CN-121978100-A - Method for rapidly determining gelatinization degree of fermented grains based on machine vision detection

CN121978100ACN 121978100 ACN121978100 ACN 121978100ACN-121978100-A

Abstract

The invention mainly relates to the technical field of white spirit brewing, and in order to improve the efficiency and accuracy of detecting the gelatinization degree of fermented grains in white spirit, the invention provides a method for rapidly measuring the gelatinization degree of fermented grains based on machine vision detection, which comprises the steps of collecting a fermented grain sample image through a machine vision detection technology, and extracting a color with a larger area as a characteristic color number; and establishing a gelatinization degree prediction model by taking the area ratio of the key characteristic color number in an image as an independent variable and the gelatinization degree as a dependent variable, thereby realizing rapid, quantitative and nondestructive detection of the gelatinization degree of the fermented grains.

Inventors

LI MI
CHEN SHIJIANG
CHEN XINYU
LI QIJUN
HUANG YULONG
SU JIAN
ZHANG YUPENG
JIANG QINAN

Assignees

宜宾五粮液股份有限公司
四川省食品发酵工业研究设计院有限公司

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (8)

1. The method for rapidly determining the gelatinization degree of the fermented grains based on the machine vision detection is characterized by comprising the following steps of: preparing fermented grain samples with different gelatinization degrees, collecting an original image of the sample, preprocessing the original image, and taking a color number with an area ratio larger than a set value as a characteristic color number; Screening key feature color numbers related to the gelatinization degree from the feature color numbers, and counting the area occupation ratio of each key feature color number; Establishing a gelatinization degree prediction model based on the key characteristic color number area occupation ratio; And extracting the key characteristic color number area ratio of the fermented grain sample to be detected, and measuring and calculating the gelatinization degree of the fermented grain sample to be detected based on a gelatinization degree prediction model.
2. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to claim 1, further comprising extracting a characteristic color number with an area ratio larger than a set value, and then performing principal component analysis on the extracted characteristic color number to verify the feasibility of measuring the gelatinization degree of the fermented grains based on the characteristic color number.
3. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to claim 2, wherein the performing principal component analysis on the extracted characteristic color number, verifying the feasibility of measuring the gelatinization degree of the fermented grains based on the characteristic color number comprises: Carrying out standardization processing on the extracted characteristic color number duty ratio data; Calculating a correlation matrix of the standardized data, and carrying out eigenvalue decomposition on the correlation matrix to obtain eigenvalues and variance contribution rates corresponding to the principal components; screening effective principal components according to the set eigenvalue threshold and the accumulated variance contribution rate threshold; Based on the scores of the effective main components, a main component score graph is generated, and the feasibility of measuring and calculating the gelatinization degree of the fermented grains based on the characteristic color numbers is verified.
4. The method for rapidly determining the gelatinization degree of fermented grains based on machine vision inspection according to claim 1, wherein the gelatinization degree range is 50% -95% when the fermented grains samples with different gelatinization degrees are prepared.
5. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to claim 1, wherein the step of screening the key feature color numbers related to the gelatinization degree from the feature color numbers comprises the step of screening the key feature color numbers related to the gelatinization degree from the feature color numbers based on a partial least square method.
6. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to claim 1, wherein the screening of key feature color numbers related to the gelatinization degree from the feature color numbers comprises: Obtaining a plurality of fermented grain sample images with known gelatinization degree, and extracting characteristic color numbers with area occupation ratios larger than a set threshold value in each sample and the area occupation ratios of the characteristic color numbers; calculating the space gradient value of the characteristic color number Wherein And Respectively is a characteristic color number Average area ratio in the high-degree-of-gelatinization sample group and the low-degree-of-gelatinization sample group, Is the characteristic color number Standard deviation of area ratio in all samples; calculating characteristic color number pasting degree sensitivity Calculating local sensitivity between adjacent gelatinization degree fermented grain samples Will be Is taken as the degree of gelatinization sensitivity Wherein, the method comprises the steps of, And Is the characteristic color number In the first place The area ratio and the degree of gelatinization in each sample, And Is characterized by color number i in the first Area ratio and gelatinization degree in each sample; Spatial gradient value based on characteristic color number And degree of gelatinization sensitivity Calculating characteristic color numbers Weight of (2): , wherein, Is the characteristic color number Is a normalized value of the frequency of occurrence of (a), 、 And For the preset weighting coefficient to be used, And The characteristic color number space gradient values and the maximum value of the gelatinization degree sensitivity in all samples are respectively; according to the characteristic color number And (3) sorting all the feature color numbers according to the weights, and taking the feature color numbers with the first M bits or the weights higher than a set threshold T as key feature color numbers.
7. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to claim 6, wherein the gelatinization degree of the high-gelatinization degree sample group is 85% -95% and the gelatinization degree of the low-gelatinization degree sample group is 50% -60%.
8. The method for rapidly determining the gelatinization degree of fermented grains based on the machine vision inspection according to any one of claims 1 to 6, wherein the establishing a gelatinization degree prediction model based on the area ratio of key feature color numbers comprises establishing a gelatinization degree multiple linear regression model with the area ratio of key feature color numbers as independent variables by taking the gelatinization degree of the fermented grains as the dependent variable, wherein the model expression is: wherein In order to achieve a degree of gelatinization, For the area ratio of each key feature color number, For the weight coefficient corresponding to each key feature color number, Is a constant term.

Description

Method for rapidly determining gelatinization degree of fermented grains based on machine vision detection Technical Field The invention mainly relates to the technical field of white spirit brewing, in particular to a method for rapidly determining the gelatinization degree of fermented grains based on machine vision detection. Background The degree of gelatinization of fermented grains is a key quality index in the process of brewing white spirit, and directly influences the starch conversion efficiency, the fermentation effect of the fermented grains and the final yield and quality of the white spirit, so that the fermented grains are required to be accurately and timely monitored. At present, the method for measuring the gelatinization degree of the fermented grains in the white wine brewing industry mainly comprises a chemical analysis method, including an enzymolysis method, a titration method and the like. The method needs complex pretreatment on the sample, consumes a large amount of chemical reagents, has long detection period, usually needs several hours to tens of hours, has low detection efficiency, can generate chemical waste liquid to cause environmental pollution, and meanwhile, the chemical method can not realize on-line real-time monitoring on a production site, has obvious delay on detection results, so that the process parameters can not be adjusted according to the change of gelatinization degree in time in the production process, and the brewing efficiency and the product stability are affected. In the spectroscopic analysis technique, although the near infrared spectrometer, the ultraviolet-visible spectrometer, the raman spectrometer and the like can realize quantitative analysis of part of substances, the method has obvious limitation in industrial field detection of complex substrate samples such as fermented grains and the like. The fermented grain sample has complex components and contains various substances such as starch, moisture, cellulose, microorganism metabolites and the like, the existing spectrum technology is easy to be interfered by sample matrixes, huge basic data is needed to be used as support, data correction is carried out through a complex chemometric algorithm, the correction effect is unstable, the fluctuation of detection precision is large, meanwhile, the existing spectrum technology has extremely high requirement on the uniformity of the sample, and the difference of the grain size and the tightness degree of the fermented grain can obviously influence the accuracy of spectrum data. The machine vision detection system (electronic eye) is used as a rapid detection technology, is widely applied to the field of qualitative analysis of samples, is mainly used for detecting surface features such as color difference and appearance form of the samples, and has the advantages of high detection speed, simplicity and convenience in operation, no contact pollution and the like. However, the existing electronic eye technology can only realize qualitative distinction between samples, cannot perform quantitative analysis on specific quality indexes, and is difficult to meet the accurate quantification requirements of key parameters such as the gelatinization degree of fermented grains. Disclosure of Invention The invention aims to solve the technical problem of providing a method for rapidly determining the gelatinization degree of fermented grains based on machine vision detection, and aims to improve the efficiency and accuracy of the detection of the gelatinization degree of the fermented grains. The invention solves the technical problems by adopting the technical scheme that: a method for rapidly determining the gelatinization degree of fermented grains based on machine vision detection, the method comprising: preparing fermented grain samples with different gelatinization degrees, collecting an original image of the sample, preprocessing the original image, and taking a color number with an area ratio larger than a set value as a characteristic color number; Screening key feature color numbers related to the gelatinization degree from the feature color numbers, and counting the area occupation ratio of each key feature color number; Establishing a gelatinization degree prediction model based on the key characteristic color number area occupation ratio; And extracting the key characteristic color number area ratio of the fermented grain sample to be detected, and measuring and calculating the gelatinization degree of the fermented grain sample to be detected based on a gelatinization degree prediction model. Further, the method further comprises the steps of extracting the characteristic color number with the area occupation ratio larger than a set value, performing principal component analysis on the extracted characteristic color number, and verifying the feasibility of measuring and calculating the gelatinization degree of the fermented grains based on the characte