An image machine learning-based traditional Chinese medicine specimen collection and informatization processing method

Abstract

The application belongs to the technical field of image recognition, and discloses a traditional Chinese medicine specimen collection and information processing method based on image machine learning, which comprises the following steps: collecting traditional Chinese medicine specimens through a special collection platform in multiple spectrums, using a black and white base plate and a scale ruler to ensure data accuracy; using image processing technology to perform geometric correction, color correction and enhancement processing on the collected images; extracting multi-dimensional features based on an improved SIFT algorithm and a deep convolutional neural network, and constructing a specimen feature map; identifying specimen properties in combination with a traditional Chinese medicine knowledge base, and establishing a traditional Chinese medicine specimen information model; realizing specimen identification through multi-dimensional feature matching; evaluating the growth state of the specimen in the collection environment based on an ecological adaptability model, and verifying the accuracy of the identification result; and finally generating a traditional Chinese medicine specimen digital archive containing comprehensive information; the application realizes the digitization, intelligentization and standardization of traditional Chinese medicine specimen collection, identification and management, and improves the accuracy and efficiency of traditional Chinese medicine specimen identification.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, more particularly, to a traditional Chinese medicine specimen collection and information processing method based on image machine learning. BACKGROUND

[0002] As an important part of traditional Chinese medicine system, traditional Chinese medicine has a history of thousands of years of application and a rich theoretical system; accurate identification and standardized management of traditional Chinese medicine specimens are the basis for traditional Chinese medicine research, development and clinical application. However, the traditional collection and identification of traditional Chinese medicine specimens mainly rely on expert experience, which has strong subjectivity, non-uniform standards and simple recording methods, and cannot meet the needs of modern traditional Chinese medicine research and industrial development.

[0003] The traditional collection and information processing of traditional Chinese medicine specimens mainly have the following problems:

[0004] Firstly, the traditional collection method of traditional Chinese medicine specimens lacks standardized collection process and equipment; the environmental parameter record is not comprehensive, and the light condition is unstable during the collection process, which leads to uneven quality of the obtained specimen images, and it is difficult to use for subsequent accurate analysis.

[0005] Secondly, the existing digital technology of traditional Chinese medicine specimens mainly adopts simple photographing and archiving method, which lacks multi-dimensional and multi-spectral image acquisition capability, which leads to that the obtained images cannot comprehensively reflect the morphological characteristics, texture characteristics and color characteristics of traditional Chinese medicine specimens and the feature differences under different spectral conditions;

[0006] Thirdly, the traditional identification of traditional Chinese medicine specimens mainly relies on expert experience, and lacks objective quantitative feature extraction and pattern recognition method; with the increase of traditional Chinese medicine species and the lack of professional talents, pure manual identification cannot meet the large-scale and high-precision classification requirements of traditional Chinese medicine specimens;

[0007] Fourthly, the traditional management method of traditional Chinese medicine specimens lacks objective verification mechanism for the authenticity and reliability of the specimens; the specimen identification result lacks credibility evaluation, which is easy to misjudge and confuse, and affects the accuracy of subsequent research and application.

[0008] In view of the above problems, the present application provides a traditional Chinese medicine specimen collection and information processing method based on image machine learning. SUMMARY

[0009] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0010] A traditional Chinese medicine specimen collection and information processing method based on image machine learning, comprising:

[0011] Step one: based on the pre-constructed specimen collection platform, multi-dimensional data of traditional Chinese medicine specimens are collected, corresponding specimen sample images are obtained, and specimen metadata are recorded; and the specimen metadata are stored in the pre-constructed specimen database;

[0012] Step two: feature extraction is performed on the collected specimen sample images, and a corresponding specimen feature map is constructed in combination with the constructed specimen database;

[0013] Step three: specimen attribute recognition is performed on the obtained specimen feature map, and a corresponding traditional Chinese medicine specimen informationization model is constructed based on the specimen attribute recognition result;

[0014] Step four: multi-dimensional feature matching processing is performed on the constructed traditional Chinese medicine specimen informationization model, and a corresponding specimen recognition result is obtained;

[0015] Step five: based on the specimen recognition result and the collected environmental information, the growth environment of the corresponding traditional Chinese medicine specimen is simulated, and the accuracy of the specimen recognition result is evaluated based on the growth environment simulation result, and a corresponding credibility index is obtained;

[0016] Step six: the credibility index of the specimen recognition result is multi-dimensionally evaluated, and the evaluation result is integrated with the traditional Chinese medicine specimen information to obtain a corresponding informationization traditional Chinese medicine specimen data file.

[0017] Further, the process of multi-dimensional data collection of traditional Chinese medicine specimens includes:

[0018] The specimen collection platform is constructed; the image collection terminal collects images of the traditional Chinese medicine specimens to be collected under different light supplement modes based on pre-set collection requirements, and obtains corresponding specimen sample images and specimen sample videos;

[0019] Geographic information is collected based on the positioning terminal at the same time; metadata information corresponding to each specimen sample image is recorded, and the metadata information is associated with the collected specimen image set to generate specimen metadata; the collected specimen metadata are uploaded and stored in the pre-constructed specimen database.

[0020] Further, the light supplement mode refers to a natural light mode and a fluorescent mode provided by a light supplement terminal, the natural light mode refers to guiding an environmental light source to irradiate a specimen through a natural light guiding structure built in the light supplement terminal; and the fluorescent mode refers to making a corresponding image collection terminal collect images of the traditional Chinese medicine specimens to be collected based on ultraviolet light and infrared light respectively through a wavelength-adjustable ultraviolet light LED array and a multi-waveband infrared light source.

[0021] Further, the process of feature extraction on the obtained specimen sample images includes:

[0022] An acquired specimen sample image is obtained, a corresponding specimen sample image is geometrically corrected based on a scale ruler on a corresponding black-and-white color base plate during an image acquisition process, and the geometrically corrected specimen sample image is subjected to polynomial color correction and white balance adjustment; an initial specimen image is obtained;

[0023] The obtained initial specimen image is subjected to local mean filtering processing to remove Gaussian noise in the image while preserving edge and texture details, wherein a formula for the non-local mean filtering processing is: In the formula, BF[I] p represents a pixel value at a pixel point p after local mean filtering; S represents a spatial domain, i.e., a neighborhood range of the corresponding pixel point p; q represents a pixel point index in the neighborhood range; and respectively represent a spatial domain kernel function and a value domain kernel function; I p and I q respectively represent pixel values at the corresponding pixel point p and the pixel point q, and p≠q; W p represents a normalization coefficient;

[0024] After the local mean filtering is completed, a specimen type corresponding to the corresponding traditional Chinese medicine specimen is obtained, and the corresponding traditional Chinese medicine specimen is subjected to adaptive enhancement based thereon to obtain an enhanced specimen image, and a standard specimen image is obtained based thereon;

[0025] Feature points in the corresponding standard specimen image are extracted based on the improved SIFT algorithm to obtain image feature points;

[0026] The feature points in the initial specimen image after the adaptive enhancement are labeled based on the obtained image feature points, and a corresponding feature distribution map is generated;

[0027] Morphological features, texture features, and color features of the corresponding standard specimen image are extracted based on the obtained feature distribution map to obtain morphological feature vectors, texture feature vectors, and color feature vectors; at the same time, a 3D convolutional neural network is used to process the acquired specimen sample video to capture change features of the traditional Chinese medicine specimen in the time dimension to obtain video dynamic features; for example, a fluorescence decay process or appearance changes of the specimen at different angles;

[0028] The morphological feature vectors, the texture feature vectors, the color feature vectors, the video dynamic features, and the acquired geographical location information are fused to obtain a corresponding comprehensive feature vector.

[0029] Further, the process of extracting feature points in the corresponding standard specimen image based on the improved SIFT algorithm to obtain corresponding image feature points includes:

[0030] respectively based on Gaussian convolution kernels under different scales to obtain Gaussian convolution images under different scales; performing image difference calculation on Gaussian convolution images under adjacent scales to obtain corresponding difference images; obtaining pixel values corresponding to pixel points in the corresponding difference images, and comparing the pixel values with pixel values corresponding to all pixel points in a neighborhood pixel set, determining whether the pixel values corresponding to the corresponding pixel points are local extreme points based on a comparison result, if not, not performing any other operation, if yes, marking the corresponding pixel points as candidate key points;

[0031] Further, curvature coefficients corresponding to each candidate key point are obtained, and the curvature coefficients are compared with a pre-set curvature threshold, if the curvature coefficients are not greater than the curvature threshold, the corresponding candidate key points are marked as image feature points, if the curvature coefficients are greater than the curvature threshold, the corresponding candidate key points are discarded.

[0032] Further, the process of obtaining the specimen feature atlas and identifying the specimen attribute, and constructing a corresponding traditional Chinese medicine specimen informationization model based on the specimen attribute identification result includes:

[0033] Based on the existing traditional Chinese medicine knowledge base, a traditional Chinese medicine attribute data set related to traditional Chinese medicine is obtained; based on the obtained specimen feature atlas, attribute features of the corresponding traditional Chinese medicine specimen are extracted to obtain corresponding specimen attribute features

[0034] The obtained specimen attribute features and the traditional Chinese medicine attribute data set are associated in attribute, and a corresponding classification framework is constructed based thereon; the construction process of the classification framework is: based on the association relationship between the key attribute information in the traditional Chinese medicine attribute data set and the specimen attribute features, the association relationship is visualized as an attribute association network, and the association strength and direction between each key attribute information and the specimen attribute features are marked; based on the attribute association network, a corresponding classification framework is built;

[0035] The classification framework and the obtained specimen feature atlas are integrated to obtain a corresponding traditional Chinese medicine specimen informationization model, the traditional Chinese medicine specimen informationization model includes feature data, attribute association and classification structure of traditional Chinese medicine specimens, and related knowledge of traditional Chinese medicine.

[0036] Further, the process of obtaining the specimen identification result includes:

[0037] The traditional Chinese medicine specimen informationization database corresponding to the corresponding traditional Chinese medicine specimen is matched with the known traditional Chinese medicine specimen in feature to generate a corresponding initial matching result, the initial matching result includes a plurality of sample matching items;

[0038] sequencing the known Chinese medicinal plant specimens based on the comprehensive similarity coefficient, obtaining a corresponding similarity sequencing sequence; pre-screening the known Chinese medicinal plant specimens in the similarity sequencing sequence based on a pre-set similarity threshold, taking the known Chinese medicinal plant specimens with a comprehensive similarity coefficient higher than the similarity threshold as candidate specimens; aggregating all the obtained candidate specimens, obtaining a corresponding specimen identification result, wherein the specimen identification result comprises specimen ID, name, and similarity data of the known Chinese medicinal plant specimens, and the like; and the similarity data comprises similarity calculation results in multiple dimensions and the comprehensive similarity coefficient.

[0039] Further, the process of obtaining the credibility index comprises:

[0040] extracting parameters from the informationized database of Chinese medicinal plant specimens to obtain ecological environment data related to the corresponding Chinese medicinal plant specimens; and based on the same, extracting environmental parameter ranges suitable for growth of the Chinese medicinal plant specimens, obtaining a corresponding specimen growth environment demand parameter set;

[0041] obtaining the adaptability of the corresponding Chinese medicinal plant specimens to different environmental factors based on a pre-constructed ecological adaptability model, obtaining a corresponding ecological adaptability index;

[0042] the mathematical formula of the ecological adaptability model is defined as:

[0043] wherein, μ x (i) represents the ecological adaptability index corresponding to the i-th environmental factor; x i represents the specific value of the i-th environmental factor; T i,min and T i,max respectively represent the left boundary and the right boundary of the parameter corresponding to the i-th environmental factor; obtained from the specimen growth environment demand parameter set; and δ represents the width of the transition interval;

[0044] Further, based on the ecological adaptability index and in combination with the specimen growth environment demand parameter set, the growth state and performance of the corresponding Chinese medicinal plant specimens under different environmental factors are simulated, and key data points in the simulation process are recorded, wherein the key data points comprise growth rate, morphological change, physiological index, and the like; the simulation results are integrated to generate a complete specimen ecological adaptability data set;

[0045] obtaining the collected specimen metadata, obtaining environmental data recorded in the collection process of the corresponding Chinese medicinal plant specimens based on the same, wherein the environmental data comprises environmental parameters such as temperature, humidity, illumination, and altitude, and the like, and the same is standardized to obtain a corresponding environmental condition parameter set;

[0046] Based on the ecological adaptability model, ecological adaptability indexes corresponding to each environmental parameter in the corresponding environmental data are obtained, and the growth state and performance of the traditional Chinese medicine specimen under the current environmental data are simulated based on the ecological adaptability indexes, to obtain a corresponding environmental specimen ecological data set;

[0047] The obtained environmental specimen ecological data set and specimen ecological adaptability data set are acquired, and the matching degree between the two is evaluated to obtain a corresponding environmental matching degree index; the process of obtaining the corresponding environmental matching degree index is as follows: the environmental factors and environmental parameters in the corresponding environmental specimen ecological data set and specimen ecological adaptability data set are one-to-one corresponding, the parameter difference value and the difference value of the ecological adaptability index between the two are obtained, and the adaptability score is calculated by combining a pre-set matching degree function, to obtain the corresponding environmental matching degree index;

[0048] The obtained environmental matching degree index is integrated to obtain a corresponding specimen ecological environment evaluation data, which includes ecological adaptability indexes, environmental matching degree indexes, and an optimal growth interval, etc.

[0049] Further, based on the sample identification result, the ecological growth demand corresponding to each candidate specimen is obtained, and a consistency comparison is made between the ecological growth demand and the obtained specimen ecological environment evaluation data;

[0050] If the comparison is inconsistent, the corresponding candidate specimen is discarded;

[0051] If the comparison is consistent, the matching degree between the environmental parameters and the ecological growth demand corresponding to each candidate specimen obtained based on the sample identification result is obtained, to obtain a corresponding environmental verification result;

[0052] Further, based on the environmental verification result, the sample identification result corresponding to the corresponding candidate specimen is corrected to adjust the error in the sample identification result and improve the accuracy;

[0053] Further, based on the traditional Chinese medicine expert knowledge base, classification rules and identification points of each type of traditional Chinese medicine are obtained, and based on the same, evaluation factors of the corresponding specimen identification result are obtained, which include morphological feature coincidence degree, ecological environment consistency, and distribution area overlap rate, etc.

[0054] The obtained evaluation factors are weighted and summed to obtain a corresponding credibility index.

[0055] Further, the construction process of the information-based traditional Chinese medicine specimen data file includes:

[0056] The identification criteria of the credibility index are defined, and the credibility level corresponding to the identification result of the corresponding sample is obtained based on the same; the informationization model of the traditional Chinese medicine sample corresponding to the candidate sample in the corresponding sample identification result is obtained, and is compared with the informationization model of the traditional Chinese medicine sample corresponding to the traditional Chinese medicine sample to be collected, to obtain the corresponding data coverage and accuracy, and generate the corresponding data integrity evaluation value based on the same;

[0057] The obtained credibility level and data integrity evaluation value are integrated, and the related information involved in the corresponding traditional Chinese medicine sample is integrated synchronously, to generate the digital file of the informationization traditional Chinese medicine sample.

[0058] The technical effects and advantages of the image machine learning-based traditional Chinese medicine sample collection and informationization processing method of the present application are as follows:

[0059] Through the standardized sample collection platform and process in the collection link, the drawbacks of the lack of standardization in the traditional collection are overcome, the telescopic box body is suitable for different samples, the stable light supplement mode and the comprehensive environmental parameter record ensure that the image quality of the collected sample is high and the data is comprehensive and accurate, which provides a reliable basis for the subsequent accurate analysis; the multi-dimensional and multi-spectrum image collection capability makes up for the deficiency of the traditional simple photographing and archiving, and can comprehensively reflect the key characteristics such as the morphology, texture and color of the sample and the differences under different spectra; in the processing link, the feature extraction and pattern recognition method based on image machine learning changes the subjective identification mode relying on expert experience, and realizes high-precision sample classification through objective quantitative means, to meet the needs of large-scale traditional Chinese medicine sample identification. At the same time, the sample feature atlas, informationization model and credibility evaluation mechanism constructed not only help to deeply study the characteristics of the traditional Chinese medicine sample, but also objectively verify the sample identification result, avoid misjudgment and confusion, and improve the authenticity and reliability of the sample management; in addition, the finally generated informationization traditional Chinese medicine sample data file is convenient for storage, retrieval and sharing, and effectively promotes the development of modern traditional Chinese medicine research and industry, so that the traditional Chinese medicine can better serve the fields of clinical application and innovation in the inheritance and innovation. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 It is a schematic diagram of the image machine learning-based traditional Chinese medicine sample collection and informationization processing method of the present application;

[0061] Figure 2 It is a schematic diagram of the image machine learning-based traditional Chinese medicine sample collection and informationization processing system of the present application. DETAILED DESCRIPTION

[0062] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below, obviously, the described embodiments are only a 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 work belong to the scope of protection of the present application.

[0063] Embodiment 1

[0064] Please refer to Figure 1 The present embodiment is an image machine learning-based traditional Chinese medicine specimen collection and information processing method, which comprises:

[0065] Step one: based on the pre-constructed specimen collection platform, multi-dimensional data of traditional Chinese medicine specimens are collected, corresponding specimen sample images are obtained, and specimen metadata are recorded; and they are stored in the pre-constructed specimen database;

[0066] It needs to be further explained that, in the specific implementation process, the process of obtaining the corresponding specimen sample images and recording the specimen metadata comprises:

[0067] The specimen collection platform is constructed, and the specimen collection platform is composed of a scalable box body, an image collection terminal, a positioning terminal, a light supplement terminal and a replaceable black and white bottom plate; wherein the black and white bottom plate is provided with a high-precision scale ruler for scale reference;

[0068] The size of the traditional Chinese medicine specimen to be collected is estimated, and the size parameter of the scalable box body in the corresponding specimen collection platform is adjusted based on it, so that it is adapted to different specifications of traditional Chinese medicine specimens;

[0069] After the size adjustment is completed, the collection parameters in the image collection terminal are initialized, and after the initialization is completed, the image collection terminal uses the black and white bottom plate under different light supplement modes based on the pre-set collection requirements to collect images of the traditional Chinese medicine specimen to be collected, and obtains corresponding specimen sample images and specimen sample videos;

[0070] Among them, the light supplement mode refers to the natural light mode and the fluorescent mode provided by the light supplement terminal, the natural light mode guides the environmental light source to irradiate the specimen through the natural light guide structure built-in the light supplement terminal; the natural light guide structure passes through the window with adjustable angle, is equipped with light intensity sensor and automatic adjusting light shield, and realizes accurate control of natural light; the fluorescent mode means that the corresponding image collection terminal respectively collects images of the traditional Chinese medicine specimen to be collected based on ultraviolet light and infrared light through the built-in wavelength-adjustable ultraviolet light LED array and multi-waveband infrared light source;

[0071] Synchronously collect geographic information based on the positioning terminal, the geographic information including accurate coordinates and elevation of the specimen collection position; record metadata information such as spectral parameters, specimen collection angle, scale value and geographic information of each image, and associate the metadata information with the collected specimen image set to generate specimen metadata; the specimen metadata including complete information such as collection time, place, light condition and camera parameter, ensuring data traceability;

[0072] After the collection is completed, the collected specimen image, specimen video and associated specimen metadata are encrypted and uploaded, and during the uploading process, the optimal transmission channel is adaptively selected according to the current network environment, and the relay nodes used during the transmission process are enhanced or reduced, so as to ensure the data uploading efficiency and the integrity of the data; after the uploading is completed, the encrypted and packaged specimen image, specimen video and associated specimen metadata are decrypted and verified for completion, and after the verification is passed, they are stored in the pre-constructed specimen database.

[0073] Step two: extracting features from the collected specimen image and constructing a corresponding specimen feature map in combination with the constructed specimen database;

[0074] It needs to be further explained that, in the specific implementation process, the process of extracting features from the obtained specimen image includes:

[0075] The collected specimen image is obtained, and the corresponding specimen image is geometrically corrected based on the scale ruler on the corresponding black and white base plate during the image collection process, wherein the process of geometric correction includes: identifying the scale ruler in the specimen image and establishing a mapping relationship between the image scale size and the actual scale size based thereon;

[0076] Further, geometric correction is performed based on the established mapping relationship, and a perspective transformation algorithm is used to correct the perspective distortion in the image, so as to ensure the geometric shape accuracy of the specimen in the image;

[0077] After the geometric correction is completed, the black and white base plate in the corresponding specimen image is used as a color reference to perform polynomial color correction and white balance adjustment on the geometrically corrected specimen image; the corresponding initial specimen image is obtained; polynomial color correction is a kind of image color consistency processing method for mapping pixel values in different color channels under different light conditions to the same reference color through constructing a non-linear mapping function, and the color channels include red, green and blue three channel colors;

[0078] The obtained initial specimen image is subjected to local mean filtering processing to remove Gaussian noise in the image while retaining edge and texture details, wherein the formula for performing non-local mean filtering processing is: In the formula, BF[I]p represents the pixel value at pixel point p after local mean filtering; S represents the spatial domain, i.e. the neighborhood range of the corresponding pixel point p; q represents the pixel point index in the neighborhood range; and respectively represent the spatial domain kernel function and the value domain kernel function; I p and I q respectively represent the pixel values at the corresponding pixel point p and the pixel point q, and p≠q; W p represents the normalization coefficient;

[0079] After the local mean filtering is completed, the specimen type corresponding to the corresponding traditional Chinese medicine specimen is obtained, and the corresponding traditional Chinese medicine specimen is adaptively enhanced based on the specimen type, to obtain a corresponding enhanced specimen image; for example, for a flower and fruit specimen with bright colors, a nonlinear contrast enhancement algorithm is used; for a root and stem specimen with complex morphological structure, a CLAHE algorithm is used to enhance the texture details; for a leaf specimen with low contrast, a multi-scale enhancement algorithm based on the Retinex theory is used to improve the visibility of details; the enhancement parameters involved in the corresponding adaptive enhancement process are dynamically adjusted according to the statistical characteristics (brightness mean value, contrast, information entropy, etc.) of the image, to avoid image distortion caused by excessive enhancement;

[0080] The specimen foreground and the standard background in the initial sample image after adaptive enhancement are subjected to image segmentation, and the pure traditional Chinese medicine specimen region is extracted based on the image segmentation result, to obtain a corresponding standard specimen image;

[0081] The feature points in the corresponding standard specimen image are extracted based on the improved SIFT algorithm, to obtain corresponding image feature points; wherein the image feature point acquisition process includes:

[0082] The standard specimen image is subjected to Gaussian convolution processing based on Gaussian convolution kernels at different scales, to obtain Gaussian convolution images at different scales, and then the Gaussian convolution images at adjacent scales are subjected to image difference calculation, to obtain a corresponding difference image;

[0083] Taking a certain pixel point in the difference image as an example, the pixel value corresponding to the corresponding pixel point is obtained, and the pixel value corresponding to all pixel points in the neighborhood pixel set is compared, based on the comparison result, it is determined whether the pixel value corresponding to the corresponding pixel point is a local extreme point, if not, no other operation is performed, if yes, the corresponding pixel point is marked as a candidate key point; the neighborhood pixel set includes other pixel points adjacent to the pixel point in the corresponding difference image and all pixel points at the same position and adjacent positions in the corresponding difference image at adjacent scales;

[0084] Further, the curvature coefficient corresponding to each candidate key point is obtained, and the curvature coefficient is compared with a pre-set curvature threshold value. If the curvature coefficient is not greater than the curvature threshold value, the corresponding candidate key point is marked as an image feature point. If the curvature coefficient is greater than the curvature threshold value, the corresponding candidate key point is discarded. The process of obtaining the curvature coefficient includes: obtaining the pixel value corresponding to the corresponding candidate key point, and obtaining the second-order partial derivatives in the horizontal direction, the vertical direction and the mixed horizontal and vertical direction respectively; and constructing the corresponding curvature matrix based on the same, and obtaining the ratio between the characteristic values corresponding to the corresponding curvature matrix, which is the required curvature coefficient.

[0085] Based on the obtained image feature points, the initial sample image after adaptive enhancement is marked with feature points, and a corresponding feature distribution map is generated. The feature distribution map records the position, intensity and direction information of each image feature point.

[0086] Further, based on the obtained feature distribution map, morphological features, texture features and color features of the corresponding standard specimen image are extracted to obtain corresponding morphological feature vectors, texture feature vectors and color feature vectors. The corresponding feature extraction is realized based on a pre-constructed feature extraction model. The feature extraction model adopts a multi-input channel deep convolutional neural network. The deep convolutional neural network adopts a multi-branch structure. Each branch corresponds to one spectral type (natural light, ultraviolet light and infrared light). Each branch contains 5-7 convolution layers, pooling layers and batch normalization layers. The convolution kernel size varies from 3x3 to 7x7 to adapt to the feature extraction requirements of different scales. Further, the deep convolutional neural network captures the feature differences of the traditional Chinese medicine specimen under each spectral type. The required morphological feature vectors, texture feature vectors and color feature vectors are obtained.

[0087] At the same time, the 3D convolutional neural network is used to process the collected specimen sample video to capture the change features of the traditional Chinese medicine specimen in the time dimension to obtain corresponding video dynamic features, such as fluorescence decay process or appearance change of the specimen under different angles.

[0088] The obtained morphological feature vectors, texture feature vectors, color feature vectors and video dynamic features under different spectral bands are fused with the collected geographical location information to obtain corresponding comprehensive feature vectors. In the feature fusion, the attention mechanism and feature connection technology are used. The self-attention module is used to weight and fuse the features from different sources to highlight important features and suppress redundant information. In the feature fusion process, the geographical location information is converted into a feature vector, which is connected with the image features to form a comprehensive feature representation.

[0089] Further, based on the atlas construction technology, and combined with the specimen database, the specimen metadata, morphological feature vector, texture feature vector, color feature vector, video dynamic feature and comprehensive feature vector corresponding to the corresponding traditional Chinese medicine specimen are displayed in the form of an atlas to obtain the corresponding specimen feature atlas; the specimen feature atlas adopts a multi-level and multi-dimensional atlas architecture, and the atlas architecture includes a basic information layer, an image resource layer and a feature data layer;

[0090] The basic information layer includes recording the basic attributes of the traditional Chinese medicine specimen, including specimen type, collection time, collection location, collection personnel, environmental parameters and other basic information;

[0091] The image resource layer includes storing original specimen sample images, standard specimen images and intermediate results of each processing link; the feature data layer includes a morphological feature sublayer, a color feature sublayer, a texture feature sublayer, a dynamic feature layer and a comprehensive feature sublayer; and the corresponding morphological feature vector, texture feature vector, color feature vector, video dynamic feature and comprehensive feature vector are stored.

[0092] Step three: performing specimen attribute recognition on the obtained specimen feature atlas, and constructing a corresponding traditional Chinese medicine specimen information model based on the specimen attribute recognition result;

[0093] It should be further explained that, in the specific implementation process, the construction process of the traditional Chinese medicine specimen information model includes:

[0094] Based on the existing traditional Chinese medicine knowledge base, a traditional Chinese medicine attribute data set related to traditional Chinese medicine is obtained, and the traditional Chinese medicine attribute data set includes key attribute information of traditional Chinese medicine, such as medicinal properties, tastes, effects, medicinal parts and chemical components;

[0095] Based on the obtained specimen feature atlas, attribute features of the corresponding traditional Chinese medicine specimen are extracted to obtain corresponding specimen attribute features, and the specimen attribute features include morphological features (such as leaf shape, flower shape, fruit morphology, etc.), anatomical features (such as tissue structure, cell characteristics, etc.), biochemical features (such as effective components, content, etc.) and other multiple dimensions;

[0096] The obtained specimen attribute features and the traditional Chinese medicine attribute data set are associated in attributes, and a corresponding classification framework is constructed based thereon; the construction process of the corresponding classification framework is: based on the association relationship between the key attribute information in the traditional Chinese medicine attribute data set and the specimen attribute features, the association relationship is visualized as an attribute association network, and the association strength and direction between each key attribute information and the specimen attribute features are marked; further, the corresponding classification framework is built based on the attribute association network;

[0097] The classification framework is data integrated with the obtained specimen feature atlas to obtain a corresponding informationization model of the medicinal material specimen, which includes feature data, attribute association and classification structure of the medicinal material specimen, and related knowledge of traditional Chinese medicine.

[0098] Step four: performing multi-dimensional feature matching processing on the constructed informationization model of the medicinal material specimen to obtain a corresponding specimen identification result.

[0099] It should be further explained that, in the specific implementation process, the process of obtaining the specimen identification result includes:

[0100] The corresponding informationization database of the medicinal material specimen is matched with the known medicinal material specimen in terms of features to generate a corresponding initial matching result, and the initial matching result includes a plurality of sample matching items; in the feature matching process, firstly, similarity calculation is performed in terms of morphological features, color features, texture features and the like, and the similarity calculation results in each dimension are weighted and summed to obtain a corresponding comprehensive similarity coefficient; in actual application, due to the differences between features, different specimens adopt different similarity calculation methods, for example, morphological features can be evaluated in terms of similarity by calculating Hausdorff distance, and feature similarity calculation of color features needs to collect a histogram intersection algorithm.

[0101] Based on the comprehensive similarity coefficient, the known medicinal material specimen is sorted to obtain a corresponding similarity sorting sequence; based on a pre-set similarity threshold, the known medicinal material specimen in the similarity sorting sequence is pre-screened, and the known medicinal material specimen with a comprehensive similarity coefficient higher than the similarity threshold is taken as a candidate specimen; all the obtained candidate specimens are summarized to obtain a corresponding specimen identification result, which includes specimen ID, name and similarity data of the known medicinal material specimen and the like; the similarity data includes similarity calculation results in multiple dimensions and a comprehensive similarity coefficient.

[0102] Step five: simulating the growth environment of the corresponding medicinal material specimen based on the specimen identification result and the collected environmental information, and evaluating the accuracy of the corresponding specimen identification result based on the growth environment simulation result to obtain a corresponding credibility index.

[0103] It should be further explained that, in the specific implementation process, the process of obtaining the credibility index includes:

[0104] The parameter extraction is performed on the information database of traditional Chinese medicine specimens to obtain ecological environment data related to the corresponding traditional Chinese medicine specimens, and the ecological environment data includes the requirements for environmental factors such as temperature, humidity, illumination, and soil; based on the known professional traditional Chinese medicine plant ecology knowledge base, the environmental parameter range suitable for the growth of the traditional Chinese medicine specimens is extracted to obtain a corresponding specimen growth environment requirement parameter set; the specimen growth environment requirement parameter set includes environmental factors such as temperature adaptation range, humidity requirement, illumination intensity, altitude range, and soil pH value;

[0105] Further, based on the pre-constructed ecological adaptability model, the adaptability of the corresponding traditional Chinese medicine specimens to different environmental factors is obtained to obtain corresponding ecological adaptability indexes;

[0106] The mathematical formula of the ecological adaptability model is defined as:

[0107] In the formula, μ x (i) represents the ecological adaptability index corresponding to the ith environmental factor; x i represents the specific value of the ith environmental factor; T i,min and T i,max respectively represent the left boundary and the right boundary of the parameter corresponding to the ith environmental factor; obtained from the specimen growth environment requirement parameter set; and δ represents the transition interval width.

[0108] Further, based on the ecological adaptability indexes and in combination with the specimen growth environment requirement parameter set, the growth state and performance of the corresponding traditional Chinese medicine specimens under different environmental factors are simulated, and key data points in the simulation process are recorded, including growth rate, morphological change, physiological indexes, etc. The simulation results are integrated to generate a complete specimen ecological adaptability data set.

[0109] The specimen metadata collected is obtained, based on which the environmental data recorded in the collection process of the corresponding traditional Chinese medicine specimens is obtained, and the environmental data includes environmental parameters such as temperature, humidity, illumination, and altitude, and the environmental data is standardized to obtain a corresponding environmental condition parameter set.

[0110] Based on the ecological adaptability model, the ecological adaptability indexes corresponding to each environmental parameter in the corresponding environmental data are obtained, and based on the ecological adaptability indexes, the growth state and performance of the traditional Chinese medicine specimens under the current environmental data are simulated to obtain a corresponding environmental specimen ecological data set.

[0111] The obtained environment sample ecological data set and sample ecological adaptability data set are acquired, and the matching degree between the two is evaluated to obtain a corresponding environment matching degree index; the acquisition process of the corresponding environment matching degree index is that the environment factors and environment parameters in the corresponding environment sample ecological data set and sample ecological adaptability data set are one-to-one corresponding, the parameter difference value and the difference value of the ecological adaptability index between the two are obtained, and the adaptability score is calculated in combination with a pre-set matching degree function to obtain the corresponding environment matching degree index;

[0112] The obtained environment matching degree index is integrated to obtain corresponding sample ecological environment evaluation data, which includes ecological adaptability index, environment matching degree index and optimal growth interval, etc.

[0113] Further, the ecological growth demand of each candidate sample corresponding to the sample recognition result is obtained, and it is compared with the obtained sample ecological environment evaluation data for consistency;

[0114] If the two are inconsistent, the corresponding candidate sample is discarded;

[0115] If the two are consistent, the matching degree between the environment parameters and the ecological growth demand of each candidate sample corresponding to the sample recognition result is obtained to obtain a corresponding environment verification result;

[0116] Further, the sample recognition result corresponding to the corresponding candidate sample is corrected based on the environment verification result to adjust the error in the sample recognition result and improve the accuracy;

[0117] Further, the classification rules and identification points of each type of traditional Chinese medicine are obtained based on the traditional Chinese medicine expert knowledge base, and the evaluation factors of the corresponding sample recognition result are obtained based thereon, including morphological feature coincidence degree, ecological environment consistency, distribution area overlap rate, etc.

[0118] The obtained evaluation factors are weighted and summed to obtain a corresponding credibility index.

[0119] Step six: The credibility index of the sample recognition result is multi-dimensionally comprehensively evaluated, and the evaluation result is integrated with the traditional Chinese medicine sample information to obtain corresponding informationized traditional Chinese medicine sample data archives;

[0120] It needs to be further explained that in the specific implementation process, the construction process of the informationized traditional Chinese medicine sample data archives includes:

[0121] The identification standard of the credibility index is defined, and the credibility level corresponding to the sample recognition result is obtained based thereon, including high credibility, medium credibility and low credibility;

[0122] obtain the data coverage rate and the accuracy rate by comparing the Chinese medicinal material specimen informationization model corresponding to the Chinese medicinal material specimen to be collected with the Chinese medicinal material specimen informationization model corresponding to the candidate specimen in the corresponding sample identification result, and generate a corresponding data integrity evaluation value based on the data coverage rate and the accuracy rate, wherein the higher the data coverage rate and the accuracy rate, the greater the corresponding data integrity evaluation value;

[0123] Integrate the obtained credibility level and data integrity evaluation value, and integrate the relevant information related to the corresponding Chinese medicinal material specimen, generate an informationized Chinese medicinal material specimen digital archive, the informationized Chinese medicinal material specimen digital archive contains specimen basic information, morphological characteristics, classification information, ecological characteristics, value evaluation and digital image, etc. The generated informationized Chinese medicinal material specimen digital archive is arranged in a standard format for easy storage, retrieval and sharing.

[0124] The embodiment realizes accurate and standardized collection of different specifications of Chinese medicinal material specimens by integrating the scalable box body design, multi-spectral lighting technology and intelligent image acquisition; the multi-dimensional features such as morphology, texture and color of the specimen images obtained under multi-spectral conditions are extracted by the deep convolutional neural network, and the complete specimen feature map is constructed by combining the specimen video dynamic characteristics and geographical information. The extracted features are further associated with the traditional Chinese medicine knowledge base to form the mapping relationship between the specimen properties and medicinal properties, and the reliability of the specimen identification result is verified by the ecological adaptability model, and finally the structured Chinese medicinal material specimen digital archive is generated. It has the advantages of strong adaptability of collection equipment, comprehensive data collection dimension, intelligent feature extraction and analysis, systematic knowledge association, etc. It provides a scientific and effective information solution for the digital protection, research and utilization of traditional Chinese medicine resources, and has important value for promoting the modernization development of traditional Chinese medicine.

[0125] Embodiment 2

[0126] Please refer to Figure 2 The embodiment does not describe some parts in detail, see the description of embodiment 1, and provides a Chinese medicinal material specimen collection and information processing system based on image machine learning, which comprises:

[0127] The collection module collects multi-dimensional data of Chinese medicinal material specimens based on the pre-constructed specimen collection platform, obtains corresponding specimen sample images, records specimen metadata, and stores them in the pre-constructed specimen database;

[0128] The analysis module is used for feature extraction of the collected specimen sample images, and constructs a corresponding specimen feature map in combination with the constructed specimen database; based on the feature map, the specimen property is identified, and a corresponding Chinese medicinal material specimen informationization model is constructed;

[0129] The evaluation module is configured to perform multi-dimensional feature matching processing on the constructed informationization model of the traditional Chinese medicine specimen to obtain a corresponding specimen identification result; simulate a growth environment of the corresponding traditional Chinese medicine specimen based on the specimen identification result and the collected environmental information, and perform accuracy evaluation on the corresponding specimen identification result based on the simulation result of the growth environment to obtain a corresponding credibility index;

[0130] The archiving module is configured to perform multi-dimensional comprehensive evaluation on the credibility index of the specimen identification result, and perform information integration based on the evaluation result to obtain a corresponding informationization traditional Chinese medicine specimen data archive.

[0131] The various modules are connected through wired and / or wireless means to realize data transmission between the modules.

[0132] Embodiment 3

[0133] The embodiment discloses an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-provided method for traditional Chinese medicine specimen collection and informationization processing based on image machine learning is realized.

[0134] Since the electronic device introduced in the embodiment is the electronic device used to implement the method for traditional Chinese medicine specimen collection and informationization processing based on image machine learning in the embodiment, the specific implementation mode of the electronic device and its various changes can be understood by those skilled in the art based on the method for traditional Chinese medicine specimen collection and informationization processing based on image machine learning introduced in the embodiment. Therefore, how the electronic device implements the method in the embodiment will not be introduced in detail. As long as the electronic device used to implement the method for traditional Chinese medicine specimen collection and informationization processing based on image machine learning in the embodiment is implemented by those skilled in the art, it belongs to the scope of protection of the present application.

[0135] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the nearest real situation. The preset parameters and threshold values in the formula are set by those skilled in the art according to the actual situation.

[0136] The above is only a preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution falling within the concept of the present application belongs to the protection scope of the present application. It should be noted that, for ordinary technical users in the technical field, some improvements and refinements without departing from the principle of the present application are also considered to be within the protection scope of the present application.

Claims

1. A traditional Chinese medicine specimen collection and information processing method based on image machine learning, characterized in that, Comprise: Step one: based on the pre-constructed specimen collection platform, multi-dimensional data of traditional Chinese medicine specimens are collected, and the image acquisition terminal is based on natural light, ultraviolet light and infrared light to collect images of the specimens under different light spectrum bands, and the specimen sample images are obtained and the specimen metadata are recorded; Store them in the pre-constructed specimen database; Step two: the features of the collected specimen sample images are extracted, the morphological feature vector, texture feature vector, color feature vector and video dynamic feature under different light spectrum bands are obtained, and the geographical location information is fused to obtain the corresponding comprehensive feature vector, and the specimen feature atlas is constructed combined with the constructed specimen database; the atlas architecture includes a basic information layer, an image resource layer and a feature data layer; The image resource layer includes the storage of original specimen sample images, standard specimen images and intermediate results of each processing link; the feature data layer includes a morphological feature sublayer, a color feature sublayer, a texture feature sublayer, a dynamic feature layer and a comprehensive feature sublayer; the corresponding morphological feature vector, texture feature vector, color feature vector, video dynamic feature and comprehensive feature vector are stored respectively; Step three: the specimen attribute recognition is performed on the obtained specimen feature atlas, and the corresponding traditional Chinese medicine specimen information model is constructed based on the specimen attribute recognition result; Step four: the multi-dimensional feature matching processing is performed on the constructed traditional Chinese medicine specimen information model, and the corresponding specimen recognition result is obtained; Step five: based on the specimen recognition result and the collected environmental information, the growth environment of the corresponding traditional Chinese medicine specimen is simulated, and the accuracy of the specimen recognition result is evaluated based on the growth environment simulation result, and the corresponding credibility index is obtained; Step six: the credibility index of the specimen recognition result is comprehensively evaluated in multiple dimensions, and the evaluation result is integrated with the traditional Chinese medicine specimen information to obtain the corresponding informationized traditional Chinese medicine specimen data file; The construction process of the traditional Chinese medicine specimen information model comprises: Based on the existing traditional Chinese medicine knowledge base, the traditional Chinese medicine attribute data set related to traditional Chinese medicine is obtained, which includes the medicinal properties, tastes, effects, medicinal parts and chemical components of traditional Chinese medicine; Based on the obtained specimen feature atlas, the attribute features of the corresponding traditional Chinese medicine specimen are extracted to obtain the corresponding specimen attribute features, including morphological features, anatomical features and biochemical features; The obtained specimen attribute features and traditional Chinese medicine attribute data set are associated with attributes, and a corresponding classification framework is constructed based on the same; the classification framework is integrated with the obtained specimen feature map to obtain a corresponding traditional Chinese medicine specimen informatization model, wherein the traditional Chinese medicine specimen informatization model comprises the , attribute association and classification structure of the traditional Chinese medicine specimen, and related knowledge of traditional Chinese medicine.

2. The method according to claim 1, wherein the traditional Chinese medicine specimen informatization model comprises the following steps: (1) obtaining the traditional Chinese medicine specimen attribute data set; (2) obtaining the traditional Chinese medicine specimen feature map; (3) associating the obtained specimen attribute features and the traditional Chinese medicine attribute data set with attributes, and constructing a corresponding classification framework based on the same; (4) integrating the classification framework with the obtained specimen feature map to obtain a corresponding traditional Chinese medicine specimen informatization model, wherein the traditional Chinese medicine specimen informatization The construction process of the classification framework is: based on the association between the key attribute information in the traditional Chinese medicine attribute data set and the specimen attribute features, the association network is visualized, and the association strength and direction between the key attribute information and the specimen attribute features are marked; based on the attribute association network, the corresponding classification framework is built.

2. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 1, characterized in that, The process of collecting multi-dimensional data of traditional Chinese medicine specimens comprises: Construct a specimen collection platform; the image acquisition terminal adopts a black and white color plate under different light supplement modes based on the pre-set collection requirements to collect images of the specimens to be collected, and the specimen sample images and specimen sample videos are obtained; Synchronously collect geographical information based on the positioning terminal; record metadata information corresponding to each specimen sample image, and associate the metadata information with the collected specimen image set to generate specimen metadata; and upload and store the collected specimen metadata in a pre-constructed specimen database.

3. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 2, characterized in that, The light supplement mode refers to a natural light mode and a fluorescent mode provided by the light supplement terminal. The natural light mode refers to guiding an environmental light source to irradiate the specimen through a natural light guide structure built in the light supplement terminal. The fluorescent mode refers to causing the corresponding image acquisition terminal to perform multispectral band image acquisition on the traditional Chinese medicine specimen to be collected based on ultraviolet light and infrared light respectively through a built-in ultraviolet light LED array with adjustable wavelength and a multi-band infrared light source.

4. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 3, characterized in that, The process of feature extraction on the obtained specimen sample image includes: Obtaining the collected specimen sample image, performing geometric correction on the corresponding specimen sample image based on the scale ruler on the corresponding black and white color base plate in the image acquisition process, and performing polynomial color correction and white balance adjustment on the geometrically corrected specimen sample image to obtain the initial specimen image; Performing local mean filtering processing on the obtained initial specimen image; after the local mean filtering is completed, obtaining the specimen type corresponding to the corresponding traditional Chinese medicine specimen, and performing adaptive enhancement on the corresponding traditional Chinese medicine specimen based on the specimen type to obtain the enhanced specimen image and the standard specimen image based on the enhanced specimen image; Extracting feature points in the corresponding standard specimen image based on the improved SIFT algorithm to obtain the image feature points; Based on the obtained image feature points, marking the feature points of the adaptively enhanced initial specimen image, and generating the corresponding feature distribution map; Based on the obtained feature distribution map, extracting morphological features, texture features and color features of the corresponding standard specimen image to obtain morphological feature vectors, texture feature vectors and color feature vectors; at the same time, processing the collected specimen sample video by using a 3D convolutional neural network to capture the change features of the traditional Chinese medicine specimen in the time dimension to obtain the video dynamic features; including the fluorescent decay process or the appearance change of the specimen at different angles.

5. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 4, characterized in that, The process of extracting feature points in the corresponding standard specimen image based on the improved SIFT algorithm to obtain the image feature points includes: Respectively performing Gaussian convolution processing on the standard specimen image based on Gaussian convolution kernels in different scales to obtain Gaussian convolution images in different scales; performing image difference calculation on the Gaussian convolution images in adjacent scales to obtain the corresponding difference image; obtaining the pixel values corresponding to the pixel points in the corresponding difference image, and comparing the pixel values with the pixel values corresponding to all pixel points in the neighborhood pixel set, determining whether the pixel value corresponding to the corresponding pixel point is a local extreme point based on the comparison result, if not, no other operation is performed, if yes, the corresponding pixel point is marked as a candidate key point; Further, the curvature coefficient corresponding to each candidate key point is obtained, and the curvature coefficient is compared with a pre-set curvature threshold value. If the curvature coefficient is not greater than the curvature threshold value, the corresponding candidate key point is marked as an image feature point. If the curvature coefficient is greater than the curvature threshold value, the corresponding candidate key point is discarded.

6. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 1, characterized in that, The sample identification result acquisition process comprises: The corresponding traditional Chinese medicine specimen information database and the known traditional Chinese medicine specimen are subjected to feature matching, to generate an initial matching result, which comprises a plurality of sample matching items. Based on the comprehensive similarity coefficient, the known traditional Chinese medicine specimen is sorted to obtain a similarity sorting sequence. Based on a pre-set similarity threshold value, the known traditional Chinese medicine specimen in the similarity sorting sequence is pre-screened, and the known traditional Chinese medicine specimen with a comprehensive similarity coefficient higher than the similarity threshold value is taken as a candidate specimen. All the obtained candidate specimens are summarized to obtain a corresponding specimen identification result, which comprises a specimen ID, a name and similarity data of the known traditional Chinese medicine specimen. The similarity data comprises similarity calculation results in multiple dimensions and a comprehensive similarity coefficient.

7. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 6, characterized in that, The process of obtaining the credibility index comprises: The traditional Chinese medicine specimen information database is subjected to parameter extraction to obtain ecological environment data related to the corresponding traditional Chinese medicine specimen. Based on the extraction, an environmental parameter range suitable for the growth of the traditional Chinese medicine specimen is obtained to obtain a specimen growth environment demand parameter set. Based on a pre-constructed ecological adaptability model, the adaptability of the corresponding traditional Chinese medicine specimen to different environmental factors in the specimen growth environment demand parameter set is obtained to obtain an ecological adaptability index. The mathematical formula of the ecological adaptability model is defined as: ; wherein, represents the ecological adaptability index corresponding to the i-th environmental factor; represents the specific value of the i-th environmental factor; and respectively represent the left boundary and the right boundary of the parameter corresponding to the i-th environmental factor; obtained from the specimen growth environment requirement parameter set; represents the transition interval width; Based on the ecological adaptability index and in combination with the specimen growth environment demand parameter set, the growth state and performance of the corresponding traditional Chinese medicine specimen under different environmental factors are simulated, key data points in the simulation process are recorded, and a complete specimen ecological adaptability data set is generated. The specimen metadata collected is obtained, the environmental data recorded in the collection process of the corresponding traditional Chinese medicine specimen is obtained based on the specimen metadata, and the environmental data is standardized to obtain an environmental condition parameter set. Based on the ecological adaptability model, the ecological adaptability index corresponding to each environmental parameter in the corresponding environmental condition parameter set is obtained, and the growth state and performance of the traditional Chinese medicine specimen under the current environmental data are simulated based on the ecological adaptability index to obtain a corresponding environmental specimen ecological data set. The obtained environmental specimen ecological data set and specimen ecological adaptability data set are obtained, and the matching degree between the two is evaluated to obtain a corresponding environmental matching degree index. The obtained environmental matching degree index is integrated to obtain a corresponding specimen ecological environment evaluation data. Based on the sample identification result, the ecological growth demand corresponding to each candidate specimen is obtained, and the ecological growth demand is compared with the obtained specimen ecological environment evaluation data for consistency. If the two are inconsistent, the corresponding candidate specimen is discarded. If the two are consistent, the matching degree between the environmental parameter and the ecological growth demand corresponding to each candidate specimen obtained based on the sample identification result is obtained to obtain a corresponding environmental verification result. Based on the environment verification result, the sample identification result corresponding to the corresponding candidate specimen is corrected. After the correction is completed, the classification rules and identification points of the corresponding types of traditional Chinese medicines are obtained based on the traditional Chinese medicine expert knowledge base, and the evaluation factors of the corresponding specimen identification result are obtained based on the same. The obtained evaluation factors are weighted and summed to obtain the corresponding credibility index. The evaluation factors include morphological feature coincidence, ecological environment consistency and distribution area overlap rate.

8. The image-based machine learning method for traditional Chinese medicine specimen collection and informatization processing according to claim 7, characterized in that, The construction process of the information-based traditional Chinese medicine specimen data file includes: Defining the identification standard of the credibility index, and obtaining the credibility level corresponding to the specimen identification result based on the same; obtaining the information-based model of the traditional Chinese medicine specimen corresponding to the candidate specimen in the corresponding sample identification result, and comparing the information-based model of the traditional Chinese medicine specimen corresponding to the traditional Chinese medicine specimen to be collected with the information-based model of the traditional Chinese medicine specimen, to obtain the corresponding data coverage and accuracy, and generate the corresponding data integrity evaluation value based on the same; The obtained credibility level and data integrity evaluation value are integrated, and the related information involved in the corresponding traditional Chinese medicine specimen is integrated synchronously to generate the information-based traditional Chinese medicine specimen digital file.

Images