A quality evaluation method and system for cigar images

Abstract

The present application relates to the field of image recognition technology, and in particular to a quality evaluation method and system for cigar images, which collects the current fresh-keeping state index of the cigar image to be detected during picking and transportation, outputs multiple standard image feature information associated with the cigar image to be detected at each time in the expected cycle according to the current fresh-keeping state index and through cyclic iterative experimental analysis, and outputs multiple standard image feature information sets of the cigar image to be detected; according to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line. The present invention can accurately evaluate the quality of cigars in an all-round way, conduct all-round monitoring based on the planting, transportation, packaging and other processes of cigars, and effectively ensure the market quality of cigars.

Description

Technical Field

[0001] The present application relates to the field of image recognition technology, and in particular to a quality evaluation method and system for cigar images. Background Art

[0002] As a high-end tobacco product, the quality of cigars is directly determined by the quality of the tobacco leaves used. Among the multiple quality indicators of cigar tobacco leaves, freshness is the primary factor that determines the flavor of tobacco leaves. High-quality cigar tobacco leaves are dark green, shiny, and uniform in color.

[0003] The appearance quality of cigar core tobacco leaves directly reflects the field growth conditions, curing and fermentation status of tobacco leaves, so it has a strong guiding significance for industrial use. For the core tobacco leaves, the main indicators to be investigated are color, leaf size, smoothness, pests and diseases, leaf vein thickness, leaf flatness and redox status. In the study of tobacco science, comprehensive evaluation methods occupy a very important position. Tobacco scholars are increasingly using statistical and mathematical methods to comprehensively evaluate the quality of tobacco leaves.

[0004] Therefore, how to explore a fast and efficient quality evaluation method for cigar images is an urgent problem that technicians in this field need to solve. Summary of the invention

[0005] To achieve the above objectives, this application provides the following technical strategies:

[0006] According to a first aspect of the present invention, the present invention claims a method for evaluating the quality of a cigar image, comprising the following steps:

[0007] Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and output multiple standard image feature information associated with the cigar image to be detected at each time in the expected period through cyclic iterative experimental analysis according to the current fresh-keeping state index, and output multiple standard image feature information sets of the cigar image to be detected;

[0008] Collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period;

[0009] If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced back and analyzed, and the traceability process of the cigar image to be detected and the traceability node of the traceability process are output;

[0010] According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line.

[0011] Furthermore, the current fresh-keeping state index of the cigar image to be detected during picking and transportation is collected, and multiple standard image feature information of the cigar image to be detected within the expected period is output through cyclic iterative experimental analysis according to the current fresh-keeping state index, and multiple standard image feature information sets of the cigar image to be detected are output, specifically:

[0012] Collecting the preservation history records of the cigar image to be detected, and obtaining the historical image feature information of the cigar image to be detected when it is displayed under various historical preservation status index conditions in the preservation history records; wherein the preservation status index includes a fertilization index, a pest index, a maturity index, a transportation index, and a content index;

[0013] Performing denoising analysis on the historical image feature information and outputting a data set after denoising analysis; dividing the data set after denoising analysis into historical features and expected features; the historical features include features extracted from the historical image feature information, and the expected features are the expected freshness period of the cigar image to be detected;

[0014] configuring a cyclic iterative experiment, importing the historical features and the expected features into the cyclic iterative experiment, learning parameters of the cyclic iterative experiment by fitting the historical features and the expected features, until the parameters of the cyclic iterative experiment meet standard requirements, and outputting a tested cyclic iterative experiment;

[0015] Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, import the current fresh-keeping state index into the tested cyclic iteration experiment for analysis, output multiple standard image feature information of the cigar image to be detected within the expected period, and output multiple standard image feature information sets of the cigar image to be detected;

[0016] Among them, the image feature information includes color, tobacco leaf size, smoothness, pests and diseases, tobacco leaf vein thickness, tobacco leaf flatness and redox state.

[0017] Furthermore, multiple real image feature information sets of the cigar image to be detected are compared with the corresponding standard image feature information sets, and the preservation quality of the cigar image to be detected in the expected period is output, specifically:

[0018] Calculating the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets one by one, and comparing the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets with the standard threshold value respectively;

[0019] The real image feature information set associated with the cosine similarity value in the cigar image to be detected being less than or equal to the standard threshold is marked as a deteriorated image feature information set; the real image feature information set associated with the cosine similarity value in the cigar image to be detected being greater than the standard threshold is marked as a fresh image feature information set;

[0020] A statistical analysis is performed on the deteriorated image feature information set and the fresh image feature information set of the cigar image to be detected in the expected period; if there is no deteriorated image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is fresh; if there is a deteriorated image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is not fresh.

[0021] Furthermore, if the preservation quality of the cigar image to be detected is not fresh, a traceability analysis is performed on the cigar image to be detected, and the traceability process of the cigar image to be detected and the traceability nodes of the traceability process are output, specifically:

[0022] If the preservation quality of the cigar image to be detected is not fresh, then a spoiled image feature information set is collected, and a standard image feature information set associated with the spoiled image feature information set is collected;

[0023] configuring a relational database, and generating a first feature change table in the relational database according to the deteriorated image feature information set, and generating a second feature change table in the relational database according to the standard image feature information set associated with the deteriorated image feature information set;

[0024] Comparing and analyzing the first feature change table and the second feature change table in the relational database, outputting tuples that do not overlap between the first feature change table and the second feature change table, and collecting the period associated with the tuples that do not overlap between the first feature change table and the second feature change table;

[0025] Mark the period associated with the tuples that do not overlap in the first feature change table and the second feature change table as a deterioration period; and obtain the real change attributes of the cigar image to be detected in the deterioration period in the preservation history record of the cigar image to be detected; wherein the real change attributes include real color, real smoothness, real pests and diseases, real thickness of tobacco leaf veins, real flatness of tobacco leaves, and real oxidation-reduction state;

[0026] Configure a deep learning model, import the real change attributes of the cigar image to be tested during the deterioration period into the deep learning model for traceability analysis, and output the traceability process and traceability nodes of the traceability process of the cigar image to be tested.

[0027] Furthermore, a deep learning model is configured to import the real change attributes of the cigar image to be detected during the deterioration period into the deep learning model for traceability analysis, and the traceability process and traceability nodes of the traceability process of the cigar image to be detected are output, specifically:

[0028] Collect various processes and their possible traceability nodes, and collect the probabilistic dependencies between various processes and their possible traceability nodes; collect the traceability history features of various traceability nodes of the cigar image to be detected in the preserved historical records of the cigar image to be detected;

[0029] The deep learning network is configured according to the probabilistic dependencies between various processes and their possible traceability nodes. The nodes in the deep learning network represent processes and traceability nodes, and the edges represent the probabilistic dependencies between nodes.

[0030] The traceability history features of various traceability nodes of the cigar image to be detected are imported into the deep learning network, and the traceability history features are regarded as a number of chromosomes, and a chromosome set is formed according to the number of chromosomes; the mutation probability of each chromosome in the chromosome set is calculated based on the likelihood of the traceability history features;

[0031] For each chromosome in the chromosome set, the associated fitness value is calculated according to the mutation probability represented by it; the chromosomes with fitness values ​​greater than the standard value are retained in the chromosome set, and the chromosomes with fitness values ​​less than or equal to the standard value are removed from the chromosome set, and the final population is output;

[0032] Calculate the mutation probability of each remaining chromosome in the final population, output the optimal mutation probability, apply the optimal mutation probability to each edge of the deep learning network, and output the deep learning model;

[0033] The real change attributes of the cigar image to be tested during the deterioration period are imported into the deep learning model for traceability analysis, and the traceability process of the cigar image to be tested and the traceability nodes of the traceability process are output.

[0034] Furthermore, according to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line, specifically:

[0035] Formulate in advance the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected, configure a strategy mapping table, and import the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected into the strategy mapping table;

[0036] Collecting the traceability nodes of the traceability process of the cigar image to be detected, and matching the cigar image to be detected in the strategy mapping table according to the traceability nodes of the traceability process of the cigar image to be detected to adjust the adjustment strategy;

[0037] The positioning data of the traceability process of the cigar image to be detected is collected, the positioning data of the traceability process of the cigar image to be detected is segmented with the adjustment strategy for adjusting the cigar image to be detected, an evaluation data packet is output, and the evaluation data packet is transmitted to the cigar production line.

[0038] According to a second aspect of the present invention, the present invention claims protection for a quality evaluation system for cigar images, the quality evaluation system for cigar images comprising a memory and a processor, the memory storing a quality evaluation method program for cigar images, and when the quality evaluation method program for cigar images is executed by the processor, the following steps are implemented:

[0039] Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and output multiple standard image feature information associated with the cigar image to be detected at each time in the expected period through cyclic iterative experimental analysis according to the current fresh-keeping state index, and output multiple standard image feature information sets of the cigar image to be detected;

[0040] Collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period;

[0041] If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced back and analyzed, and the traceability process of the cigar image to be detected and the traceability node of the traceability process are output;

[0042] According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line;

[0043] The quality assessment system for cigar images is used to execute the quality assessment method for cigar images.

[0044] The present application relates to the field of image recognition technology, and in particular to a quality evaluation method and system for cigar images, which collects the current fresh-keeping state index of the cigar image to be detected during picking and transportation, outputs multiple standard image feature information associated with the cigar image to be detected at each time in the expected cycle according to the current fresh-keeping state index and through cyclic iterative experimental analysis, and outputs multiple standard image feature information sets of the cigar image to be detected; according to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line. The present invention can accurately evaluate the quality of cigars in an all-round way, conduct all-round monitoring based on the planting, transportation, packaging and other processes of cigars, and effectively ensure the market quality of cigars. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flowchart of a quality evaluation method for cigar images claimed in an embodiment of the present application;

[0046] Figure 2 A second working flow chart of a quality evaluation method for cigar images claimed in an embodiment of the present application;

[0047] Figure 3 This is a structural diagram of a quality evaluation system for cigar images as claimed in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical strategies in the embodiments of the present application through the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0049] The terms "first", "second" and "third" in this application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined as "first", "second" and "third" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. All directional indications (such as up, down, left, right, front, back...) in the embodiments of this application are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0050] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described by the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase at each location in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0051] like Figure 1 As shown, the first aspect of the present invention discloses a quality evaluation method for a cigar image, comprising the following steps:

[0052] S102: collecting the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and outputting a plurality of standard image feature information associated with the cigar image to be detected at each time in an expected period according to the current fresh-keeping state index through cyclic iterative experimental analysis, and outputting a plurality of standard image feature information sets of the cigar image to be detected;

[0053] S104: collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period;

[0054] S106: If the freshness quality of the cigar image to be detected is not fresh, perform a traceability analysis on the cigar image to be detected, and output a traceability process and a traceability node of the traceability process of the cigar image to be detected;

[0055] S108: Match and output a corresponding adjustment strategy in the strategy mapping table according to the traceability node of the traceability process of the cigar image to be detected, and segment the adjustment strategy and the positioning data of the traceability process and transmit them to the cigar production line.

[0056] like Figure 2 As shown, further, in a preferred embodiment of the present invention, the current fresh-keeping state index of the cigar image to be detected during picking and transportation is collected, and multiple standard image feature information of the cigar image to be detected within the expected period is output through cyclic iterative experimental analysis according to the current fresh-keeping state index, and multiple standard image feature information sets of the cigar image to be detected are output, specifically:

[0057] S202: Collecting the preservation history records of the cigar image to be detected, and obtaining the historical image feature information of the cigar image to be detected when it is displayed under various historical preservation status index conditions in the preservation history records; wherein the preservation status index includes a fertilization index, a pest index, a maturity index, a transportation index, and a content index;

[0058] S204: performing denoising analysis on the historical image feature information, and outputting a data set after denoising analysis; dividing the data set after denoising analysis into historical features and expected features; the historical features include features extracted from the historical image feature information, and the expected features are the expected freshness period of the cigar image to be analyzed;

[0059] S206: configuring a loop iterative experiment, importing the historical features and the expected features into the loop iterative experiment, learning parameters of the loop iterative experiment by fitting the historical features and the expected features, until the parameters of the loop iterative experiment meet the standard requirements, and outputting the tested loop iterative experiment;

[0060] S208: collecting the current fresh-keeping state index of the cigar image to be detected during picking and transportation, importing the current fresh-keeping state index into the tested cyclic iteration experiment for analysis, outputting a plurality of standard image feature information of the cigar image to be detected within the expected period, and outputting a plurality of standard image feature information sets of the cigar image to be detected;

[0061] Among them, the image feature information includes color, tobacco leaf size, smoothness, pests and diseases, tobacco leaf vein thickness, tobacco leaf flatness and redox state.

[0062] Redox reaction refers to the oxidation of organic matter contained in tobacco leaves under natural conditions with oxygen in the air under the catalytic action of inorganic elements (Fe, Mg, etc.), resulting in changes in chemical substances in tobacco. The stacking and fermentation of cigar tobacco leaves involves the enzymatic action of a variety of microorganisms and is an efficient and environmentally friendly process for degrading macromolecular organic compounds. The chemical reaction effects mainly include redox reactions and Maillard reactions (non-enzymatic browning reactions). Among them, the Maillard reaction is a browning reaction of compounds containing free amino groups, reducing sugars or carbonyl compounds at room temperature or when heated.

[0063] Image feature recognition of redox reactions can help to assist in controlling the storage conditions of tobacco leaves.

[0064] Iterative experiment is a statistical method used to analyze or estimate the relationship between two or more variables, where one variable is independent and the other one or more variables are dependent. Iterative experiment assumes that there is semantic similarity between the dependent variable and one or more independent variables. Through the above steps, the regression model can be tested based on the denoised data and used to analyze the freshness period of the cigar image to be tested in the expected period.

[0065] Furthermore, in a preferred embodiment of the present invention, a plurality of real image feature information sets of the cigar image to be detected are compared with the corresponding standard image feature information sets, and the preservation quality of the cigar image to be detected in the expected period is output, specifically:

[0066] Calculating the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets one by one, and comparing the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets with the standard threshold value respectively;

[0067] The real image feature information set associated with the cosine similarity value in the cigar image to be detected being less than or equal to the standard threshold is marked as a deteriorated image feature information set; the real image feature information set associated with the cosine similarity value in the cigar image to be detected being greater than the standard threshold is marked as a fresh image feature information set;

[0068] A statistical analysis is performed on the deteriorated image feature information set and the fresh image feature information set of the cigar image to be detected in the expected period; if there is no deteriorated image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is fresh; if there is a deteriorated image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is not fresh.

[0069] Among them, the cosine similarity value is a statistic used to measure the strength and direction of the semantic similarity between two variables.

[0070] It should be noted that the real image feature information associated with the cigar image to be detected at each time in the expected period is collected by an acoustic sensor, and then the cosine similarity values ​​between multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets are calculated one by one, such as calculating the cosine similarity value between the standard smoothness and the real smoothness of the cigar image to be detected within the standard time. Then, the real image feature information set associated with the cosine similarity value less than or equal to the standard threshold in the cigar image to be detected is marked as a deteriorated image feature information set, and the real image feature information set associated with the cosine similarity value greater than the standard threshold in the cigar image to be detected is marked as a fresh image feature information set. If there is no deteriorated image feature information set in the cigar image to be detected within the expected period, it means that the fresh-keeping quality of the cigar image to be detected in the expected period is fresh; if there is a deteriorated image feature information set in the cigar image to be detected within the expected period, it means that the fresh-keeping quality of the cigar image to be detected in the expected period is not fresh.

[0071] Furthermore, in a preferred embodiment of the present invention, if the freshness quality of the cigar image to be detected is not fresh, a traceability analysis is performed on the cigar image to be detected, and a traceability process and a traceability node of the traceability process of the cigar image to be detected are output, specifically:

[0072] If the preservation quality of the cigar image to be detected is not fresh, then a spoiled image feature information set is collected, and a standard image feature information set associated with the spoiled image feature information set is collected;

[0073] configuring a relational database, and generating a first feature change table in the relational database according to the deteriorated image feature information set, and generating a second feature change table in the relational database according to the standard image feature information set associated with the deteriorated image feature information set;

[0074] Comparing and analyzing the first feature change table and the second feature change table in the relational database, outputting tuples that do not overlap between the first feature change table and the second feature change table, and collecting the period associated with the tuples that do not overlap between the first feature change table and the second feature change table;

[0075] Mark the period associated with the tuples that do not overlap in the first feature change table and the second feature change table as a deterioration period; and obtain the real change attributes of the cigar image to be detected in the deterioration period in the preservation history record of the cigar image to be detected; wherein the real change attributes include real color, real smoothness, real pests and diseases, real thickness of tobacco leaf veins, real flatness of tobacco leaves, and real oxidation-reduction state;

[0076] Configure a deep learning model, import the real change attributes of the cigar image to be tested during the deterioration period into the deep learning model for traceability analysis, and output the traceability process and traceability nodes of the traceability process of the cigar image to be tested.

[0077] Among them, deep learning networks are a powerful analytical model that can capture the probabilistic dependencies between variables. In traceability analysis, there are complex probabilistic relationships between various traceability nodes and symptoms. Deep learning networks can effectively model these relationships, thereby realizing traceability analysis.

[0078] It should be noted that the acquisition time node is used as the horizontal axis and the image feature information is used as the vertical axis to configure the relational database, and a first feature change table is generated in the relational database according to the deteriorated image feature information set, and a second feature change table is generated in the relational database according to the standard image feature information set associated with the deteriorated image feature information set; the period associated with the tuples that do not overlap between the first feature change table and the second feature change table is marked as the deterioration period, that is, the period of deterioration of the image feature information of the cigar image to be detected during the life cycle. Then, the real change attributes of the cigar image to be detected during the deterioration period are collected, such as the real color, real smoothness, and real pests and diseases of the cigar leaves during this period. Finally, the real change attributes of the cigar image to be detected during the deterioration period are imported into the deep learning model for analysis and output of the traceability process and the traceability nodes of the traceability process of the cigar image to be detected.

[0079] Furthermore, in a preferred embodiment of the present invention, a deep learning model is configured to import the real change attributes of the cigar image to be detected during the deterioration period into the deep learning model for traceability analysis, and the traceability process of the cigar image to be detected and the traceability nodes of the traceability process are output, specifically:

[0080] Collect various processes and their possible traceability nodes, and collect the probabilistic dependencies between various processes and their possible traceability nodes; collect the traceability history features of various traceability nodes of the cigar image to be detected in the preserved historical records of the cigar image to be detected;

[0081] The deep learning network is configured according to the probabilistic dependencies between various processes and their possible traceability nodes. The nodes in the deep learning network represent processes and traceability nodes, and the edges represent the probabilistic dependencies between nodes.

[0082] The traceability history features of various traceability nodes of the cigar image to be detected are imported into the deep learning network, and the traceability history features are regarded as a number of chromosomes, and a chromosome set is formed according to the number of chromosomes; the mutation probability of each chromosome in the chromosome set is calculated based on the likelihood of the traceability history features;

[0083] For each chromosome in the chromosome set, the associated fitness value is calculated according to the mutation probability represented by it; the chromosomes with fitness values ​​greater than the standard value are retained in the chromosome set, and the chromosomes with fitness values ​​less than or equal to the standard value are removed from the chromosome set, and the final population is output;

[0084] Calculate the mutation probability of each remaining chromosome in the final population, output the optimal mutation probability, apply the optimal mutation probability to each edge of the deep learning network, and output the deep learning model;

[0085] The real change attributes of the cigar image to be tested during the deterioration period are imported into the deep learning model for traceability analysis, and the traceability process of the cigar image to be tested and the traceability nodes of the traceability process are output.

[0086] It should be noted that the variables involved are first defined, including various cigar tobacco leaf parts and their possible traceability nodes, as well as the traceability history features of various traceability nodes of the cigar tobacco image to be detected in the preservation history record of the cigar tobacco image to be detected. The traceability history features of cigar tobacco leaves refer to the data related to the state, value and health status of cigar tobacco leaves collected by various sensors and monitoring equipment during the life cycle of cigar tobacco leaves. After collecting a large number of traceability history features, these data contain many aspects of the preservation quality of cigar tobacco leaves. However, not all traceability history features are equally important for traceability analysis, so it is necessary to screen out useful data from a large number of data for analysis and analysis. Therefore, by introducing genetic algorithms to screen out useful traceability history features, the efficiency, accuracy and interpretability of traceability analysis and diagnosis can be improved, while saving resources and simplifying the model building process. Specifically, a set of initial solutions is randomly generated as a set of chromosomes, each chromosome represents the collected traceability history features, and then the fitness function is defined in advance. The fitness function is used to evaluate the fitness value of each chromosome, and the selection is made according to the fitness value of the chromosome, so as to screen out the optimal chromosome, that is, the traceability history features that are important for traceability analysis. When the genetic algorithm terminates, the chromosome with high fitness is selected as the best solution, which represents the optimal mutation probability estimate, and then this optimal mutation probability is applied to the deep learning network to generate a deep learning model. Through the above steps, using the genetic algorithm to estimate the mutation probability of each edge in the deep learning network using historical traceability data can not only improve the accuracy of parameter estimation, but also effectively explore the solution space and find a network structure that is more in line with the data distribution. It is especially suitable for analyzing large-scale and complex network structures, and can quickly and accurately screen out useful data from a large number of data, which can improve the efficiency of cigar tobacco leaf evaluation and the reliability of evaluation results.

[0087] Furthermore, in a preferred embodiment of the present invention, the corresponding adjustment strategy is matched and output in the strategy mapping table according to the traceability node of the traceability process of the cigar image to be detected, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line, specifically:

[0088] Formulate in advance the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected, configure a strategy mapping table, and import the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected into the strategy mapping table;

[0089] Collecting the traceability nodes of the traceability process of the cigar image to be detected, and matching the cigar image to be detected in the strategy mapping table according to the traceability nodes of the traceability process of the cigar image to be detected to adjust the adjustment strategy;

[0090] The positioning data of the traceability process of the cigar image to be detected is collected, the positioning data of the traceability process of the cigar image to be detected is segmented with the adjustment strategy for adjusting the cigar image to be detected, an evaluation data packet is output, and the evaluation data packet is transmitted to the cigar production line.

[0091] It should be noted that the technical personnel formulate in advance the adjustment strategies associated with various traceability nodes in each process of the cigar image to be detected, so as to configure the output strategy mapping table, and then when the traceability node of the traceability process of the cigar image to be detected is output, the corresponding adjustment strategy is matched in the strategy mapping table, and the positioning data of the traceability process of the cigar image to be detected is divided with the adjustment strategy for adjusting the cigar image to be detected, and the evaluation data packet is output for reference by the adjustment personnel.

[0092] In summary, by analyzing the image feature information and fresh-keeping period of cigar tobacco leaves through this method, the traceability node and location can be quickly determined, which helps to reduce the time and cost of traceability investigation and can effectively improve the evaluation efficiency and reliability of cigar tobacco leaves.

[0093] In addition, the method further comprises the following steps:

[0094] The traceability process and traceability nodes of the traceability process for collecting the cigar image to be tested, and the process functionally associated with the traceability process in the collected cigar image to be tested;

[0095] The process that is functionally related to the traceability process is regarded as the first independent event, and the traceability node of the traceability process is regarded as the second independent event;

[0096] configuring a correlation analysis model, mapping the first independent event and the second independent event to the correlation analysis model, using a particle swarm algorithm to perform sorting optimization analysis on the first independent event and the second independent event in the correlation analysis model, and outputting an optimal correlation analysis model;

[0097] Inputting the traceability process of the cigar image to be detected into the optimal correlation analysis model to perform correlation analysis and output multiple correlation scores;

[0098] Compare multiple relevance scores with the standard relevance score; collect processes with functional associations whose relevance scores are greater than the standard relevance scores, and mark the processes with functional associations whose relevance scores are greater than the standard relevance scores as traceability processes.

[0099] It should be noted that in cigar tobacco leaves, processes that are functionally associated with a certain process usually refer to other processes that have a close functional connection with the process in the cigar tobacco leaf system. These processes are usually displayed in coordination with each other to achieve the fresh life cycle and value optimization of cigar tobacco leaves. After analyzing and outputting the traceability process of the cigar image to be tested and the traceability node of the traceability process, this step further analyzes whether the process that is functionally associated with the traceability process has also been traced, thereby realizing the function of comprehensive evaluation of cigar tobacco leaves.

[0100] In addition, the method further comprises the following steps:

[0101] Collect the standard planting production index line change graph of the cigar image to be detected at different planting production stages through the big data network, configure the second strategy mapping table, and import the standard planting production index line change graph of the cigar image to be detected at different planting production stages into the second strategy mapping table;

[0102] Collecting the actual planting and production time of the cigar image to be detected, and matching and outputting a line chart of the standard planting and production index of the cigar image to be detected in the second strategy mapping table according to the actual planting and production time of the cigar image to be detected;

[0103] Collecting the expected planting production values ​​of the cigar images to be detected during picking and transportation at multiple standard position nodes, and configuring a line chart of the actual planting production index of the cigar images to be detected during picking and transportation according to the expected planting production values;

[0104] Calculate the similarity between the actual planting production index line change graph of the cigar image to be detected during picking and transportation and the matching output standard planting production index line change graph by using the cosine similarity algorithm; determine whether the similarity is greater than the standard similarity;

[0105] If the similarity is less than or equal to the standard similarity, feature extraction and analysis are performed on the real planting production index line change graph of the cigar image to be detected during picking and transportation and the matched output standard planting production index line change graph, and a first planting production index line change feature change table and a second planting production index line change feature change table are output;

[0106] configuring a virtual space, importing the first planting production index broken line change characteristic change table and the second planting production index broken line change characteristic change table into the virtual space, and performing registration analysis on the first planting production index broken line change characteristic change table and the second planting production index broken line change characteristic change table;

[0107] After the registration is completed, the curve parts that overlap with the first planting production index line change feature change table and the second planting production index line change feature change table are removed, and the non-overlapping curve parts are retained, and the electromagnetic distribution deviation map is output; the wear and deterioration position of the cigar image to be detected is determined according to the electromagnetic distribution deviation map, and the wear and deterioration position is transmitted to the cigar production line.

[0108] A chromosome represents a combination of planting preservation conditions and transportation preservation conditions.

[0109] Set the population size to N, for example N = 100.

[0110] N chromosomes are randomly generated within the value range of each variable, that is, N groups of combinations of planting and preservation conditions and transportation and preservation conditions are randomly generated.

[0111] For each chromosome in the population (i.e. each set of preservation conditions), calculate its corresponding deterioration rate F C .

[0112] The fitness function is defined as the metamorphic rate F C The reciprocal of Fitness=1 / F C The purpose is to make chromosomes with slower deterioration rates have higher fitness.

[0113] According to the fitness value of each chromosome, the probability of its being selected is calculated.

[0114] Using roulette wheel selection, chromosome i The probability of selection ;

[0115] in, Fitness i is the fitness value of the i-th chromosome; Represents the sum of the fitness values ​​of all individuals in the population. This formula shows that the larger the individual fitness value, the greater the probability of being selected.

[0116] According to the calculated selection probability, N chromosomes are selected from the population to form a new population. During the selection process, chromosomes with high fitness have a greater probability of being selected.

[0117] Set the crossover probability to P c ,For example P c = 0.8.

[0118] For each pair of chromosomes in the new population (randomly paired), the crossover probability P c Determines whether to perform crossover operation.

[0119] If a crossover operation is performed, a crossover point is randomly selected and some gene fragments of the two chromosomes are exchanged at the crossover point to obtain two new chromosomes.

[0120] Set the mutation probability to P m ,For example P m = 0.1.

[0121] For each chromosome in the new population, the mutation probability P m Decide whether to perform a mutation operation.

[0122] If a mutation operation is performed, a mutation point (i.e., a preservation condition variable) is randomly selected, and a new value is randomly generated within the value range of the variable to replace the original value.

[0123] You can set the maximum number of iterations Max Iter As a termination condition, such as Max Iter = 500.

[0124] Alternatively, the iteration can be stopped when the average fitness of the population has not improved significantly in several consecutive generations.

[0125] If the termination condition is met, the iteration stops, and the combination of preservation conditions represented by the chromosome with the highest fitness in the current population is the optimal combination of planting preservation conditions and transportation preservation conditions obtained by optimization.

[0126] If the termination condition is not met, the selection operation is returned to continue the next round of iteration.

[0127] Based on the above content, what is finally obtained is the optimal combination of planting preservation conditions and transportation preservation conditions to achieve the best preservation effect under the best preservation conditions.

[0128] like Figure 3 As shown, the second aspect of the present invention discloses a quality evaluation system for cigar images, the quality evaluation system for cigar images comprises a memory 20 and a processor 30, the memory 20 stores a quality evaluation method program for cigar images, and when the quality evaluation method program for cigar images is executed by the processor 30, the following steps are implemented:

[0129] Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and output multiple standard image feature information associated with the cigar image to be detected at each time in the expected period through cyclic iterative experimental analysis according to the current fresh-keeping state index, and output multiple standard image feature information sets of the cigar image to be detected;

[0130] Collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period;

[0131] If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced back and analyzed, and the traceability process of the cigar image to be detected and the traceability node of the traceability process are output;

[0132] According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line.

[0133] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be passed or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0134] In addition, each functional unit in each embodiment of the present application can be integrated into an analysis unit, each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the content of the specification and drawings of this application, or directly or indirectly used in other related technical fields, is also included in the patent protection scope of the present application.

[0135] The specific implementation methods of the invention are described in detail above, but they are only examples, and the present application is not limited to the specific implementation methods described above. For those skilled in the art, any equivalent modification or substitution of the invention is also within the scope of the present application, and therefore, the equalization, modification, and improvement made without departing from the spirit and principle of the present application should be included in the scope of the present application.

Claims

1. A quality evaluation method for cigar images, characterized in that: The following steps are involved: Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and output multiple standard image feature information associated with the cigar image to be detected at each time in the expected period through cyclic iterative experimental analysis according to the current fresh-keeping state index, and output multiple standard image feature information sets of the cigar image to be detected; Collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period; If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced back and analyzed, and the traceability process of the cigar image to be detected and the traceability node of the traceability process are output; According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line; The method collects the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and outputs multiple standard image feature information of the cigar image to be detected within an expected period according to the current fresh-keeping state index and through cyclic iterative experimental analysis, and outputs multiple standard image feature information sets of the cigar image to be detected, specifically: Collecting the preservation history records of the cigar image to be detected, and obtaining the historical image feature information of the cigar image to be detected when it is displayed under various historical preservation status index conditions in the preservation history records; wherein the preservation status index includes a fertilization index, a pest index, a maturity index, a transportation index, and a content index; Performing denoising analysis on the historical image feature information and outputting a data set after denoising analysis; dividing the data set after denoising analysis into historical features and expected features; the historical features include features extracted from the historical image feature information, and the expected features are the expected freshness period of the cigar image to be detected; configuring a cyclic iterative experiment, importing the historical features and the expected features into the cyclic iterative experiment, learning parameters of the cyclic iterative experiment by fitting the historical features and the expected features, until the parameters of the cyclic iterative experiment meet standard requirements, and outputting a tested cyclic iterative experiment; Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, import the current fresh-keeping state index into the tested cyclic iteration experiment for analysis, output multiple standard image feature information of the cigar image to be detected within the expected period, and output multiple standard image feature information sets of the cigar image to be detected; Among them, the image feature information includes color, tobacco leaf size, smoothness, pests and diseases, tobacco leaf vein thickness, tobacco leaf flatness and redox state.

2. The quality evaluation method for cigar images according to claim 1, characterized in that: Compare multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and output the preservation quality of the cigar image to be detected in the expected period, specifically: Calculating the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets one by one, and comparing the cosine similarity values ​​between the multiple real image feature information sets of the cigar image to be detected and the corresponding standard image feature information sets with the standard threshold value respectively; The real image feature information set associated with the cosine similarity value in the cigar image to be detected being less than or equal to the standard threshold is marked as a deteriorated image feature information set; the real image feature information set associated with the cosine similarity value in the cigar image to be detected being greater than the standard threshold is marked as a fresh image feature information set; Statistically analyzing the deterioration image feature information set and the fresh image feature information set of the cigar image to be detected in the expected period; If there is no deterioration image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is fresh; if there is a deterioration image feature information set, it means that the preservation quality of the cigar image to be detected in the expected period is not fresh.

3. The quality evaluation method for cigar images according to claim 2, characterized in that: If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced and analyzed, and the traceability process and the traceability node of the traceability process of the cigar image to be detected are output, specifically: If the preservation quality of the cigar image to be detected is not fresh, then a spoiled image feature information set is collected, and a standard image feature information set associated with the spoiled image feature information set is collected; configuring a relational database, and generating a first feature change table in the relational database according to the deteriorated image feature information set, and generating a second feature change table in the relational database according to the standard image feature information set associated with the deteriorated image feature information set; Comparing and analyzing the first feature change table and the second feature change table in the relational database, outputting tuples that do not overlap between the first feature change table and the second feature change table, and collecting the period associated with the tuples that do not overlap between the first feature change table and the second feature change table; Mark the period associated with the tuples that do not overlap in the first feature change table and the second feature change table as a deterioration period; and obtain the real change attributes of the cigar image to be detected in the deterioration period in the preservation history record of the cigar image to be detected; wherein the real change attributes include real color, real smoothness, real pests and diseases, real thickness of tobacco leaf veins, real flatness of tobacco leaves, and real oxidation-reduction state; Configure a deep learning model, import the real change attributes of the cigar image to be tested during the deterioration period into the deep learning model for traceability analysis, and output the traceability process and traceability nodes of the traceability process of the cigar image to be tested.

4. The quality assessment method for cigar images according to claim 3, characterized in that: Configure a deep learning model, import the real change attributes of the cigar image to be tested during the deterioration period into the deep learning model for traceability analysis, and output the traceability process and traceability nodes of the traceability process of the cigar image to be tested, specifically: Collect various processes and their possible traceability nodes, and collect the probabilistic dependencies between various processes and their possible traceability nodes; collect the traceability history features of various traceability nodes of the cigar image to be detected in the preserved historical records of the cigar image to be detected; The deep learning network is configured according to the probabilistic dependencies between various processes and their possible traceability nodes. The nodes in the deep learning network represent processes and traceability nodes, and the edges represent the probabilistic dependencies between nodes. The traceability history features of various traceability nodes of the cigar image to be detected are imported into the deep learning network, and the traceability history features are regarded as a number of chromosomes, and a chromosome set is formed according to the number of chromosomes; the mutation probability of each chromosome in the chromosome set is calculated based on the likelihood of the traceability history features; For each chromosome in the chromosome set, the associated fitness value is calculated according to the mutation probability represented by it; the chromosomes with fitness values ​​greater than the standard value are retained in the chromosome set, and the chromosomes with fitness values ​​less than or equal to the standard value are removed from the chromosome set, and the final population is output; Calculate the mutation probability of each remaining chromosome in the final population, output the optimal mutation probability, apply the optimal mutation probability to each edge of the deep learning network, and output the deep learning model; The real change attributes of the cigar image to be tested during the deterioration period are imported into the deep learning model for traceability analysis, and the traceability process of the cigar image to be tested and the traceability nodes of the traceability process are output.

5. The quality assessment method for cigar images according to claim 1, characterized in that: According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line, specifically: Formulate in advance the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected, configure a strategy mapping table, and import the adjustment strategies for various traceability node associations occurring in each process of the cigar image to be detected into the strategy mapping table; Collecting the traceability nodes of the traceability process of the cigar image to be detected, and matching the cigar image to be detected in the strategy mapping table according to the traceability nodes of the traceability process of the cigar image to be detected to adjust the adjustment strategy; The positioning data of the traceability process of the cigar image to be detected is collected, the positioning data of the traceability process of the cigar image to be detected is segmented with the adjustment strategy for adjusting the cigar image to be detected, an evaluation data packet is output, and the evaluation data packet is transmitted to the cigar production line.

6. A quality evaluation system for cigar images, characterized in that: The cigar image quality evaluation system includes a memory and a processor. The memory stores a cigar image quality evaluation method program. When the cigar image quality evaluation method program is executed by the processor, the following steps are implemented: Collect the current fresh-keeping state index of the cigar image to be detected during picking and transportation, and output multiple standard image feature information associated with the cigar image to be detected at each time in the expected period through cyclic iterative experimental analysis according to the current fresh-keeping state index, and output multiple standard image feature information sets of the cigar image to be detected; Collecting real image feature information associated with the cigar image to be detected at each time in the expected period, and outputting multiple real image feature information sets of the cigar image to be detected; comparing the multiple real image feature information sets of the cigar image to be detected with the corresponding standard image feature information sets, and outputting the freshness quality of the cigar image to be detected in the expected period; If the freshness quality of the cigar image to be detected is not fresh, then the cigar image to be detected is traced back and analyzed, and the traceability process of the cigar image to be detected and the traceability node of the traceability process are output; According to the traceability node of the traceability process of the cigar image to be detected, the corresponding adjustment strategy is matched and output in the strategy mapping table, and the adjustment strategy and the positioning data of the traceability process are segmented and transmitted to the cigar production line; The quality assessment system for cigar images is used to execute the quality assessment method for cigar images as described in any one of claims 1 to 5.

Images