Ink aging degree detection method and system based on oxidation kinetics

Abstract

The application discloses an ink aging degree detection method and system based on oxidation kinetics, relates to the field of ink detection, and aims to solve the technical problems of strong subjectivity, insufficient accuracy and reliability of ink aging degree detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ink detection, and in particular to an ink aging degree detection method and system based on oxidation kinetics. BACKGROUND

[0002] In the fields of cultural relic protection, judicial authentication, and archive management, accurately determining the aging degree of ink is crucial. The aging degree of ink not only affects the preservation state and value assessment of cultural relics, but also directly relates to the accuracy of judicial authentication and the reliability of archives. However, traditional methods for detecting the aging degree of ink mostly rely on manual observation and experience-based judgment. This method is not only time-consuming and labor-intensive, but also susceptible to human factors, resulting in a high degree of subjectivity and uncertainty in the detection results. Ink can age due to oxidation kinetics and other reasons, leading to the volatilization and penetration of ink components within the paper. Over time and with environmental changes, the aging process of ink exhibits complex and variable characteristics. Traditional detection methods often struggle to accurately capture these subtle changes, thus failing to provide precise aging degree information and affecting the stability and reliability of the detection. SUMMARY

[0003] The present application provides an ink aging degree detection method and system based on oxidation kinetics to address the technical problems of strong subjectivity, insufficient accuracy, and reliability in detecting the aging degree of ink in the prior art.

[0004] The technical solutions of the present application to solve the above technical problems are as follows:

[0005] In a first aspect, the present application provides an ink aging degree detection method based on oxidation kinetics, which includes: dividing a target handwriting to obtain a plurality of handwriting regions, collecting spectral parameters and images of the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array; predicting the aging of the spectral parameter array to obtain a spectral aging parameter array, identifying the handwriting image array to obtain a handwriting structure parameter array, and classifying to obtain an image aging parameter array; performing relative error analysis and global error analysis based on the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, performing sample error analysis based on the target handwriting to obtain a sample error coefficient, calculating the spectral aging parameter array and the image aging parameter array to obtain an ink aging parameter, performing error compensation based on the relative error coefficient, the global error coefficient, and the sample error coefficient to obtain a compensated ink aging parameter interval as the ink aging degree detection result.

[0006] In a second aspect, the present application provides an ink aging degree detection system based on oxidation kinetics, which comprises: a handwriting division module, configured to divide target handwriting to obtain a plurality of handwriting regions, and configured to perform spectral parameter acquisition and image acquisition on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array; a parameter array acquisition module, configured to perform aging prediction on the spectral parameter array to obtain a spectral aging parameter array, and configured to perform identification on the handwriting image array to obtain a handwriting structure parameter array, and configured to classify to obtain an image aging parameter array; a sample error analysis module, configured to perform relative error analysis and global error analysis according to the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, and configured to perform sample error analysis according to the target handwriting to obtain a sample error coefficient; and a detection result generation module, configured to perform calculation according to the spectral aging parameter array and the image aging parameter array to obtain an ink aging parameter, and configured to perform error compensation according to the relative error coefficient, the global error coefficient and the sample error coefficient to obtain a compensated ink aging parameter interval as an ink aging degree detection result.

[0007] The present application has the following beneficial effects: by collecting spectral parameters and image information of handwriting, combining aging prediction and identification technology, spectral aging parameters and image aging parameters are obtained. Further, by relative error analysis, global error analysis and sample error analysis, the obtained parameters are compensated for errors, and finally accurate ink aging parameter intervals are obtained as detection results, effectively improving the accuracy and reliability of ink aging degree detection. BRIEF DESCRIPTION OF DRAWINGS

[0008] Figure 1 The present application provides a flowchart of an ink aging degree detection method based on oxidation kinetics.

[0009] Figure 2 The present application provides a structural diagram of an ink aging degree detection system based on oxidation kinetics.

[0010] The reference signs are explained as follows: handwriting division module 11, parameter array acquisition module 12, sample error analysis module 13, and detection result generation module 14. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0012] In the description of the present application, the terms "first", "second" are only for descriptive purposes and cannot be understood to indicate or imply relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly specified.

[0013] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can recognize that the present application can be implemented without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope consistent with the principles and characteristics disclosed.

[0014] Embodiment one:

[0015] As Figure 1 shown, the present application provides an ink aging degree detection method based on oxidation kinetics, which comprises:

[0016] S10: dividing the target handwriting to obtain a plurality of handwriting regions, collecting spectral parameters and images of the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array.

[0017] S20: predicting aging of the spectral parameter array to obtain a spectral aging parameter array, identifying the handwriting image array to obtain a handwriting structure parameter array, and classifying to obtain an image aging parameter array.

[0018] S30: performing relative error analysis and global error analysis according to the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, and performing sample error analysis according to the target handwriting to obtain a sample error coefficient.

[0019] S40: calculating according to the spectral aging parameter array and the image aging parameter array to obtain an ink aging parameter, and performing error compensation according to the relative error coefficient, the global error coefficient and the sample error coefficient to obtain a compensated ink aging parameter interval as an ink aging degree detection result.

[0020] For example, the oxidation kinetics mainly refers to the rate and mechanism of the aging process of the ink material over time under the action of oxidation. In this process, the ink material reacts with oxygen in the surrounding environment, causing changes in its chemical and physical properties, which in turn affect the appearance and performance of the ink. By analyzing the kinetics of this oxidative aging process, the degree of aging of the ink can be more accurately assessed. Thus, the ink aging degree detection method based on oxidation kinetics is proposed.

[0021] Specifically, when processing the target handwriting to assess its aging degree, the entire handwriting of the word is first carefully divided into multiple small handwriting regions. This step is similar to dividing a complex painting into several detailed parts in order to analyze more accurately. Then, spectral parameter collection is performed on these divided handwriting regions, which is done through specific spectral instruments that can capture the absorption characteristics of the handwriting material under different wavelengths of light. When the handwriting ages over time, its spectral absorption characteristics will change subtly, reflecting the degree of reaction between the ink composition and oxygen. At the same time, microscopic image collection is performed on these handwriting regions to obtain a handwriting image array. Microscopic images can reveal the fine structure of the handwriting, including the distribution of ink particles, the diffusion of handwriting edges, etc. Handwriting with different aging degrees will show different degrees of diffusion under the microscope, which is caused by the gradual penetration of ink material into the paper fibers over time. By comparing the spectral parameters and microscopic images of different handwriting regions, a more comprehensive understanding of the aging condition of the handwriting can be obtained, including the uniformity of aging and the difference in aging degree of different regions. This comprehensive analysis method provides a more accurate and detailed assessment of the aging of handwriting.

[0022] Further, advanced aging prediction algorithms are applied to the spectral parameter array. This algorithm can infer the aging trend of the handwriting over time based on the absorption characteristics of the writing material under different spectra, and then generate a spectral aging parameter array. The spectral aging parameter array not only reflects the degree of handwriting aging, but also reveals the change law of spectral characteristics during the aging process. Then, in-depth identification of the handwriting image array is performed. Through high-precision image processing technology, subtle structural features of the handwriting can be extracted, such as ink distribution, line width, and the blurring degree of the handwriting edge, etc. These features are integrated into a handwriting structure parameter array. Subsequently, machine learning algorithms are used to classify and analyze these structural parameters to distinguish handwriting image features of different aging degrees. The key to this step is to identify subtle changes in the handwriting image due to aging, such as ink diffusion and line breakage, etc., thereby generating an image aging parameter array. The combination of spectral aging parameter array and image aging parameter array can provide a more comprehensive understanding of the aging condition of the handwriting. These parameters not only provide time clues for handwriting aging, but also reveal the complex mechanism of the interaction between the writing material and the paper during the aging process. By comparing different parameter arrays, the approximate aging time of the handwriting can be more accurately inferred, providing a strong scientific basis for subsequent handwriting identification, cultural relic protection, etc.

[0023] Optionally, an error analysis method is employed in the process of handwriting aging assessment to ensure the accuracy of the assessment results. First, relative error analysis and global error analysis are conducted based on the spectral aging parameter array and the image aging parameter array. The core of this step lies in the recognition that there may be certain errors in the spectral aging parameters and image aging parameters of different handwriting areas, which may be caused by factors such as uneven distribution of ink materials, slight differences in paper surface, or technical limitations in the collection process. At the same time, even within the same handwriting area, there may be errors between the spectral aging parameters and image aging parameters, which reflect the inherent differences between the two evaluation methods. Through relative error analysis, the proportion of these differences, i.e., the relative error coefficient, is calculated to quantify the error degree between different handwriting areas or different evaluation methods within the same handwriting area. Global error analysis considers the overall error situation of all handwriting areas and evaluation methods, generating a global error coefficient to comprehensively evaluate the accuracy of the entire handwriting aging assessment process. In addition, sample error analysis is conducted for the target handwriting. The starting point of this step is to recognize that when identifying the text within the handwriting, factors such as the writing habits, frequency, and force used when writing different texts may affect the aging degree and pattern of the ink. At the same time, when using machine learning algorithms for model training, the difference in sample data volume may also cause bias in the evaluation results. Therefore, through detailed sample error analysis of the target handwriting, the sample error coefficient is calculated to quantify the influence of these factors on the handwriting aging assessment results. This step not only helps to more accurately understand the complex mechanism of handwriting aging, but also provides more reliable evaluation basis for subsequent handwriting identification, cultural relic protection, etc.

[0024] Finally, in the final stage of processing the handwriting aging assessment, calculations and error compensation are taken to obtain accurate handwriting aging degree detection results. First, comprehensive calculations are performed based on the spectral aging parameter array and the image aging parameter array, and the preliminary handwriting aging parameters are obtained by taking the mean or weighted average of the two parameters. This step integrates the advantages of spectral analysis and image analysis, providing a comprehensive assessment of the degree of handwriting aging. However, due to various factors, these preliminary parameters may have some errors. Therefore, next, error compensation is performed according to the relative error coefficient, global error coefficient and sample error coefficient calculated earlier. These three coefficients reflect the errors between different handwriting regions, the errors between different evaluation methods within the same handwriting region, and the errors caused by sample data differences, respectively. By combining these error coefficients with the preliminary handwriting aging parameters, the corresponding adjustments are made to obtain more accurate handwriting aging parameters. Specifically, these error coefficients can be regarded as an adjustment factor for defining the upper and lower limits of the preliminary parameters. For example, if the preliminary calculation of the handwriting aging parameter is 1 (assuming it is a dimensionless standardized parameter), and the relative error coefficient, global error coefficient and sample error coefficient all indicate that the error range is within ±5%, then the final handwriting aging parameter can be defined as an interval, i.e. 0.95 to 1.05 (i.e. 1±5%). This interval includes the preliminary calculation result and considers the possible error range, thus providing a more reliable and comprehensive handwriting aging degree detection result. Through this series of calculation and error compensation steps, the aging degree of handwriting can be more accurately evaluated, providing strong support for subsequent handwriting identification, cultural relic protection and other work.

[0025] In a preferred embodiment, the target handwriting is divided to obtain a plurality of handwriting regions, spectral parameter acquisition and image acquisition are performed on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array, comprising: dividing the target handwriting according to a preset division window to obtain a plurality of handwriting regions; using a spectrometer to acquire spectral parameters of the plurality of handwriting regions to obtain a spectral parameter array; acquiring images of the plurality of handwriting regions to obtain a handwriting image array.

[0026] Optionally, in the process of processing the target handwriting, the target handwriting is first finely divided according to a preset division window. The purpose of this step is to divide the handwriting into multiple independent handwriting regions, so as to collect spectral parameters and images of each region separately. The size and shape of the preset division window are usually determined according to the complexity of the handwriting, the size of the paper, and the specific needs of the analysis, to ensure the rationality and accuracy of the division, for example, a rectangular window of 2mm*2mm, the target handwriting is divided into multiple rectangular regions, each handwriting region includes a partial region of the handwriting, for example, the region of a certain stroke. Then, a professional spectrometer is used to collect spectral parameters of each handwriting region. The spectrometer can measure the reflection or absorption characteristics of ink at different wavelengths, thereby generating a series of spectral data. By collecting the spectral data of multiple handwriting regions, a spectral parameter array is constructed, which contains the spectral characteristics of each handwriting region, providing key information for subsequent ink aging analysis. At the same time, the image of each handwriting region is also collected to obtain a handwriting image array. This step uses a high-resolution image acquisition device that can capture the fine structure of the handwriting, the thickness of the lines, and the distribution of the ink, etc. The handwriting image array and the spectral parameter array complement each other and together constitute a comprehensive description of the target handwriting. Through this series of steps, the target handwriting is successfully divided into multiple handwriting regions that are convenient for analysis, and the spectral parameters and image information of each region are collected, laying a foundation for subsequent ink aging analysis.

[0027] In a preferred embodiment, the spectral parameter array is subjected to aging prediction to obtain a spectral aging parameter array, including: according to the ink aging detection data in the historical time, collecting a sample spectral parameter set of sample handwriting, and collecting the actual aging time of the sample handwriting as a sample spectral aging parameter set; using the sample spectral parameter set and the sample spectral aging parameter set as training data and test data, training and testing to obtain a qualified spectral aging predictor based on deep learning; inputting multiple spectral parameters in the spectral parameter array into the spectral aging predictor to predict multiple spectral aging parameters, and obtaining a spectral aging parameter array.

[0028] Exemplarily, in the process of processing the spectral parameter array to predict the ink aging degree, the ink aging detection data in the historical time is reviewed, and representative sample handwriting is selected from it. For these sample handwriting, the spectral parameters are collected using a spectrometer to construct a sample spectral parameter set. At the same time, the actual aging time of these sample handwriting is also recorded as a sample spectral aging parameter set. This step ensures that the training data contains both the spectral characteristics of the ink and their aging time, providing a solid foundation for the subsequent prediction model. Next, using these sample spectral parameter sets and sample spectral aging parameter sets as training data and test data, based on deep learning technology, the spectral aging predictor is trained. Deep learning model can learn complex mapping relationship from a large amount of data, so it is very suitable for predicting the aging degree of ink. In the training process, the parameters of the model are constantly adjusted to improve its prediction accuracy. When the performance of the model on the test data reaches the preset qualified standard, it is considered that the spectral aging predictor has been trained. Finally, multiple spectral parameters in the spectral parameter array to be analyzed are input into the trained spectral aging predictor. The predictor will output the corresponding spectral aging parameters according to these spectral parameters. Through this step, a spectral aging parameter array is obtained, and each parameter in the array represents the predicted value of the aging degree of the corresponding handwriting area. In summary, through the collection of historical data, the training of deep learning model and the application of predictor, the aging prediction of spectral parameter array is successfully realized, which provides strong support for the subsequent ink aging analysis.

[0029] In a preferred embodiment, the handwriting image array is identified to obtain a handwriting structure parameter array, and the image aging parameter array is classified, comprising: according to the ink aging detection data in the historical time, collecting a sample handwriting image set of sample handwriting, and collecting the actual diffusion distance of the sample handwriting, and labeling it as a sample handwriting structure parameter set; using the sample handwriting image set and the sample handwriting structure parameter set as training data and test data, based on a convolutional neural network, training and testing to obtain a qualified handwriting structure identifier; inputting multiple handwriting images in the handwriting image array into the handwriting structure identifier to identify and output multiple handwriting structure parameters to obtain a handwriting structure parameter array; and classifying multiple image aging parameters according to the multiple handwriting structure parameters to obtain an image aging parameter array.

[0030] Specifically, based on the historical aging detection data of ink, a series of sample handwriting is selected, and a high-resolution image acquisition device is used to obtain a sample handwriting image set. At the same time, the diffusion distance of the ink of the sample handwriting in the actual aging process is measured, and the data is arranged into a sample handwriting structure parameter set. This step ensures that the training data contains both the image features of the handwriting and the structural changes in the actual aging process, providing strong support for the subsequent recognition model. Next, using the sample handwriting image set and the sample handwriting structure parameter set as training data and test data, based on the powerful image recognition ability of the convolutional neural network (CNN), the handwriting structure recognizer is trained. The CNN model can automatically extract features from image data and learn the complex relationship between these features and handwriting structure parameters. During the training process, the parameters of the model are continuously adjusted to improve its recognition accuracy and generalization ability. When the model's performance on the test data meets the preset qualified standard, the handwriting structure recognizer is considered to have been trained. Then, multiple handwriting images in the handwriting image array to be analyzed are input into the trained handwriting structure recognizer. The recognizer will output corresponding handwriting structure parameters based on these image features. Through this step, a handwriting structure parameter array is obtained, and each parameter in the array represents the structural features of the corresponding handwriting region. Finally, the handwriting structure parameters are further classified to obtain an image aging parameter array. This step utilizes the inherent relationship between handwriting structure parameters and image aging degree, and converts handwriting structure parameters into specific image aging parameters through classification algorithms. These parameters not only reflect the aging degree of the handwriting image, but also provide important reference information for subsequent ink aging analysis. In summary, through the collection of historical data, the training of CNN models, the application of handwriting structure recognizers, and the classification of image aging parameters, the handwriting image array is successfully analyzed and processed, providing strong support for subsequent ink aging research.

[0031] In a preferred embodiment, according to the plurality of handwriting structure parameters, a plurality of image aging parameters are classified to obtain an image aging parameter array, including: collecting a sample handwriting structure parameter set of sample handwriting, and obtaining an average aging time of sample handwriting under different sample handwriting structure parameters to obtain a sample image aging parameter set; constructing a mapping relationship between the sample handwriting structure parameter set and the sample image aging parameter set to obtain an image aging classifier; inputting the plurality of handwriting structure parameters into the image aging classifier respectively, and classifying and outputting a plurality of image aging parameters to obtain an image aging parameter array.

[0032] In detail, sample handwriting structure parameter sets of sample handwriting collected from historical data are recorded in detail, which record the structural characteristics of different sample handwriting. At the same time, for each sample handwriting structure parameter, the average aging time of the corresponding sample handwriting is measured and calculated, so as to obtain a sample image aging parameter set. This step ensures that the data contains not only the structural information of the handwriting, but also the time characteristics of the handwriting in the actual aging process, providing a solid foundation for the subsequent classification model. Next, the mapping relationship between the sample handwriting structure parameter set and the sample image aging parameter set is constructed. This step is the key to constructing the image aging classifier, which enables the classifier to predict and output the corresponding image aging parameters according to the input handwriting structure parameters. In order to achieve this goal, a machine learning algorithm is used to train and optimize the model parameters, and finally a qualified image aging classifier is obtained. Finally, multiple handwriting structure parameters to be analyzed are input into the trained image aging classifier. The classifier will use the mapping relationship learned before to classify and output the corresponding image aging parameters according to the handwriting structure parameters. Through this step, an image aging parameter array is obtained, and each parameter in the array represents the predicted value of the aging degree of the corresponding handwriting image. In summary, through the collection of sample data, the construction of mapping relationship and the application of image aging classifier, the classification processing of multiple handwriting structure parameters is successfully realized, and the image aging parameter array is obtained. This process not only improves the accuracy and efficiency of ink aging analysis, but also provides strong support for subsequent ink research and application.

[0033] In a preferred embodiment, relative error analysis and global error analysis are performed according to the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, including: randomly selecting two spectral aging parameters from the spectral aging parameter array, calculating the error amplitude as the first spectral global error coefficient; continue to randomly select multiple groups of spectral aging parameters, calculate multiple error amplitudes and calculate the mean value to obtain the spectral global error coefficient; according to the image aging parameter array, perform global error analysis calculation to obtain the image global error coefficient, combine the spectral global error coefficient to calculate the global error coefficient; combine the spectral aging parameters and the image aging parameters of the same handwriting region in the spectral aging parameter array and the image aging parameter array to obtain multiple groups of aging parameter combinations; calculate the error amplitudes of the multiple groups of aging parameter combinations respectively, and calculate the mean value of the multiple error amplitudes to obtain the relative error coefficient.

[0034] Preferably, two spectral aging parameters are randomly selected from the spectral aging parameter array, and the error amplitude between them is calculated, which is taken as the preliminary global error reference, i.e. the first spectral global error coefficient. In order to more comprehensively evaluate the error situation of spectral aging parameters, multiple groups of spectral aging parameters are further randomly selected, and multiple error amplitudes are calculated and averaged to obtain a more robust spectral global error coefficient. Next, global error analysis is performed on the image aging parameter array, and the image global error coefficient is obtained through a specific calculation method. This step provides an independent evaluation of the accuracy of image aging parameters. In order to obtain a comprehensive global error coefficient, the spectral global error coefficient and the image global error coefficient are combined, and a global error coefficient considering both spectral and image information is obtained through appropriate weighting or averaging methods. In addition, in order to further explore the relationship between spectral aging parameters and image aging parameters, the aging parameters corresponding to the same handwriting area in both are paired and combined to form multiple groups of aging parameter combinations. For each combination, the error amplitude between the spectral aging parameters and the image aging parameters is calculated, and the average of these error amplitudes is calculated to obtain a relative error coefficient. This coefficient reflects the consistency or difference between the two aging parameters in describing the aging degree of the same handwriting. In summary, through in-depth error analysis of the spectral aging parameter array and the image aging parameter array, not only the global error coefficients of the spectrum and the image are obtained, but also the correlation and consistency between the two aging parameters are revealed through the calculation of the relative error coefficient. These error analysis results provide important accuracy and reliability evaluation basis for subsequent ink aging research.

[0035] In a preferred embodiment, according to the target handwriting, sample error analysis is performed to obtain a sample error coefficient, including: identifying the text content obtained from the target handwriting; retrieving the writing proficiency coefficient of the text content; obtaining the average text occurrence rate of different sample text content in the sample handwriting; obtaining the average accuracy rate of spectral aging parameter prediction and handwriting structure parameter identification, and calculating the error rate; according to the ratio of the average text occurrence rate and the writing proficiency coefficient, the error rate is corrected and calculated to obtain the sample error coefficient.

[0036] For example, the text content of the target handwriting is identified and obtained, which is the basis for analysis. Subsequently, in order to the proficiency of the text content writing, the number of occurrences of the target handwriting text content in the 10,000 files is counted, and the writing proficiency coefficient is obtained, for example, the text content is "I", and the proportion of the appearance of "I" in the 10,000 files is 10%, which is the writing proficiency coefficient. The larger this coefficient is, the more stable the handwriting is, the more sample data is, and the smaller the error rate will be in theory. Next, the average text occurrence rate of different sample text contents in the sample handwriting is calculated. This step is based on the occurrence rate of each character (or each type of text content) in the sample handwriting in the training data, and then the average value is obtained, which is the average text occurrence rate. This value reflects the universality and representativeness of each type of text content in the sample handwriting. At the same time, the average accuracy of spectral aging parameter prediction and handwriting structure parameter identification is obtained, and the error rate is calculated according to this, for example, by subtracting the average accuracy of spectral aging parameter prediction and handwriting structure parameter identification, for example, 90%, then the error rate is 10%. This error rate preliminarily reflects the possible deviation in the prediction and identification process. In order to more accurately evaluate the sample error, a correction calculation method is adopted. Specifically, the ratio of the average text occurrence rate to the writing proficiency coefficient is taken as the correction factor, which is directly multiplied by the preliminary calculated error rate, so as to obtain the sample error coefficient. For example, assuming that the average text occurrence rate is 0.05 (i.e. 5% of the text content frequently appears in the sample), the writing proficiency coefficient is 0.10 (which means that the text content is relatively proficient for most people to write, and the writing frequency is 10%), and the preliminary error rate is 10%, then the corrected sample error coefficient is 0.05 / 0.10x10%=5%. This correction process considers the influence of the universality of the text content and the proficiency of the writer on the error rate, so that the final sample error coefficient is closer to the actual situation. In summary, through the identification of the text content of the target handwriting, the retrieval of the writing proficiency coefficient, the calculation of the average text occurrence rate, the acquisition of the error rate and the correction calculation, a comprehensive and accurate sample error coefficient is obtained, which provides an important reference for subsequent handwriting analysis and research.

[0037] In a preferred embodiment, the ink aging parameter is obtained by calculating the mean value of the spectral aging parameter array and the image aging parameter array, and the compensated ink aging parameter interval is obtained by error compensation according to the relative error coefficient, the global error coefficient and the sample error coefficient, and is taken as the ink aging degree detection result.

[0038] Specifically, in the process of detecting the ink aging degree, the information of the spectral aging parameter array and the image aging parameter array is first combined, and the ink aging parameter is obtained by calculating the mean value of the parameters in the two arrays. This step fully utilizes the advantages of spectral analysis and image processing, and provides a preliminary evaluation of the ink aging degree. In order to further improve the accuracy of the detection result, various error sources are considered, and the relative error coefficient, the global error coefficient and the sample error coefficient are calculated. These coefficients respectively reflect the relative difference between the spectral and image aging parameters, the error level of the overall data set and the error characteristics of the specific sample. Subsequently, using these error coefficients, an error compensation coefficient is calculated by a specific weighting or combination method. This coefficient aims to comprehensively consider various error factors in order to modify the preliminary ink aging parameter. Finally, the error compensation coefficient is used to perform error compensation calculation on the ink aging parameter. This step adjusts the value of the preliminary parameter to obtain a more accurate and reliable compensated ink aging parameter interval. This interval not only contains the estimated value of the ink aging degree, but also reflects the fluctuation range that may be caused by the existence of errors, and therefore can be taken as the final detection result of the ink aging degree. In summary, by combining spectral and image aging parameters, calculating error coefficients and performing error compensation, an accurate and reliable ink aging degree detection result is obtained, which provides strong support for subsequent ink research and application.

[0039] The ink aging degree detection method based on oxidation kinetics provided by the embodiment of the present application has at least the following technical effects:

[0040] 1. By combining the spectral parameter array and the handwriting image array, multi-dimensional detection of the ink aging degree is realized. Spectral parameters provide direct evidence of changes in the chemical composition of ink, while handwriting images reflect changes in the physical form of ink on paper. By fusing these two different dimensional data and using deep learning and other technologies for comprehensive analysis, the aging degree of ink can be more accurately evaluated, avoiding the one-sidedness that may be caused by single-dimensional data.

[0041] 2. The relative error coefficient, global error coefficient, and sample error coefficient are introduced innovatively to consider various error sources that may exist in the actual detection process, and an error compensation coefficient is calculated based on these indicators. By applying the error compensation coefficient to the preliminary calculated ink aging parameters, accurate compensation of errors can be achieved, thereby greatly improving the accuracy and reliability of the detection results.

[0042] 3. Intelligent technical means such as deep learning and convolutional neural networks are used to realize automatic collection, identification, and analysis of spectral parameters and handwriting images, which not only greatly improves the detection efficiency but also reduces errors caused by human intervention. In addition, this method also has the functions of automatically calculating the error compensation coefficient and generating the compensated ink aging parameter interval, making the entire detection process more convenient and efficient.

[0043] Embodiment Two:

[0044] As shown in Figure 2 Based on the same inventive concept as the ink aging degree detection method based on oxidation kinetics provided in Embodiment One, the present embodiment also provides an ink aging degree detection system based on oxidation kinetics. The system includes:

[0045] A handwriting division module 11 is used to divide the target handwriting to obtain a plurality of handwriting regions, and to perform spectral parameter collection and image collection on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array.

[0046] A parameter array acquisition module 12 is used to perform aging prediction on the spectral parameter array to obtain a spectral aging parameter array, and to perform identification on the handwriting image array to obtain a handwriting structure parameter array, and to classify to obtain an image aging parameter array.

[0047] A sample error analysis module 13 is used to perform relative error analysis and global error analysis based on the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, and to perform sample error analysis based on the target handwriting to obtain a sample error coefficient.

[0048] A detection result generation module 14 is used to calculate based on the spectral aging parameter array and the image aging parameter array to obtain ink aging parameters, and to perform error compensation based on the relative error coefficient, the global error coefficient, and the sample error coefficient to obtain a compensated ink aging parameter interval as the ink aging degree detection result.

[0049] Further, the handwriting division module 11 is also used to perform the following steps:

[0050] According to a preset division window, the target handwriting is divided to obtain a plurality of handwriting regions; a spectrometer is used to collect spectrum parameters of the plurality of handwriting regions to obtain a spectrum parameter array; and an image of the plurality of handwriting regions is collected to obtain a handwriting image array.

[0051] Further, the parameter array acquisition module 12 is further configured to perform the following steps:

[0052] According to the ink aging detection data in the historical time, a sample spectrum parameter set of sample handwriting is collected, and an actual aging time of the sample handwriting is collected as a sample spectrum aging parameter set; the sample spectrum parameter set and the sample spectrum aging parameter set are used as training data and test data, a qualified spectrum aging predictor is trained and tested based on deep learning; a plurality of spectrum parameters in the spectrum parameter array are input into the spectrum aging predictor, a plurality of spectrum aging parameters are output by prediction, and a spectrum aging parameter array is obtained.

[0053] Further, the parameter array acquisition module 12 is further configured to perform the following steps:

[0054] According to the ink aging detection data in the historical time, a sample handwriting image set of sample handwriting is collected, and an actual diffusion distance of the sample handwriting is collected and labeled as a sample handwriting structure parameter set; the sample handwriting image set and the sample handwriting structure parameter set are used as training data and test data, a qualified handwriting structure identifier is trained and tested based on a convolutional neural network; a plurality of handwriting images in the handwriting image array are input into the handwriting structure identifier, a plurality of handwriting structure parameters are output by identification, and a handwriting structure parameter array is obtained; and according to the plurality of handwriting structure parameters, a plurality of image aging parameters are classified to obtain an image aging parameter array.

[0055] Further, the parameter array acquisition module 12 is further configured to perform the following steps:

[0056] A sample handwriting structure parameter set of sample handwriting is collected, and an average aging time of sample handwriting under different sample handwriting structure parameters is obtained to obtain a sample image aging parameter set; a mapping relationship between the sample handwriting structure parameter set and the sample image aging parameter set is constructed to obtain an image aging classifier; the plurality of handwriting structure parameters are respectively input into the image aging classifier, a plurality of image aging parameters are output by classification, and an image aging parameter array is obtained.

[0057] Further, the sample error analysis module 13 is further configured to perform the following steps:

[0058] Two spectral aging parameters are randomly extracted from the spectral aging parameter array, and an error amplitude is calculated as a first spectral global error coefficient; a plurality of groups of spectral aging parameters are continuously randomly extracted, a plurality of error amplitudes are calculated, and a mean value is calculated to obtain a spectral global error coefficient; global error analysis and calculation are performed according to the image aging parameter array to obtain an image global error coefficient, the spectral global error coefficient is combined, and a global error coefficient is calculated and obtained; the spectral aging parameters and the image aging parameters in the same handwriting region in the spectral aging parameter array and the image aging parameter array are combined to obtain a plurality of groups of aging parameter combinations; error amplitudes of the plurality of groups of aging parameter combinations are respectively calculated, and a mean value of the plurality of error amplitudes is calculated to obtain a relative error coefficient.

[0059] Further, the sample error analysis module 13 is further used to perform the following steps:

[0060] The text content of the target handwriting is identified and acquired; a writing proficiency coefficient of the text content is retrieved and acquired; an average text occurrence rate of different sample text contents in the sample handwriting is acquired; an average accuracy rate of spectral aging parameter prediction and handwriting structure parameter identification is acquired, and an error rate is calculated and obtained; the error rate is corrected and calculated according to a ratio of the average text occurrence rate and the writing proficiency coefficient to obtain a sample error coefficient.

[0061] Further, the detection result generation module 14 is further used to perform the following steps:

[0062] A mean value is calculated according to the spectral aging parameter array and the image aging parameter array to obtain an ink aging parameter; an error compensation coefficient is calculated and obtained according to the relative error coefficient, the global error coefficient, and the sample error coefficient; the error compensation coefficient is used to perform error compensation calculation on the ink aging parameter to obtain a compensated ink aging parameter interval as an ink aging degree detection result.

[0063] Through the foregoing detailed description of the method for detecting the ink aging degree based on oxidation kinetics, those skilled in the art can clearly understand the ink aging degree detection system based on oxidation kinetics in the embodiment. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part is described in the method part.

[0064] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and that the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Therefore, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting the degree of aging of ink based on oxidation kinetics, characterized by, The method comprises: The target handwriting is divided to obtain a plurality of handwriting regions, spectral parameter acquisition and image acquisition are performed on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array; The spectral parameter array is subjected to aging prediction to obtain a spectral aging parameter array, the handwriting image array is subjected to recognition to obtain a handwriting structure parameter array, and an image aging parameter array is obtained through classification, comprising: According to the ink aging detection data in the historical time, a sample spectral parameter set of sample handwriting is collected, and the actual aging time of the sample handwriting is collected as a sample spectral aging parameter set; The sample spectral parameter set and the sample spectral aging parameter set are used as training data and test data, a qualified spectral aging predictor is obtained through training and testing based on deep learning; A plurality of spectral parameters in the spectral parameter array are input into the spectral aging predictor to predict and output a plurality of spectral aging parameters, thereby obtaining a spectral aging parameter array; According to the ink aging detection data in the historical time, a sample handwriting image set of sample handwriting is collected, and the actual diffusion distance of the sample handwriting is collected and labeled as a sample handwriting structure parameter set; The sample handwriting image set and the sample handwriting structure parameter set are used as training data and test data, a qualified handwriting structure recognizer is obtained through training and testing based on a convolutional neural network; A plurality of handwriting images in the handwriting image array are input into the handwriting structure recognizer to recognize and output a plurality of handwriting structure parameters, thereby obtaining a handwriting structure parameter array; According to the plurality of handwriting structure parameters, a plurality of image aging parameters are classified to obtain an image aging parameter array; According to the spectral aging parameter array and the image aging parameter array, relative error analysis and global error analysis are performed to obtain a relative error coefficient and a global error coefficient, sample error analysis is performed according to the target handwriting to obtain a sample error coefficient; According to the spectral aging parameter array and the image aging parameter array, ink aging parameters are obtained through calculation, error compensation is performed according to the relative error coefficient, the global error coefficient and the sample error coefficient, a compensated ink aging parameter interval is obtained as an ink aging degree detection result.

2. The method of claim 1, wherein the method is characterized by: The target handwriting is divided to obtain a plurality of handwriting regions, spectral parameter acquisition and image acquisition are performed on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array, comprising: The target handwriting is divided according to a preset division window to obtain a plurality of handwriting regions; A spectrometer is used to collect spectral parameters of the plurality of handwriting regions to obtain a spectral parameter array; Images of the plurality of handwriting regions are collected to obtain a handwriting image array.

3. The method of claim 1, wherein the method is characterized by: According to the plurality of handwriting structure parameters, a plurality of image aging parameters are classified to obtain an image aging parameter array, comprising: A sample handwriting structure parameter set of sample handwriting is collected, and an average aging time of sample handwriting under different sample handwriting structure parameters is obtained to obtain a sample image aging parameter set; A mapping relationship between the sample handwriting structure parameter set and the sample image aging parameter set is constructed to obtain an image aging classifier; Inputting the plurality of handwriting structure parameters into the image aging classifier respectively, and classifying to output a plurality of image aging parameters to obtain an image aging parameter array.

4. The method of claim 1, wherein the method is characterized by: Performing relative error analysis and global error analysis according to the spectral aging parameter array and the image aging parameter array to obtain a relative error coefficient and a global error coefficient, including: Randomly extracting two spectral aging parameters from the spectral aging parameter array, and calculating an error amplitude as a first spectral global error coefficient; Continuing to randomly extract a plurality of groups of spectral aging parameters, calculating a plurality of error amplitudes and calculating a mean value to obtain a spectral global error coefficient; Performing global error analysis calculation according to the image aging parameter array to obtain an image global error coefficient, and combining the spectral global error coefficient to calculate a global error coefficient; Combining the spectral aging parameters and the image aging parameters of the same handwriting region in the spectral aging parameter array and the image aging parameter array to obtain a plurality of groups of aging parameter combinations; Respectively calculating error amplitudes of the plurality of groups of aging parameter combinations, and calculating a mean value of the plurality of error amplitudes to obtain a relative error coefficient.

5. The method of claim 1, wherein the method is characterized by: According to the target handwriting, performing sample error analysis to obtain a sample error coefficient, including: Identifying the text content of the target handwriting; Retrieving the writing proficiency coefficient of the text content; Obtaining the average text occurrence rate of different sample text contents in the sample handwriting; Obtaining the average accuracy rate of spectral aging parameter prediction and handwriting structure parameter identification, and calculating an error rate; According to the ratio of the average text occurrence rate to the writing proficiency coefficient, correcting and calculating the error rate to obtain a sample error coefficient.

6. The method of claim 1, wherein the method is characterized by: According to the spectral aging parameter array and the image aging parameter array, calculating a writing ink aging parameter, and performing error compensation according to the relative error coefficient, the global error coefficient and the sample error coefficient to obtain a compensated writing ink aging parameter interval as a writing ink aging degree detection result, including: According to the spectral aging parameter array and the image aging parameter array, calculating a mean value to obtain a writing ink aging parameter; According to the relative error coefficient, the global error coefficient and the sample error coefficient, calculating an error compensation coefficient; Using the error compensation coefficient, performing error compensation calculation on the writing ink aging parameter to obtain a compensated writing ink aging parameter interval as a writing ink aging degree detection result.

7. An ink aging degree detection system based on oxidation kinetics, characterized by, The system for implementing the oxidation kinetics-based writing ink aging degree detection method of any one of claims 1-6, comprising: A handwriting division module for dividing target handwriting to obtain a plurality of handwriting regions, performing spectral parameter acquisition and image acquisition on the plurality of handwriting regions to obtain a spectral parameter array and a handwriting image array; A parameter array acquisition module for performing aging prediction on the spectral parameter array to obtain a spectral aging parameter array, and performing identification on the handwriting image array to obtain a handwriting structure parameter array, and classifying to obtain an image aging parameter array; a sample error analysis module, configured to perform relative error analysis and global error analysis according to the spectral aging parameter array and the image aging parameter array, to obtain a relative error coefficient and a global error coefficient, and to perform sample error analysis according to the target handwriting to obtain a sample error coefficient; a detection result generation module, configured to perform calculation according to the spectral aging parameter array and the image aging parameter array to obtain an ink aging parameter, and to perform error compensation according to the relative error coefficient, the global error coefficient and the sample error coefficient to obtain a compensated ink aging parameter interval as an ink aging degree detection result.

Images