis after rolling, and stable quality of tea production is guaranteed while the accuracy of wall breaking evaluation is improved in factory production.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image region detection, and in particular to a tea leaf cell wall breaking rate detection method, device and system. BACKGROUND

[0002] During the traditional processing of tea, i.e., during the withering, fixation, rolling and drying processes, cell wall breaking will occur to different degrees. Cell wall breaking can improve flavor release, and tea polyphenols and amino acids in the cells are more easily dissolved after cell wall breaking, which affects the taste and aroma of the tea soup. Cell wall breaking can also improve the dissolution rate, and moderate cell wall breaking can make the tea soup release effective components more quickly, which is convenient for brewing. At the same time, the functional utilization is improved, and in particular in products such as broken tea powder and health tea, the degree of cell wall rupture directly determines the availability of nutritional ingredients.

[0003] In actual factory production, manual judgment relies on experience, has no unified standard and is low in efficiency, and it is difficult to cover a large number of samples. In addition, there is a lack of quantitative indicators. Although high-precision microscopic detection in the laboratory is accurate, it requires a complex process, a long cycle and a high cost, and is not suitable for real-time operation on the production line. Although the existing image processing detection quantitative indicators are suitable for production line scenarios, they ignore the influence of the leaf characteristics of tea leaves themselves, which can lead to cell wall breaking misjudgment and increase the estimation and quantization error, and it is difficult to meet the quality control of tea production. SUMMARY

[0004] In order to solve the technical problem that the image processing detection in the prior art ignores the influence of the leaf characteristics of tea leaves themselves, which can lead to cell wall breaking misjudgment and increase the estimation and quantization error, and it is difficult to meet the quality control of tea production, the purpose of the present application is to provide a tea leaf cell wall breaking rate detection method, device and system, and the technical solution adopted is as follows: The first aspect of the present application provides a tea leaf cell wall breaking rate detection method, which comprises: After each batch of rolling processing, obtain leaf gray scale images of the sample tea leaves at different angles; On the leaf gray scale image at each angle, obtain an initial cell wall breaking index of the leaf gray scale image according to the area degree of pixel value difference, determine a suspected crack edge according to the uniformity degree of the edge direction and the smoothing condition of the edge width in the leaf gray scale image, and obtain a cell wall breaking sufficiency evaluation index of the leaf gray scale image according to the length position distribution uniformity of the suspected crack edge; analyze the boundary complexity of the leaf in the leaf gray scale image, adjust the initial cell wall breaking index in combination with the cell wall breaking sufficiency evaluation index, and obtain an adjusted cell wall breaking index of the leaf gray scale image. Obtain a cell wall breaking confidence of the sample tea leaves of the current batch according to the consistency of the adjusted cell wall breaking indexes of the leaf gray scale images of the sample tea leaves at different angles and the similarity degree of the adjusted cell wall breaking indexes of the current batch and historical batches, and determine a final cell wall breaking rate of the current batch based on the cell wall breaking confidence.

[0005] Further, the method for obtaining the initial wall breaking index comprises: For any one leaf gray image, a connected domain is obtained by using Otsu threshold segmentation method on the leaf gray image; a ratio between a total area of all connected domains and a total area of the leaf in the leaf gray image is taken as the initial wall breaking index.

[0006] Further, the method for determining the suspected crack edge comprises: For any one leaf gray image, an edge line of the leaf gray image is obtained; for any one edge line, a gradient direction difference between every two adjacent pixel points on the edge line is calculated, and then a mean value of all gradient direction differences is negatively correlated mapped as a direction consistency index of the edge line; a gradient value difference between every two adjacent pixel points on the edge line is calculated, and then a standard deviation of all gradient value differences is negatively correlated mapped to obtain a width smoothness index of the edge line; a product of the direction consistency index and the width smoothness index of the edge line is taken as an edge feature value of the edge line; When the edge feature value is less than a preset feature threshold, the corresponding edge line is taken as the suspected crack edge.

[0007] Further, the method for obtaining the initial wall breaking index comprises: For any one leaf gray image, the leaf gray image is divided into a preset number of sub-regions; a proportion of pixel points of the suspected crack edge in each sub-region is taken as a damage distribution degree of each sub-region; a variance of damage distribution degrees of all sub-regions is negatively correlated mapped as a local distribution stability of the leaf gray image; a product of a mean value of gradient values of all pixel points on the suspected crack edge in the leaf gray image and a total length of the suspected crack edge is taken as a cumulative distribution degree of the leaf gray image; The local distribution stability and the cumulative distribution degree of the leaf gray image are combined to obtain a wall breaking sufficient evaluation index of the leaf gray image.

[0008] Further, the method for obtaining the initial wall breaking index comprises: A ratio between a total length of the leaf boundary and a total area of the leaf in each leaf gray image is taken as a boundary tearing feature value of each leaf gray image; A product of a value negatively correlated mapped by the boundary tearing feature value of each leaf gray image and the initial wall breaking index is taken as a wall breaking area evaluation index of each leaf gray image; A sum of the wall breaking area evaluation index and the wall breaking sufficient evaluation index of each leaf gray image is normalized to obtain an adjusted wall breaking index of each leaf gray image.

[0009] Further, the method for obtaining the cell wall breaking confidence level comprises: The average cell wall breaking index of the sample tea leaves in each batch is obtained by averaging all the adjusted cell wall breaking indexes at all angles, and the average cell wall breaking index of each historical batch is obtained by averaging all the adjusted cell wall breaking indexes at all angles. The difference between the average cell wall breaking index of the current batch and the average cell wall breaking index of each historical batch is calculated, and the sum of all the average cell wall breaking index differences is negatively correlated to obtain the historical confidence level of the current batch. The variance of the adjusted cell wall breaking index at all angles of the current batch is negatively correlated to obtain the cell wall breaking confidence level of the sample tea leaves of the current batch in combination with the historical confidence level.

[0010] Further, the method for determining the final cell wall breaking rate of the current batch based on the cell wall breaking confidence level comprises: When the cell wall breaking confidence level is greater than a preset confidence threshold, the final cell wall breaking rate of the finished product after processing is obtained by image analysis; otherwise, the final cell wall breaking rate of the finished product after processing is obtained by a spectrometer.

[0011] Further, after obtaining the cell wall breaking confidence level of the sample tea leaves of the current batch, the method further comprises: When the cell wall breaking confidence level is greater than a preset confidence threshold, if the adjusted cell wall breaking index is less than a preset rolling and twisting cell wall breaking threshold, a cell wall breaking adjustment alarm is performed.

[0012] In a second aspect, the present application provides a tea leaf cell wall breaking rate detection system, which comprises: A data acquisition module is configured to obtain leaf gray scale images of sample tea leaves at different angles after rolling and twisting processing of each batch. A cell wall breaking adjustment analysis module is configured to obtain an initial cell wall breaking index of the leaf gray scale image according to the pixel value difference area degree, determine a suspected crack edge according to the edge direction consistency degree and edge width smoothing condition in the leaf gray scale image, and obtain a cell wall breaking sufficiency evaluation index according to the length position distribution uniformity of the suspected crack edge, analyze the boundary complexity of the leaf in the leaf gray scale image, adjust the initial cell wall breaking index in combination with the cell wall breaking sufficiency evaluation index, and obtain an adjusted cell wall breaking index of the leaf gray scale image. A final cell wall breaking rate analysis module is configured to obtain a cell wall breaking confidence level of the current sample tea leaves according to the consistency of the adjusted cell wall breaking index of the leaf gray scale image at different angles and the similarity of the adjusted cell wall breaking index between the current batch and the historical batch, and obtain a final cell wall breaking rate by judging the cell wall breaking confidence level.

[0013] In a third aspect, the present application provides a tea leaf broken cell wall rate detection device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method of the first aspect or any embodiment of the first aspect of the present application when executing the computer program.

[0014] In a fourth aspect, the present application provides a computer program product, comprising computer program code which, when executed, performs the method of the first aspect or any embodiment of the first aspect of the present application.

[0015] In a fifth aspect, the present application provides a computer-readable storage medium storing computer program code which, when executed, performs the method of the first aspect or any embodiment of the first aspect of the present application.

[0016] The present application has the following beneficial effects: The present application obtains different angle leaf gray scale images after each batch of rolling, so as to comprehensively cover the broken cell wall characteristics of each surface of the leaf under the condition that the image quality is limited. According to the phenomenon that the continuity of the leaf tissue is destroyed by the broken cell wall, the initial broken cell wall condition can be quantified by the pixel value difference area, and the initial broken cell wall index is obtained. Considering that the broken features such as strong folds or broken edges are affected by the existing veins of the leaf, which causes errors in the broken cell wall evaluation, by the characteristics that the vein direction is more uniform and the width gradually changes smoothly, and the crack is opposite, the suspected crack edge is determined according to the edge direction consistency and width smoothness, and the broken cell wall region recognition accuracy is improved. And through the uniform distribution of the suspected crack, the broken cell wall sufficiency is characterized, and the broken cell wall sufficiency evaluation index is obtained according to the uniform distribution of the suspected crack length position, so as to improve the accuracy of the broken cell wall evaluation and make the evaluation more comprehensive. The initial index is adjusted by analyzing the boundary complexity to obtain the adjusted broken cell wall index, and the broken cell wall evaluation is further adjusted from the leaf edge tearing condition, so that the single angle broken cell wall index evaluation is more reliable. The confidence is obtained by adjusting the consistency of the index at different angles and the similarity with the historical batches, which verifies the reliability of the multi-angle data and associates the historical process rules to improve the result reliability. The final broken cell wall rate is determined based on the confidence, which can filter reliable results and ensure that the final value accurately reflects the actual broken cell wall condition. The present application considers the influence of the leaf vein and tearing characteristics on the broken cell wall evaluation, and analyzes the historical process reference by real-time detection after rolling, so as to timely find the abnormal condition caused by process drift, improve the broken cell wall evaluation accuracy in the factory production, and ensure the quality stability of the tea leaf production. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, a brief introduction will be given to the drawings required in the embodiments or prior art description. Obviously, the drawings in the following description only show some embodiments of the present application, and for those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0018] Figure 1 A flow chart of a tea leaf cell wall breaking rate detection method provided by an embodiment of the present application; Figure 2 A local schematic diagram of a leaf gray image provided by an embodiment of the present application; Figure 3 A structural diagram of a tea leaf cell wall breaking rate detection system provided by an embodiment of the present application; Figure 4 A structural schematic diagram of a tea leaf cell wall breaking rate detection device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined purposes, the specific embodiments, structures, features and effects of the tea leaf cell wall breaking rate detection method, device and system according to the present application are described in detail as follows in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0021] The specific scheme of the tea leaf cell wall breaking rate detection method, device and system provided by the present application is described in detail below in combination with the drawings.

[0022] Please refer to Figure 1 which shows a flow chart of a tea leaf cell wall breaking rate detection method provided by an embodiment of the present application. The method comprises the following steps: S1: After each batch of rolling processing, obtain leaf gray images of sample tea leaves at different angles.

[0023] Rolling is a key process of preliminary breaking of tea leaf, the embodiment of the application avoids the situation that the subsequent process cannot be corrected by monitoring the preliminary breaking after rolling, improves the quality of tea, therefore, in the embodiment of the application, an automatic sampler is installed at the outlet of the rolling machine, such as a small shunt sampling device, to grab a small amount of tea at regular intervals and send it to the detection port to obtain sample tea. Each leaf in the sample tea is placed on a non-woven fabric or a dull black tray, and is gently pulled apart with tweezers and naturally stretched, and is fixed with a soft silicone pad and a transparent sheet for preventing displacement but not crushing the cells.

[0024] Further, the leaf image under different angles is collected by an industrial camera, and the leaf image is preprocessed to obtain a leaf gray image. In the embodiment of the application, the leaf gray image is obtained through several angles such as front view, left and right 45° side view and up and down 45° side view, so as to more comprehensively analyze the possible curling and hiding conditions. The image preprocessing process can specifically include image graying processing, filter denoising processing and background removal processing. It should be noted that the image preprocessing process is a technology known to those skilled in the art, and specific methods such as gray weighting method for graying, bilateral filter method for filter denoising and background difference method for background removal can be selected, and details are not described here. Please refer to Figure 2 Fig. 2 shows a local schematic diagram of a leaf gray image provided by an embodiment of the application.

[0025] It can be understood that the sampler setting implementer can adjust according to the specific implementation scene, such as manually sampling sample tea by an operator at the outlet of the rolling machine on the production line through fixed sampling time, etc. The sampling process is not limited here.

[0026] S2: On the leaf gray image at each angle, the initial breaking index of the leaf gray image is obtained according to the area degree of pixel value difference; the suspected crack edge is determined according to the consistent degree of edge direction and the smoothness of edge width in the leaf gray image; the breaking full evaluation index of the leaf gray image is obtained according to the length position distribution uniformity of the suspected crack edge; the boundary complexity of the leaf in the leaf gray image is analyzed, the initial breaking index is adjusted combined with the breaking full evaluation index, and the adjusted breaking index of the leaf gray image is obtained.

[0027] The macro aspect of the breaking mainly reflects the destruction of leaf tissue continuity, and the representative phenomena appearing on the image can include: cracks or breaks, holes, edge tears, abnormal light transmission, abnormal color spots, exudation or bruise-like and strong crease accompanied by texture mutation, etc. Under normal circumstances, the color of the breaking position will be different from the normal position due to damage, so the pixel value difference can be used for preliminary quantitative analysis of the breaking condition.

[0028] In the embodiment of the present application, the method for obtaining the initial wall breaking index comprises: for any one leaf gray image, the connected domain is obtained by using the Otsu threshold segmentation method on the leaf gray image, the threshold segmentation is directly performed according to the pixel difference of the damage, and the damage degree is analyzed as a whole. The ratio between the total area of all connected domains and the total area of the leaf in the leaf gray image is taken as the initial wall breaking index. It should be noted that the Otsu threshold segmentation method is a well-known technical means known to those skilled in the art, and will not be described here.

[0029] In one specific embodiment of the present application, the total number of pixel points in the connected domain is taken as the total area of the connected domain, and the total number of all pixel points in the leaf gray image after background removal is the total area of the leaf. When the ratio is larger, that is, the initial wall breaking index is larger, it means that the proportion of the damaged part is higher and the wall breaking rate is higher.

[0030] However, there is a certain degree of damaged tea during the processing of tea. The damaged tea is curled, and it is extremely possible that the total area cannot fully represent the damage and errors occur. When the crease in the tea is heavy but the area of the connected domain is small, the wall breaking rate is large, but the accuracy of the evaluated wall breaking rate is not high due to the small crease area.

[0031] Therefore, the strong crease or broken mouth, which is a damaged feature, is considered to modify the wall breaking evaluation. However, since the leaf itself also has a certain vein feature, when analyzing the edge of the crack with a texture distribution, the vein interference is first excluded. The vein of the tea itself presents a tree branch shape, the leaf vein radiates outward from the main vein, the local direction is smooth in a long range, the direction presents good consistency, and the width changes slowly along the vein direction. The cell rupture caused by rolling and twisting produces irregular broken or bifurcated cracks, so the crack often presents a non-symmetrical strong gradient section and the direction changes randomly with high dispersion.

[0032] By the direction distribution and width smoothness of the edge, the vein information edge is filtered out to determine the crack edge. In the embodiment of the present application, the method for obtaining the suspected crack edge comprises: First, for any one leaf gray image, the edge line of the leaf gray image is obtained. For any one edge line, the gradient direction difference between every two adjacent pixel points on the edge line is calculated, then the mean value of all gradient direction differences is negatively correlated mapped as the direction consistency index of the edge line. Through the continuous direction approximation analysis, when the overall difference is lower, it means that the random change on the edge is smaller, the direction consistency is good, and the vein possibility is higher.

[0033] It should be noted that the negative correlation mapping and edge acquisition are well-known technical means familiar to those skilled in the art, the negative correlation mapping can adopt, for example, an inverse proportional value or a negative exponential power form with a natural constant as a base, and the edge detection can adopt, for example, a Canny algorithm, and details are not repeated and limited herein.

[0034] Further, after calculating the gradient value difference between every two adjacent pixel points on the edge line, the standard deviation of all gradient value differences is negatively correlated mapped to obtain the width smoothness index of the edge line. When the standard deviation is smaller, it indicates that the width change of adjacent points is gentle, the smoothness is higher, and the possibility of the choroid is higher.

[0035] Finally, the product of the direction consistency index and the width smoothness index of the edge line is taken as the edge feature value of the edge line, which reflects the significant of the choroid feature through the edge direction and the width smoothness distribution. When the edge feature value is larger, it indicates that the edge information of the choroid is more significant. Therefore, when the edge feature value is smaller than a preset feature threshold, the edge is considered as a texture with possible damage, and the corresponding edge line is taken as a suspected crack edge. In the embodiment of the present application, the preset feature threshold can be set to 0.68, and the specific value can be adjusted by the implementer, which is not limited herein.

[0036] The suspected crack edge can be evaluated according to the distribution to improve the accuracy of the wall breaking evaluation. When the length-width overall proportion of the crack damage in the leaf is higher, it indicates that the wall breaking effect is larger. At the same time, when the damage is more evenly distributed on the leaf, the crack wall breaking effect is better, which indicates that the wall breaking condition is more sufficient.

[0037] Therefore, in the embodiment of the present application, the method for obtaining the wall breaking sufficiency evaluation index comprises: First, for any one leaf gray image, the leaf gray image is divided into a preset number of sub-regions, and the distribution uniformity is judged according to the distribution in the local region. The proportion of the pixel points of the suspected crack edge in each sub-region is taken as the damage distribution degree of each sub-region to quantify the distribution degree in each region. The variance of the damage distribution degrees of all sub-regions is negatively correlated mapped to be taken as the local distribution stability of the leaf gray image. When the variance is smaller, it indicates that the distribution degrees in the sub-regions are similar, and the distribution uniformity is better. In the embodiment of the present application, the preset number can be set to 30, and the specific value can be adjusted by the implementer, which is not limited herein.

[0038] Further, the product of the mean value of the gradient values of all pixel points on the suspected crack edge in the leaf gray image and the total length of the suspected crack edge is taken as the cumulative distribution degree of the leaf gray image, which reflects the damage degree of the crack damage in the overall distribution through the overall significant width level and length. When the cumulative distribution degree is larger, it indicates that the distribution degree is higher, and the wall breaking level is higher.

[0039] Finally, the broken wall sufficiency evaluation index of the leaf gray image is obtained in combination with the local distribution stability and the cumulative distribution of the leaf gray image. In the embodiment of the present application, the product of the local distribution stability and the cumulative distribution of the leaf gray image is taken as the broken wall sufficiency evaluation index of the leaf gray image. The greater the local distribution stability and the cumulative distribution are, the more and more uniform the crack edge distribution is, and the more sufficient the broken wall is.

[0040] After the area analysis and the edge analysis, a more comprehensive broken wall evaluation result can be obtained. At this time, the edge tearing of the leaf is considered. The more serious the tearing of the leaf edge is, the more serious the real area damage is, and the lower the broken wall result reflected by the area is. Therefore, the initial broken wall index of the area analysis is limited by the edge tearing.

[0041] In the embodiment of the present application, the method for adjusting the broken wall index comprises: Firstly, the ratio of the total length of the leaf boundary to the total area of the leaf in each leaf gray image is taken as the boundary tearing characteristic value of each leaf gray image. The ratio of the total length of the outer contour boundary to the total area in the leaf gray image reflects the boundary complexity. The greater the ratio is, that is, the greater the boundary tearing characteristic value is, the higher the complexity of the outer contour is, and the more serious the tearing of the boundary is, and the lower the accuracy of the area calculation is.

[0042] Therefore, the product of the value of the boundary tearing characteristic value of each leaf gray image which is negatively correlated and mapped and the initial broken wall index is taken as the broken wall area evaluation index of each leaf gray image. The greater the boundary tearing characteristic value is, the lower the broken wall credibility of the area analysis is, and therefore the lower the broken wall area evaluation index is.

[0043] The broken wall area evaluation index and the broken wall sufficiency evaluation index of each leaf gray image are normalized to obtain the adjusted broken wall index of each leaf gray image. The greater the adjusted broken wall index is, the higher and more significant the broken wall rate is.

[0044] It should be noted that normalization is a technology known to those skilled in the art. The normalization can be hyperbolic function, linear normalization or standard normalization, and the specific normalization method is not limited herein.

[0045] S3: According to the consistency of the adjusted broken wall index of the leaf gray image of the tea leaf of the current sample at different angles and the similarity degree of the adjusted broken wall index of the current batch and the historical batch, the broken wall confidence of the tea leaf of the current batch is obtained. The final broken wall rate of the current batch is determined based on the broken wall confidence.

[0046] For actual factory processing, the production quality under the same processing technology should be relatively consistent, so the adjusted broken wall indicators estimated after rolling and twisting should be relatively consistent. When there is a high deviation from the historical situation, it means that the evaluation result of image analysis is not reliable, and the broken wall situation may be more complex. It may be due to the production residues or wear and tear caused by long-term processing, resulting in processing drift and evaluation deviation. Therefore, by analyzing the results of historical batches and multiple angles, the credibility of the current broken wall analysis is quantified.

[0047] Preferably, in the embodiments of the present application, the method for obtaining the broken wall confidence degree comprises: The average of all adjusted broken wall indicators of the sample tea leaves at all angles in each batch is taken as the average broken wall indicator of the sample tea leaves, representing the broken wall situation of each batch. Then, the average broken wall indicators of each historical batch with the same processing parameters as the current batch are obtained. In the embodiments of the present application, the average broken wall indicators of the historical batches can be the broken wall rate results under multiple test processes, or can include the final broken wall rate of iterative batches. The processing parameters include rolling and twisting time, rotation speed, pressure, moisture content, and baking temperature, etc. The tea broken wall rate should also be approximately consistent with the processing parameters.

[0048] Then, the difference between the average broken wall indicator of the current batch and the average broken wall indicator of each historical batch is calculated, and the sum of all average broken wall indicator differences is negatively correlated to map the historical confidence degree of the current batch. When the overall difference is smaller, the evaluation result of the current image analysis is more reliable through historical correlation analysis.

[0049] At the same time, when the broken wall evaluation difference under different angles is high, it reflects that the error of the evaluation result may be higher, and the confidence degree is lower. Therefore, the variance of the adjusted broken wall indicators at all angles of the current batch is negatively correlated to map the broken wall confidence degree of the sample tea leaves of the current batch in combination with the historical confidence degree. In the embodiments of the present application, the product of the value obtained by negatively correlating the variance of the adjusted broken wall indicators at all angles of the current batch and the historical confidence degree is normalized to obtain the broken wall confidence degree of the sample tea leaves of the current batch. When the broken wall confidence degree is higher, it means that the image evaluation analysis result is more reliable, and the adjusted broken wall indicator is more accurate.

[0050] The broken wall rate of the finished product after tea processing is the final required broken wall rate result. Considering that the drying process does not affect the broken wall trend, when the image analysis credibility is high, the broken wall rate of the finished product can also be obtained through image analysis to improve the efficiency of actual broken wall rate analysis.

[0051] In an embodiment of the present invention, when the wall-breaking confidence is greater than a preset confidence threshold, it indicates that the image analysis result has a high credibility, and the analysis result of the finished product after final drying is more credible. The final wall-breaking rate of the finished product after processing is obtained through image analysis, which can improve the efficiency of the wall-breaking assessment of the factory production process. Otherwise, it means that the wall-breaking situation of the sample tea leaves at this time will produce a large wall-breaking assessment error during image analysis. Therefore, multiple verifications are performed by means of a spectrometer to obtain the final wall-breaking rate of the finished product after processing, so that the wall-breaking rate analysis is more accurate and provides samples for subsequent wall-breaking analysis. In a specific embodiment of the present invention, the method of obtaining the wall-breaking rate of the finished tea leaves through image analysis is consistent with the method of adjusting the wall-breaking index through analysis after rolling, which will not be further elaborated here. The preset confidence threshold can be set to 0.6, and the specific numerical value can be adjusted by the implementer.

[0052] Considering that the drying process is time-consuming and irreversible, if an abnormal wall-breaking rate is required for adjustment and analysis, a lot of product waste will be caused. The wall-breaking index and credibility after rolling can be used for evaluation first, and the efficiency of tracing the problem can be improved. In an embodiment of the present invention, when the wall-breaking confidence is greater than the preset confidence threshold, if the adjusted wall-breaking index is less than the preset rolling wall-breaking threshold, it means that the wall breaking is not satisfied during the rolling period, and the subsequent processing quality may be lower. Therefore, a wall-breaking adjustment alarm is issued, and rolling adjustments or rework are performed, etc., to improve the quality stability of the actual production line. Among them, the preset rolling wall-breaking threshold can be set to 15%, and the specific value can be adjusted by the implementer.

[0053] In summary, the present invention obtains grayscale images of leaves at different angles after each batch of rolling, so as to comprehensively cover and analyze the wall-breaking characteristics of each surface of the leaf under the condition that the quality of the acquired image is limited. According to the phenomenon that the continuity of the leaf tissue is destroyed due to the macroscopic wall-breaking, the wall-breaking situation can be initially quantified by the area of ​​pixel value difference to obtain the initial wall-breaking index. Considering that damage features such as strong creases or fractures will be affected by the veins of the leaf itself, resulting in errors in the wall-breaking assessment, the vein direction is more uniform, the width is gradually smoothed, and the cracks are the opposite. The suspected crack edge is determined according to the consistency of the edge direction and the smoothness of the width, thereby improving the accuracy of the wall-breaking area identification. The sufficient wall-breaking situation is characterized by the influence of the uniform distribution of the suspected cracks, and the sufficient wall-breaking assessment index is obtained by the uniform distribution of the length and position of the suspected cracks, so as to improve the accuracy of the wall-breaking assessment and make the assessment more comprehensive. The wall-breaking index is adjusted by analyzing the complexity of the boundary to adjust the initial index, and the wall-breaking assessment is further adjusted based on the tearing of the leaf edge, so that the evaluation of the wall-breaking index at a single angle is more reliable. By adjusting the consistency of indicators at different angles and the degree of similarity with historical batches, confidence is obtained, which not only verifies the reliability of multi-angle data, but also associates with historical process laws to improve the credibility of the results. The final wall-breaking rate is determined based on the confidence level, which can screen reliable results and ensure that the final value accurately reflects the actual wall-breaking situation. The present invention adjusts the wall-breaking assessment taking into account the influence of leaf veins and tearing characteristics, and through real-time detection after rolling and associating with historical process reference analysis, it can promptly detect abnormal situations caused by process drift, thereby improving the accuracy of wall-breaking assessment in factory production while ensuring the stable quality of tea production.

[0054] This application also provides a tea leaf wall breaking rate detection system, please refer to Figure 3 , which shows a structural diagram of a tea wall-breaking rate detection system provided by an embodiment of the present invention. The system includes: a data acquisition module 201, a wall-breaking adjustment analysis module 202 and a final wall-breaking rate analysis module 203.

[0055] The data acquisition module 201 is used to obtain grayscale images of sample tea leaves at different angles after each batch of rolling process; The wall breaking adjustment analysis module 202 is used to obtain an initial wall breaking index of the leaf grayscale image based on the area of ​​pixel value differences in the leaf grayscale image at each angle; determine suspected crack edges based on the consistency of edge directions and the smoothness of edge widths in the leaf grayscale image; obtain a wall breaking sufficiency evaluation index of the leaf grayscale image based on the uniformity of the length and position distribution of the suspected crack edges; analyze the boundary complexity of the leaf in the leaf grayscale image, adjust the initial wall breaking index based on the wall breaking sufficiency evaluation index, and obtain an adjusted wall breaking index of the leaf grayscale image; The final wall-breaking rate analysis module 203 is used to obtain the wall-breaking confidence of the current sample tea according to the consistency of the adjusted wall-breaking index of the grayscale images of the leaves of the current sample tea at different angles, and the similarity of the adjusted wall-breaking index between the current batch and the historical batch; and obtain the final wall-breaking rate through the wall-breaking confidence judgment.

[0056] It should be noted that the system provided in the above embodiment is merely an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the tea leaf wall breakage rate detection system and the tea leaf wall breakage rate detection method embodiment provided in the above embodiment are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0057] The present application also provides a device for detecting the rate of tea leaf wall breakage. Figure 4 , which shows a structural schematic diagram of a tea wall-breaking rate detection device provided by an embodiment of the present invention. The computer device includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302. When the processor 302 executes the computer program 303, the computer device can execute any one of the tea wall-breaking rate detection methods introduced above.

[0058] An embodiment of the present application also provides a computer program product. When the computer program product is run on a computer device, the computer device can execute any one of the tea wall-breaking rate detection methods introduced above.

[0059] An embodiment of the present application also provides a computer-readable storage medium, in which computer program code is stored. When the computer program code runs on a computer device, the computer device can execute any of the tea wall breaking rate detection methods introduced above.

[0060] In the embodiments provided in the present application, it should be understood that the provided computer devices, computer program products and computer-readable storage media are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the methods provided above and will not be repeated here.

[0061] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for detecting tea leaf wall breaking rate, characterized in that: The method comprises: After each batch of rolling process, grayscale images of the sample tea leaves at different angles are obtained; On the leaf grayscale image at each angle, the initial wall breaking index of the leaf grayscale image is obtained based on the area of ​​pixel value difference; the suspected crack edge is determined based on the consistency of edge direction and the smoothness of edge width in the leaf grayscale image; the wall breaking sufficiency evaluation index of the leaf grayscale image is obtained based on the uniformity of the length and position distribution of the suspected crack edge; the boundary complexity of the leaf in the leaf grayscale image is analyzed, and the initial wall breaking index is adjusted based on the wall breaking sufficiency evaluation index to obtain the adjusted wall breaking index of the leaf grayscale image; Based on the consistency of the adjusted wall-breaking indicators of the grayscale images of the leaves of the current sample tea at different angles, and the similarity of the adjusted wall-breaking indicators of the current batch and the historical batches, the wall-breaking confidence of the current batch of sample tea is obtained; and the final wall-breaking rate of the current batch is determined based on the wall-breaking confidence.

2. The method for detecting tea leaf wall breaking rate according to claim 1, wherein: The method for obtaining the initial cell wall breaking index includes: For any leaf grayscale image, the Otsu threshold segmentation method is used to obtain the connected domains; the ratio of the total area of ​​all connected domains to the total area of ​​the leaf in the leaf grayscale image is used as the initial wall breaking index.

3. The method for detecting tea leaf wall breaking rate according to claim 1, wherein: The method for determining the suspected crack edge includes: For any leaf grayscale image, the edge line of the leaf grayscale image is obtained; for any edge line, the gradient direction difference between every two adjacent pixel points on the edge line is calculated, and the mean of all gradient direction differences is negatively correlated and mapped as the direction consistency index of the edge line; After calculating the gradient difference between every two adjacent pixels on the edge line, the standard deviation of all gradient value differences is negatively correlated to obtain the width smoothness index of the edge line; the product of the direction consistency index and the width smoothness index of the edge line is used as the edge feature value of the edge line; When the edge feature value is less than the preset feature threshold, the corresponding edge line is regarded as a suspected crack edge.

4. The method for detecting tea leaf wall breaking rate according to claim 1, wherein: The method for obtaining the cell wall breaking sufficient evaluation index includes: For any leaf grayscale image, the leaf grayscale image is divided into a preset number of sub-regions; the proportion of pixels suspected to be crack edges in each sub-region is used as the damage distribution degree of each sub-region; the variance of the damage distribution degree of all sub-regions is negatively correlated and mapped as the local distribution stability of the leaf grayscale image; The product of the mean value of the gradient values ​​of all pixels on the suspected crack edge in the leaf grayscale image and the total length of the suspected crack edge is used as the cumulative distribution degree of the leaf grayscale image; Combining the local distribution stability and cumulative distribution of the leaf grayscale image, the wall breaking sufficiency evaluation index of the leaf grayscale image is obtained.

5. The method for detecting tea leaf wall breaking rate according to claim 1, wherein: The method for obtaining the adjusted cell wall breaking index includes: The ratio of the total length of the leaf boundary to the total area of ​​the leaf in each leaf grayscale image is used as the boundary tearing feature value of each leaf grayscale image; The product of the negative correlation mapping value of the boundary tearing feature value of each leaf grayscale image and the initial wall breaking index is used as the evaluation index of the wall breaking area of ​​each leaf grayscale image; The sum of the wall-breaking area evaluation index and the wall-breaking sufficiency evaluation index of each leaf grayscale image is normalized to obtain the adjusted wall-breaking index of each leaf grayscale image.

6. A method for detecting tea leaf wall breaking rate according to claim 1, characterized in that: The method for obtaining the wall-breaking confidence includes: The average of all adjusted wall-breaking indices at all angles of the sample tea leaves in each batch is taken as the average wall-breaking index of the sample tea leaves; the average wall-breaking index of each historical batch with the same processing parameters as the current batch is obtained; After calculating the difference between the average cell-breaking index of the current batch and the average cell-breaking index of each historical batch, the sum of all the differences in the average cell-breaking indexes is negatively correlated and mapped as the historical confidence of the current batch; The variance of the adjusted wall-breaking index at all angles of the current batch is negatively correlated and mapped, and combined with the historical confidence level to obtain the wall-breaking confidence level of the current batch of sample tea leaves.

7. A method for detecting tea leaf wall breaking rate according to claim 1, characterized in that: The method of determining the final cell wall breaking rate of the current batch based on the cell wall breaking confidence level includes: When the wall-breaking confidence is greater than the preset confidence threshold, the final wall-breaking rate of the finished product after processing is obtained through image analysis; otherwise, the final wall-breaking rate of the finished product after processing is obtained through a spectrometer.

8. A method for detecting tea leaf wall breaking rate according to claim 1, characterized in that: After obtaining the tea wall breaking confidence of the current batch of sample tea leaves, the method further includes: When the wall-breaking confidence is greater than the preset confidence threshold, if the adjusted wall-breaking index is less than the preset kneading wall-breaking threshold, a wall-breaking adjustment alarm will be issued.

9. A tea leaf wall breaking rate detection system, characterized in that: The system comprises: The data acquisition module is used to obtain grayscale images of sample tea leaves at different angles after each batch of rolling processing; The wall breaking adjustment analysis module is used to obtain the initial wall breaking index of the leaf grayscale image based on the area of ​​pixel value difference in each angle of the leaf grayscale image; determine the suspected crack edge based on the consistency of edge direction and edge width smoothness in the leaf grayscale image; obtain the wall breaking sufficiency evaluation index of the leaf grayscale image based on the uniformity of the length and position distribution of the suspected crack edge; analyze the boundary complexity of the leaf in the leaf grayscale image, adjust the initial wall breaking index based on the wall breaking sufficiency evaluation index, and obtain the adjusted wall breaking index of the leaf grayscale image; The final wall-breaking rate analysis module is used to obtain the wall-breaking confidence of the current sample tea based on the consistency of the adjusted wall-breaking indicators of the grayscale images of the leaves of the current sample tea at different angles, as well as the similarity of the adjusted wall-breaking indicators between the current batch and the historical batches; and the final wall-breaking rate is obtained through the wall-breaking confidence judgment.

10. A tea leaf wall breaking rate detection device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, a method for detecting the tea wall breaking rate as described in any one of claims 1 to 8 is implemented.

Images