CN-122016665-A - Selection method and device for identifying shooting range of cordyceps sinensis hyperspectrum

CN122016665ACN 122016665 ACN122016665 ACN 122016665ACN-122016665-A

Abstract

The present disclosure provides a selection method for identifying photographing range of Cordyceps sinensis hyperspectrum, and relates to the technical field of hyperspectral imaging and intelligent identification of Chinese medicinal materials. The method is concretely realized by shooting and collecting hyperspectral data of a complete cordyceps sinensis sample by using a hyperspectral non-imaging instrument, wherein the shooting range completely covers the parts of the worm body and the sub-seat of the complete cordyceps sinensis sample, and the selection method of the hyperspectral identification shooting range of the cordyceps sinensis can be used for construction of a hyperspectral identification model of the cordyceps sinensis and hyperspectral identification of the cordyceps sinensis. According to the technical scheme, the accuracy and the reliability of the hyperspectral identification of the cordyceps sinensis are improved by definitely defining the hyperspectral shooting range.

Inventors

CHEN XINGFENG
XIE SHIHAO
FANG JUNYONG
WANG JIANWEI
Du Hejuan
Davacha Mar
CUI XINLE
LI JIAGUO
ZHAO LIMIN
LIU JUN
LAN JUN
LIU KUN

Assignees

西藏民族大学
中国科学院空天信息创新研究院
西藏自治区食品药品检验研究院（西藏自治区医疗器械检测中心）

Dates

Publication Date: 20260512
Application Date: 20251216

Claims (7)

1. A selection method for identifying shooting range of cordyceps sinensis hyperspectrum is characterized by comprising the following steps: shooting and collecting hyperspectral data of a complete cordyceps sinensis sample by using a hyperspectral non-imaging instrument, wherein the shooting range completely covers the parts of the worm body and the sub-seat of the complete cordyceps sinensis sample; the selection method of the photographing range for identifying the hyperspectrum of the cordyceps sinensis can be used for establishing a hyperspectral identification model of the cordyceps sinensis and identifying the hyperspectrum of the cordyceps sinensis.
2. The method of claim 1, wherein the constructing of the cordyceps sinensis hyperspectral identification model comprises: shooting and collecting original hyperspectral data of a cordyceps sinensis sample set, wherein the sample set consists of two cordyceps sinensis samples known as wild products or artificial breeds; Preprocessing the original hyperspectral data of the sample set, wherein the preprocessing refers to smooth denoising, baseline correction and normalization processing; Labeling the hyperspectral data of the sample set after pretreatment with wild or artificial breeding article labels, and constructing an independent training data set and an independent verification data set; training the cordyceps sinensis hyperspectral identification model by using the independent training data set, wherein the cordyceps sinensis hyperspectral identification model is a logistic regression model, the classification threshold of the logistic regression model is set to be 0.5, positive class is defined as a wild product, and negative class is defined as an artificial breeding product; using the independent verification data set to evaluate the identification precision of the trained cordyceps sinensis hyperspectral identification model and judging whether the identification precision meets the preset precision requirement or not, wherein, If the preset precision requirement is not met, adjusting parameters of the trained cordyceps sinensis hyperspectral identification model, and reevaluating the identification precision by using the independent verification data set, wherein the parameters comprise regularization parameters, regularization types, optimizers and maximum iteration times; If the preset precision requirement is met, the establishment of the cordyceps sinensis hyperspectral identification model is completed.
3. The method of claim 2, wherein the hyperspectral identification of Cordyceps sinensis comprises: shooting and collecting original hyperspectral data of a cordyceps sinensis sample to be detected; Carrying out pretreatment on the original hyperspectral data of the sample to be detected; Inputting the hyperspectral data of the sample to be tested after pretreatment into the cordyceps sinensis hyperspectral identification model to finish identification.
4. A method according to claim 3, wherein the hyperspectral data comprises: the hyperspectral data wavelength range is 350nm to 2500nm.
5. The device for selecting the photographing range for identifying the hyperspectrum of the cordyceps sinensis is characterized by comprising the following components: The sample placing unit is used for placing and fixing the whole cordyceps sinensis sample; the non-imaging hyperspectral data acquisition unit is used for shooting and acquiring hyperspectral data of the whole cordyceps sinensis sample, and the shooting range completely covers the parts of the worm body and the sub-seat of the whole cordyceps sinensis sample; The data processing unit is in communication connection with the non-imaging hyperspectral data acquisition unit and is used for establishing a hyperspectral identification model of the cordyceps sinensis and identifying the hyperspectrum of the cordyceps sinensis.
6. The apparatus of claim 5, wherein the data processing unit further comprises: the data acquisition module is used for acquiring original hyperspectral data of the cordyceps sinensis sample set and the cordyceps sinensis sample to be detected by utilizing the communication connection; The preprocessing module is used for preprocessing the original hyperspectral data of the sample set and the sample to be detected, wherein the preprocessing refers to smooth denoising, baseline correction and normalization processing; the data set construction module is used for labeling wild products or artificial breeding product labels for the hyperspectral data of the sample set after the pretreatment, and constructing an independent training data set and an independent verification data set; The model training module is used for training the cordyceps sinensis hyperspectral identification model by using the independent training data set, wherein the cordyceps sinensis hyperspectral identification model is a logistic regression model, the classification threshold of the logistic regression model is set to be 0.5, positive class is defined as a wild product, and negative class is defined as an artificial breeding product; the precision judging and parameter adjusting module is used for evaluating the identification precision of the trained cordyceps sinensis hyperspectral identification model by using the independent verification data set and judging whether the identification precision meets the preset precision requirement or not, wherein, If the preset precision requirement is not met, adjusting parameters of the trained cordyceps sinensis hyperspectral identification model, and reevaluating the identification precision by using the independent verification data set, wherein the parameters comprise regularization parameters, regularization types, optimizers and maximum iteration times; If the preset precision requirement is met, the establishment of the cordyceps sinensis hyperspectral identification model is completed; and the identification module is used for inputting the hyperspectral data of the sample to be tested after the pretreatment into the cordyceps sinensis hyperspectral identification model to finish identification.
7. The apparatus of claim 6, wherein the non-imaging hyperspectral data collection unit further comprises a wavelength range of 350nm to 2500nm for the hyperspectral data that can be collected by the non-imaging hyperspectral data collection unit.

Description

Selection method and device for identifying shooting range of cordyceps sinensis hyperspectrum Technical Field The present disclosure relates to the technical field of hyperspectral imaging and the field of intelligent identification of traditional Chinese medicinal materials, and in particular to a method and an apparatus for selecting a photographing range for hyperspectral identification of Cordyceps sinensis. Background Cordyceps sinensis is a precious traditional Chinese medicinal material, has extremely high health care and medicinal value, and has been listed as a national secondary protection endangered species. Research shows that the metabolite composition is obviously influenced by factors such as temperature, soil pressure, illumination intensity and the like in a growth environment, so that the wild cordyceps sinensis and the artificial breeding have obvious differences in the aspects of total amino acid, fatty acid and mineral content, and the differences directly influence the market value and medicinal efficacy. Because cordyceps sinensis is scarce and expensive, the phenomenon that artificially bred cordyceps sinensis is impersonated as wild cordyceps sinensis to be sold in the market, the rights and interests of consumers are seriously damaged, and the medicinal reputation of the cordyceps sinensis is also negatively influenced. Therefore, the method has important practical significance for realizing the accurate and efficient identification of the cordyceps sinensis. Previous identification methods have relied primarily on morphological observation, chemical analysis, or molecular biology techniques, which are generally time consuming, costly, and may destroy the sample, and are difficult to accommodate for the needs of large-scale market supervision. The hyperspectral technology is widely applied to agricultural product and medicinal material identification as a nondestructive and rapid object spectrum information acquisition means, and can realize high-efficiency automatic classification by combining a machine learning algorithm. However, in the actual harvesting and circulation process of Cordyceps sinensis, the separation of the Cordyceps sinensis stroma from the Cordyceps sinensis body is often caused by uncertainty in links such as manual harvesting and transportation. At present, due to the lack of clear specification and theoretical guidance on a shooting range, the data acquisition standard is not uniform, so that the integrity of a constructed spectrum feature set is difficult to ensure, and the accuracy and reliability of a follow-up identification model are further affected. Therefore, a method for clearly guiding the hyperspectral photographing range is needed to solve the problem of insufficient identification accuracy caused by incomplete sample structure and inconsistent data acquisition. Disclosure of Invention The present disclosure provides a selection method and apparatus for identifying photographing range of Cordyceps sinensis hyperspectrum. In a first aspect, the present disclosure provides a method for selecting a photographing range for identifying hyperspectrum of Cordyceps sinensis, including: shooting and collecting hyperspectral data of a complete cordyceps sinensis sample by using a hyperspectral non-imaging instrument, wherein the shooting range completely covers the parts of the worm body and the sub-seat of the complete cordyceps sinensis sample; the selection method of the photographing range for identifying the hyperspectrum of the cordyceps sinensis can be used for establishing a hyperspectral identification model of the cordyceps sinensis and identifying the hyperspectrum of the cordyceps sinensis. In a second aspect, the present disclosure provides a device for selecting a photographing range for identifying hyperspectral of Cordyceps sinensis, including: The sample placing unit is used for placing and fixing the whole cordyceps sinensis sample; the non-imaging hyperspectral data acquisition unit is used for shooting and acquiring hyperspectral data of the whole cordyceps sinensis sample, and the shooting range completely covers the parts of the worm body and the sub-seat of the whole cordyceps sinensis sample; The data processing unit is in communication connection with the non-imaging hyperspectral data acquisition unit and is used for establishing a hyperspectral identification model of the cordyceps sinensis and identifying the hyperspectrum of the cordyceps sinensis. The matters described in this section are not intended to identify key or critical features of the embodiments of the disclosure, nor are they to be construed as limiting the scope of the disclosure. Other features of the present disclosure will be described in detail in the following specification to aid understanding. Drawings FIG. 1 is a schematic diagram of an implementation scenario of an embodiment of the present disclosure; FIG. 2 is a diagram showing a whole Cordyceps sinensi