Search

CN-121982414-A - Ginseng age lossless classification model construction method and ginseng age lossless classification system

CN121982414ACN 121982414 ACN121982414 ACN 121982414ACN-121982414-A

Abstract

A ginseng age lossless classification model construction method and a ginseng age lossless classification system, in particular to a ginseng growth age classification method based on hyperspectral imaging technology and improved group intelligent optimization algorithm. The method solves the problems of high-dimensional data redundancy, insufficient map feature utilization, insufficient model generalization capability, low precision of fine classification aiming at adjacent years of cultivated ginseng and the like in the prior art. The method comprises the following steps of 1, obtaining hyperspectral image data of ginseng of different ages and preprocessing, 2, screening spectral features and texture features based on the hyperspectral image data, 3, carrying out feature level fusion on the spectral features and the texture features to obtain fusion features, and 4, inputting the fusion features into a IHSO-RF model constructed for training and testing. The classification model is embedded in the system, so that classification of ginseng ages is realized. The invention is suitable for the fields of intelligent perception and analysis based on hyperspectrum and machine vision, nondestructive detection and classification of quality of agricultural products/Chinese medicinal materials and the like.

Inventors

  • XUE MINGXUAN
  • LIU BIN
  • YU HELONG
  • SHI LEI
  • ZHANG SHUQI
  • HAN QINGYANG
  • ZHANG TONGYIN

Assignees

  • 吉林农业大学

Dates

Publication Date
20260505
Application Date
20260130

Claims (10)

  1. 1. A method for constructing a ginseng age lossless classification model is characterized by comprising the following steps: Step 1, obtaining hyperspectral image data of ginseng of different ages, and preprocessing the hyperspectral image data; step 2, screening spectral features and texture features of the preprocessed hyperspectral image data; Step 3, carrying out feature level fusion on the spectral features and the texture features to obtain fusion features; Step 4, improving the overall group optimization algorithm HSO, constructing IHSO-RF model, Inputting the fusion characteristics into the IHSO-RF model to obtain classification of ginseng ages.
  2. 2. The method for constructing a lossless classification model of ginseng age according to claim 1, wherein in step4, the overall group optimization algorithm HSO is improved as follows: In the HSO algorithm population initialization stage, a cascading Logistic-Tent chaotic mapping mechanism is adopted; in the position updating stage of the HSO algorithm, a self-adaptive polynomial and an index updating fusion strategy are adopted; in the self-adaptive variation stage of the HSO algorithm, self-adaptive attenuation is adopted And (5) a flight fusion sinusoidal variation strategy.
  3. 3. The method for constructing a ginseng age lossless classification model according to claim 2, wherein the cascade logic-Tent chaotic mapping mechanism is a control parameter for dynamically generating a logic-Tent chaotic mapping system through a two-dimensional sine-logic modulation mapping, and specifically comprises the following steps: In the formula, Indicating time of day Is used to determine the state variable of (1), ∈[0,1], A state variable representing the next time; Representing modulo arithmetic; And Is a time-varying parameter.
  4. 4. The method for constructing a ginseng age lossless classification model according to claim 3, wherein the Logistic-Tent chaotic mapping system is as follows: In the formula, A state variable representing the state of the system, ∈[0,1], Represent the first State variables of the second iteration; Is a control parameter and ∈(0,4); A decimal portion representing the result; the two-dimensional sine-logic modulation map is: In the formula, And E (0, 1) represents the time of day Is used to determine the state variable of (1), And E (0, 4) represents a control parameter; Depending on And The weights of the Logistic term weighted combination of (a) are respectively as follows And By means of The modulation of the function is performed, Dependent on the new calculation And the original The weights are respectively And By means of And (5) modulating a function.
  5. 5. The method for constructing a ginseng age lossless classification model according to claim 2, wherein the adaptive polynomial and index updating fusion strategy is to introduce adaptive weights The polynomial and index updating fusion strategy is improved, and specifically comprises the following steps: In the formula, Is the lower limit value of the step size, Is the upper limit value of the step size, For the current number of iterations, For the total number of iterations, Is the coefficient of the exponential decay, Representing the adaptive weights.
  6. 6. The method for constructing a ginseng age lossless classification model according to claim 2, wherein the adaptive attenuation The flight fusion sinusoidal variation strategy is specifically as follows The global exploration advantage of the flight and the local development advantage of the sinusoidal variation are intelligently fused through a self-adaptive attenuation mechanism, and the method specifically comprises the following steps: In the formula, The current position is indicated and the current position is indicated, The position after the update of the mutation is indicated, The adaptive weights are represented by the weights, Representing the use for long-distance jumps, Indicating the decay over time of the light, Representing sinusoidal variation terms, wherein As a random vector, randomness is introduced.
  7. 7. A ginseng age nondestructive classification detection system is characterized by comprising a data acquisition device, a data processing device and a data classification device: the data acquisition module is used for acquiring hyperspectral image data of the ginseng of different ages, wherein the hyperspectral image data comprises hyperspectral data and image data, and the hyperspectral data and the image data are transmitted to the data processing device; The data processing device is embedded with a data processing module realized by a computer program, and the data processing module comprises the following data processing units: the preprocessing unit is used for respectively preprocessing the hyperspectral data and the image data; The screening unit is used for screening important features of the preprocessed hyperspectral data and the preprocessed image data respectively; The fusion unit is used for carrying out feature level fusion on the important features of the hyperspectral data and the image data to obtain fusion features, and outputting the fusion features to the data classification device; The data classification device is embedded with the IHSO-RF model constructed by the ginseng age lossless classification model construction method, and is used for receiving the fusion characteristics and obtaining ginseng age classification based on the fusion characteristics.
  8. 8. A computer apparatus/device/system comprising a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to implement the steps of the method of any of claims 1-6.
  9. 9. A computer-readable storage medium having stored thereon computer programs/instructions, which are specific to Characterized in that the computer program/instructions, when executed by a processor, implement the steps of the method of any of claims 1 to 6.
  10. 10. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any of claims 1 to 6.

Description

Ginseng age lossless classification model construction method and ginseng age lossless classification system Technical Field The invention relates to the technical field of nondestructive testing of quality of traditional Chinese medicinal materials, in particular to a ginseng growth age classification method based on a hyperspectral imaging technology and an improved group intelligent optimization algorithm. Background Ginseng is a precious Chinese medicinal material, and its quality is closely related to growth years. With the increasing exhaustion of wild resources, ginseng cultivation has become a mainstream of the market. However, there is a fraudulent act on the market that impersonates low-age ginseng as high-age ginseng sales, severely affecting market order and consumer equity. Therefore, the accurate and rapid identification of the growth years of the ginseng is realized, and the method has important significance for guaranteeing the quality of medicinal materials and normalizing market behaviors. At present, the identification method of ginseng age mainly comprises the modern analysis technologies of traditional character identification, microscopic identification, chromatography, spectrum and the like. The traditional method relies on manual experience, has strong subjectivity and low efficiency, and has higher accuracy in chromatographic techniques and the like, but the sample pretreatment is complex and takes longer time, and can damage the sample, so that the requirements of large-scale nondestructive detection are difficult to meet. In recent years, the hyperspectral imaging technology has become a research hotspot for nondestructive identification of traditional Chinese medicinal materials because of the capability of acquiring spectrum and spatial information of a sample at the same time. The technology provides possibility for realizing rapid and nondestructive classification of ginseng years by extracting the intrinsic characteristics related to the years. However, hyperspectral data has high dimensionality and high information redundancy, and direct use of full-band data modeling tends to result in heavy model calculation burden and weak generalization capability. In addition, the existing research still has limitations in terms of the number of samples, the coverage of years and the feature fusion strategy, for example, most methods only use spectrum information, but neglect spatial features such as textures and the like, and further improvement of classification accuracy is affected. In summary, the technical problems faced by the current classification of ginseng age are that the prior art can acquire spectrum and image information at the same time, but the prior art still faces the problems of high-dimensional data redundancy, insufficient utilization of spectrum features, insufficient model generalization capability, low precision of fine classification aiming at adjacent ages of cultivated ginseng and the like in practical application. Disclosure of Invention The invention solves the problems of high-dimensional data redundancy, insufficient map feature utilization, insufficient model generalization capability, low precision of fine classification aiming at adjacent years of cultivated ginseng and the like in the prior art. A method for constructing a ginseng age lossless classification model comprises the following steps: Step 1, obtaining hyperspectral image data of ginseng of different ages, and preprocessing the hyperspectral image data; step 2, screening spectral features and texture features of the preprocessed hyperspectral image data; Step 3, carrying out feature level fusion on the spectral features and the texture features to obtain fusion features; Step 4, improving the overall group optimization algorithm HSO, constructing IHSO-RF model, Inputting the fusion characteristics into the IHSO-RF model to obtain classification of ginseng ages. In a further preferred embodiment, in step 4, the improvement of the overall group optimization algorithm HSO specifically includes: In the HSO algorithm population initialization stage, a cascading Logistic-Tent chaotic mapping mechanism is adopted; in the position updating stage of the HSO algorithm, a self-adaptive polynomial and an index updating fusion strategy are adopted; in the self-adaptive variation stage of the HSO algorithm, self-adaptive attenuation is adopted And (5) a flight fusion sinusoidal variation strategy. In a further preferred scheme, the cascade logic-Tent chaotic mapping mechanism dynamically generates control parameters of the logic-Tent chaotic mapping system through a two-dimensional sine-logic modulation mapping, and specifically comprises the following steps: In the formula, Indicating time of dayIs used to determine the state variable of (1),∈[0,1],A state variable representing the next time; Representing modulo arithmetic; And Is a time-varying parameter. Further preferably, the Logistic-Tent chaotic ma