Search

CN-122016713-A - Near infrared hyperspectral imaging method for real-time detection of microbial colony number of salmon fillet

CN122016713ACN 122016713 ACN122016713 ACN 122016713ACN-122016713-A

Abstract

The invention discloses a near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillets in real time, which comprises the following steps of S1, preparing salmon fillets samples exposed by different time air, detecting the microbial colony number, S2, carrying out spectral imaging measurement, collecting hyperspectral imaging HSI data, S3, carrying out region of interest identification and spectral data extraction based on the collected data in S2, S4, dividing the samples into a training set, a prediction set and an independent external test set, S5, establishing a salmon fillet APC spectrum quantitative analysis model and a salmon fillet APC standard exceeding discrimination spectrum identification model, S6, carrying out standard normal variable transformation (SNV) and Norris Derivative Filtering (NDF) pretreatment on the spectral data, S7, optimizing a wavelength combination by a two-stage wavelength selection method of moving a window and multi-wavelength phase-out, and S8, and verifying model accuracy by using the independent test set. Compared with the prior art, the near infrared hyperspectral imaging method has the advantages that the near infrared hyperspectral imaging method for real-time detection of the microbial colony number of the salmon fillet is provided for developing a small special NIR-HSI detector for detecting the salmon fillet.

Inventors

  • PAN TAO
  • Pan Chubing

Assignees

  • 广州源谱成像科技有限公司

Dates

Publication Date
20260512
Application Date
20260213

Claims (10)

  1. 1. The invention discloses a near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillet in real time, which comprises the following steps: s1, preparing salmon samples exposed by air at different time, and detecting the number of microbial colonies; s2, performing spectral imaging measurement and collecting hyperspectral imaging HSI data; s3, identifying an interested region and extracting spectral data based on the acquired data in the S2; S4, dividing the sample into a training set, a prediction set and an independent external test set; s5, establishing a salmon fillet APC (automatic control) spectrum quantitative analysis model and a spectrum identification model for the out-of-standard judgment of the salmon fillet APC; s6, carrying out standard normal variable transformation SNV and Norris derivative filtering NDF pretreatment on the optical data; s7, optimizing wavelength combination by a two-stage wavelength selection method of moving window and multi-wavelength phase-out; And S8, verifying model accuracy by using the independent test set. Compared with the prior art, the near infrared hyperspectral imaging method has the advantages that the near infrared hyperspectral imaging method for real-time detection of the microbial colony number of the salmon fillet is provided for developing a small special NIR-HSI detector for detecting the salmon fillet.
  2. 2. The near infrared hyperspectral imaging method for detecting the microbial colony count of salmon fillet in real time according to claim 1, wherein the step S1 is characterized in that the salmon fillet sample is exposed to air for 0-24 hours, 4 fillets are sampled per hour, and the microbial colony count APC is measured by a microbial plate counting method.
  3. 3. The near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillet in real time according to claim 1 is characterized in that a near infrared hyperspectral imager with a spectral range of 883-2492 nm is adopted in the step S2, 4 samples are placed on background white paper for reference in an open mode, and placed on a movable objective table, and the objective table moves at a speed of 10mm/S to acquire HSI data.
  4. 4. The near infrared hyperspectral imaging method for real-time detection of the microbial colony number of salmon fillet according to claim 2, wherein S3 comprises the steps of constructing a pseudo-color image in an NIR double spectrum region to identify the salmon fillet region, and extracting average spectrums in a plurality of regions of interest (ROI).
  5. 5. The near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillets in real time according to claim 1 is characterized in that partial least squares is adopted as a primitive algorithm in S5, an APC spectrum quantitative analysis model of the salmon fillets is built, and partial least squares-discriminant analysis is adopted as a classifier primitive algorithm.
  6. 6. The near infrared hyperspectral imaging method for real time detection of the microbial colony number of salmon fillet according to claim 5, wherein S6 comprises preprocessing by adopting standard normal variable transformation SNV and Norris derivative filtering NDF, and selecting preprocessing parameters based on modeling effects of the two models.
  7. 7. The near infrared hyperspectral imaging method for real-time detection of the microbial colony number of salmon fillet according to claim 5, wherein the moving window band selection method based on the variable starting point wavelength and the variable wavelength number in S7 is used for the wavelength selection in the first stage; On the basis of the wavelength selection of the first stage, a wavelength selection method of multi-wavelength phase-out is used for the wavelength selection of the second stage; the first and second phases are based on modeling effects of two models, preferably wavelength combinations.
  8. 8. The method for near infrared hyperspectral imaging of real time detection of microbial colony count of salmon fillet according to claim 7, wherein the wavelength combination for quantitative analysis of APC in step S7 comprises 31 wavelengths ：1276nm、1285nm、1295nm、1304nm、1371nm、1410nm、1477nm、1572nm、1611nm、1620nm、1630nm、1697nm、1707nm、1860nm、1927nm、1965nm、2003nm、2042nm、2051nm、2090nm、2099nm、2109nm、2118nm、2128nm、2137nm、2204nm、2214nm、2252nm、2262nm、2271nm、2338nm.
  9. 9. The method for near infrared hyperspectral imaging of real time detection of microbial colony count of salmon fillet according to claim 7, wherein the wavelength combination for out-of-standard discrimination in step S7 comprises 15 wavelengths ：998nm、1017nm、1055nm、1094nm、1113nm、1151nm、1161nm、1189nm、1218nm、1237nm、1324nm、1343nm、1352nm、1362nm、1381nm.
  10. 10. The near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillet in real time according to claim 4, wherein in the S3, a pseudo-color image is constructed by selecting three wavelengths of 1103nm, 1448nm and 1601nm in an NIR double spectrum region of 1000-1700 nm, and an average spectrum in 4 regions of interest (ROI) is extracted.

Description

Near infrared hyperspectral imaging method for real-time detection of microbial colony number of salmon fillet Technical Field The invention relates to the technical field of spectrum of food safety detection, in particular to a near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillet in real time. Background The raw salmon fillet is rich in nutrition and delicious in taste, is a mass food with huge sales scale, and is also a very easily-degraded food, and the salmon fillet is exposed in the air in the links of making, eating and the like, so that microorganisms can grow on the surface, and the food safety problem is caused. Traditional detection of total bacterial count (APC) of fish microorganisms mainly adopts a plate colony counting method, which requires reagents, aseptic operation and microorganism culture, is complex and time-consuming, and is not suitable for on-site real-time detection of salmon fillet. Near Infrared (NIR) spectroscopy has the advantage of rapid, simple detection. Meat products contain a large amount of hydrogen-containing groups X-H in the molecule and have absorption characteristics in the NIR spectrum. The salmon fillet sample is exposed to air and polluted by environmental microorganisms, and the related microorganism metabolic processes generate a plurality of novel metabolites containing hydrogen groups, so that the overall spectrum patterns before and after pollution are changed. Thus, NIR spectral pattern recognition has molecular rationality for detecting salmon fillet APC. The hyperspectral imaging (HSI) technology has the functions of spectrum detection and image analysis, has the analysis advantages of nondestructive and non-contact detection, and has demonstrated application potential in a plurality of fields. According to literature search, near infrared-hyperspectral imaging (NIR-HSI) has not been applied to real-time detection of the number of microbial colonies of salmon fillet. The invention discloses a modeling and wavelength selection method for real-time detection of the microbial colony number of salmon fillet based on NIR-HSI. Disclosure of Invention The invention aims to overcome the technical defects and provide a near infrared hyperspectral imaging method for real-time detection of the microbial colony number of salmon fillet, which provides a solution for developing a small special NIR-HSI detector for detecting salmon fillet. In order to solve the technical problems, the technical scheme provided by the invention is that the near infrared hyperspectral imaging method for detecting the microbial colony number of salmon fillet in real time comprises the following steps: s1, preparing salmon samples exposed by air at different time, and detecting the number of microbial colonies; s2, performing spectral imaging measurement and collecting hyperspectral imaging HSI data; s3, identifying an interested region and extracting spectral data based on the acquired data in the S2; S4, dividing the sample into a training set, a prediction set and an independent external test set; s5, establishing a salmon fillet APC (automatic control) spectrum quantitative analysis model and a spectrum identification model for the out-of-standard judgment of the salmon fillet APC; s6, carrying out standard normal variable transformation SNV and Norris derivative filtering NDF pretreatment on the optical data; s7, optimizing wavelength combination by a two-stage wavelength selection method of moving window and multi-wavelength phase-out; And S8, verifying model accuracy by using the independent test set. Preferably, the step S1 includes exposing the salmon side sample to air for 0-24 hours, sampling 4 pieces per hour, and determining the total colony count APC by using a microbial plate count method. Preferably, a near infrared hyperspectral imager with a spectral range of 883-2492 nm is adopted in the step S2; the 4 samples were placed on background white paper for reference at intervals, and placed on a moving stage, which was moved at 10mm/s, and HSI data collection was performed. Preferably, the step S3 includes constructing a pseudo-color image to identify salmon fillet area in the NIR double spectrum area, and extracting average spectra in multiple regions of interest ROIs. Preferably, in the step S5, partial least square is adopted as a primitive algorithm, and an APC spectrum quantitative analysis model of the salmon fillet is established; partial least squares-discriminant analysis is used as the classifier primitive algorithm. Preferably, the step S6 includes preprocessing with standard normal variable transformation SNV and Norris derivative filtering NDF, and selecting preprocessing parameters based on modeling effects of the two models. Preferably, the moving window band selection method based on the variable starting point wavelength and the number of wavelengths in S7 is used for the wavelength selection in the first stage; O