CN-121978096-A - Method and system for rapidly detecting nutritional ingredient defects of non-fried instant noodles

CN121978096ACN 121978096 ACN121978096 ACN 121978096ACN-121978096-A

Abstract

The invention discloses a method and a system for rapidly detecting nutritional ingredient defects of non-fried instant noodles, and relates to the technical field of food detection, wherein the method comprises the steps of adopting a dual-band light source to synchronously scan multiple types of non-fried instant noodles, constructing a three-dimensional spectral database of nutritional ingredient defects, and establishing a nutritional ingredient defect prediction channel; and the two channels are subjected to weighted fusion to form a multi-mode defect detection channel, the spectrum and image data of the target instant noodle sample are subjected to joint analysis, and an nutritional ingredient defect detection result is output. The invention solves the technical problems of long detection period and low accuracy of the prior art that the product is subjected to spot check by depending on laboratory physicochemical analysis or single spectrum technology, and achieves the technical effect of high-precision and rapid detection of the nutritional defects of the non-fried instant noodles by combining a multi-mode detection mechanism of spectrum analysis and image analysis.

Inventors

PAN GUOQIANG
ZHAN SHAOWEI

Assignees

南通昶昊机电制造有限公司

Dates

Publication Date: 20260505
Application Date: 20251202

Claims (10)

1. The method for rapidly detecting the nutritional ingredient defects of the non-fried instant noodles is characterized by comprising the following steps of: synchronously scanning a plurality of types of non-fried instant noodles by a dual-band light source to establish a nutritional ingredient defect three-dimensional spectrum database, wherein the dual-band light source comprises a near infrared spectrum and a Raman spectrum; Performing nutritional ingredient prediction fitting based on the nutritional ingredient defect three-dimensional spectrum database, and constructing a nutritional ingredient defect prediction channel; Acquiring a visual auxotroph image set of the multi-type non-fried instant noodles, and performing feature identification and defect evaluation on the visual auxotroph image set to obtain a visual auxotroph evaluation channel; Combining the nutritional ingredient defect prediction channel and the visual nutritional ingredient defect assessment channel to establish a nutritional ingredient defect detection multi-mode channel; and performing defect rapid detection on the spectrum data and the image data of the target non-fried instant noodles based on the nutritional ingredient defect detection multi-mode channel, and determining a nutritional ingredient defect detection result of the target instant noodles.
2. The method for rapidly detecting nutritional defects in non-fried instant noodles according to claim 1, wherein the creating a three-dimensional spectral database of nutritional defects comprises: Randomly extracting the non-fried instant noodles in the multiple types to obtain a non-fried instant noodle sample set, wherein the non-fried instant noodle sample set comprises a noodle cake, a vegetable packet and a sauce packet; Preprocessing the flour cakes, the vegetable bags and the sauce bags in the non-fried instant noodle sample set to obtain a usable non-fried instant noodle sample set; Installing a high-precision spectrometer, wherein the high-precision spectrometer is a dual-band light source spectrometer with an infrared spectrum and a Raman spectrum, and acquiring an instant noodle spectrum data set of the available non-fried instant noodle sample set through the high-precision spectrometer; And carrying out alignment fusion and nutrition association on the spectrum data set of the instant noodles to obtain a nutritional ingredient defect three-dimensional spectrum database.
3. The method for rapidly detecting nutritional defects in non-fried instant noodles according to claim 2, wherein the obtaining a three-dimensional spectral database of nutritional defects comprises: setting a spectrum data preprocessing program, wherein the spectrum data preprocessing program comprises filtering processing, baseline correction, scattering correction and spectrum normalization; preprocessing the spectrum data set of the instant noodles according to the spectrum data preprocessing program to obtain a standard spectrum data set of the instant noodles; performing band alignment and fusion processing on the standard instant noodle spectrum data set to obtain a usable instant noodle spectrum data set; And carrying out nutrient content measurement and data association on the available instant noodle spectrum data set according to the instant noodle nutrient content standard to obtain a nutrient content defect three-dimensional spectrum database.
4. A method for rapid detection of nutritional defects in non-fried instant noodles according to claim 3, wherein said obtaining a three-dimensional spectral database of nutritional defects comprises: Extracting the nutrition index of the instant noodle nutrition ingredient standard to obtain an instant noodle nutrition index set, and determining a nutrition ingredient measuring program according to the instant noodle nutrition index set; carrying out multiple nutrient determination on each instant noodle sample in the available instant noodle spectrum data set by using the nutrient determination program to obtain a nutrient test data set; Performing marking recovery verification and nutrition comparison calculation on the nutrition component test data set based on the nutrition component standard of the instant noodles to obtain a nutrition component defect parameter set; and carrying out association processing on the nutritional ingredient defect parameter set and the available instant noodle spectrum data set to obtain the nutritional ingredient defect three-dimensional spectrum database.
5. The method for rapidly detecting nutritional defects in non-fried instant noodles according to claim 1, wherein the constructing a nutritional defect prediction channel comprises: Performing dimensional data extraction on the nutritional ingredient defect three-dimensional spectrum database to generate a spectrum data matrix and a nutritional ingredient defect matrix; Performing continuous projection dimension reduction and characteristic wavelength selection based on the spectrum data matrix to obtain an available spectrum data matrix; Decomposing the available spectrum data matrix and the nutritional ingredient defect matrix to obtain a spectrum score matrix and a nutritional ingredient score matrix; And carrying out data regression fitting based on the spectrum score matrix and the nutritional ingredient score matrix to construct the nutritional ingredient defect prediction channel.
6. The method for rapid detection of nutritional defects in non-fried instant noodles of claim 1, wherein said obtaining a visual nutritional defect assessment channel comprises: Performing edge detection and anchor frame marking on each defect region in the visual auxotroph image set to obtain an instant noodle auxotroph region set; Performing multidimensional feature extraction based on the instant noodle nutrition defect region set to obtain an instant noodle multidimensional defect feature vector set; And performing feature recognition and defect evaluation on the multi-dimensional defect feature vector set of the instant noodles by using a YOLO algorithm network to obtain a visual nutritional defect evaluation channel.
7. The method for rapid detection of nutritional defects in non-fried instant noodles of claim 6, wherein said obtaining a visual nutritional defect assessment channel comprises: performing feature classification and identification on the multi-dimensional defect feature vector set of the instant noodles according to an instant noodle nutritional defect tag library to obtain a nutritional defect type feature set; performing defect area integration on the multi-dimensional defect feature vector set of the instant noodles to obtain an auxotroph area feature set; Performing defect evaluation training on the auxotroph type feature set and the auxotroph area feature set by using a YOLO algorithm network, and constructing the visual auxotroph evaluation channel.
8. The method for rapid detection of nutritional defects in non-fried instant noodles of claim 7, wherein said constructing said visual nutritional defect assessment channel comprises: Performing influence degree evaluation on each defect type in the nutritional defect type feature set, and determining a defect type influence weight factor; Weighting the nutritional defect type feature set and the nutritional defect area feature set according to the defect type influence weight factors to obtain a nutritional defect degree feature set; Performing defect evaluation training on the nutritional defect degree characteristic set based on the visual nutritional defect image set by using a YOLO algorithm network, and constructing the visual nutritional defect evaluation channel.
9. The method for rapid detection of nutritional defects in non-fried instant noodles of claim 1, wherein establishing a multi-modal pathway for nutritional defect detection comprises: verifying and acquiring decision coefficient information of the nutritional ingredient defect prediction channel and the visual nutritional defect assessment channel; and carrying out parallel weighted combination on the nutritional ingredient defect prediction channel and the visual nutritional ingredient defect evaluation channel based on the decision coefficient information, and establishing the nutritional ingredient defect detection multi-mode channel.
10. A rapid detection system for nutritional deficiencies in non-fried instant noodles, the system comprising: the three-dimensional spectrum database construction module is used for synchronously scanning the multi-type non-fried instant noodles through a dual-band light source to establish a nutritional ingredient defect three-dimensional spectrum database, wherein the dual-band light source comprises a near infrared spectrum and a Raman spectrum; The nutritional ingredient prediction fitting module is used for performing nutritional ingredient prediction fitting based on the nutritional ingredient defect three-dimensional spectrum database to construct a nutritional ingredient defect prediction channel; The visual nutrition defect evaluation module is used for acquiring visual nutrition defect image sets of the multi-type non-fried instant noodles, and performing feature recognition and defect evaluation on the visual nutrition defect image sets to obtain a visual nutrition defect evaluation channel; The multi-modal channel establishing module is used for combining the nutritional ingredient defect prediction channel and the visual nutritional defect assessment channel to establish a nutritional ingredient defect detection multi-modal channel; And the defect rapid detection module is used for rapidly detecting defects of the spectrum data and the image data of the target non-fried instant noodles based on the nutritional ingredient defect detection multi-mode channel and determining a nutritional ingredient defect detection result of the target instant noodles.

Description

Method and system for rapidly detecting nutritional ingredient defects of non-fried instant noodles Technical Field The invention relates to the technical field of food detection, in particular to a method and a system for rapidly detecting nutritional ingredient defects of non-fried instant noodles. Background The non-fried instant noodles are used as health instant food, and the balance and stability of the nutritional ingredients of the instant noodles are required to be strictly controlled in the production process. However, due to factors such as raw material difference, fluctuation of drying process, uneven distribution of seasonings and the like, abnormal key nutrition indexes such as protein, fat, carbohydrate and the like often appear in the finished product, and the quality and the compliance of nutrition labels are affected. At present, the industry mainly relies on laboratory chemical detection or artificial image recognition to carry out spot check on products, and the problems of long detection period, complex operation, sample damage, strong subjectivity of judging results and the like exist in the modes, so that the requirements of quick, batch and accurate identification on nutritional defects are difficult to meet. Disclosure of Invention The application provides a rapid detection method and a rapid detection system for nutritional ingredient defects of non-fried instant noodles, which are used for solving the technical problems that the prior art relies on laboratory physicochemical analysis or single spectrum technology to carry out spot check on products, and the detection period is long and the accuracy is low. The application provides a method for rapidly detecting nutritional ingredient defects of non-fried instant noodles, which comprises the steps of synchronously scanning multiple types of non-fried instant noodles through a dual-band light source, establishing a nutritional ingredient defect three-dimensional spectrum database, wherein the dual-band light source comprises a near infrared spectrum and a Raman spectrum, performing nutritional ingredient prediction fitting on the basis of the nutritional ingredient defect three-dimensional spectrum database, constructing a nutritional ingredient defect prediction channel, acquiring a visual nutritional ingredient image set of the multiple types of non-fried instant noodles, performing feature recognition and defect assessment on the visual nutritional ingredient image set to obtain a visual nutritional ingredient assessment channel, combining the nutritional ingredient defect prediction channel and the visual nutritional ingredient assessment channel, establishing a nutritional ingredient defect detection multi-mode channel, rapidly detecting defects on spectral data and image data of target non-fried instant noodles on the basis of the nutritional ingredient defect detection multi-mode channel, and determining a target instant noodle nutritional ingredient defect detection result. The application provides a rapid detection system for nutritional ingredient defects of non-fried instant noodles, which comprises a three-dimensional spectral database construction module, a multi-mode channel establishment module and a rapid defect detection module, wherein the three-dimensional spectral database construction module is used for synchronously scanning multi-type non-fried instant noodles through a dual-band light source to establish a nutritional ingredient defect three-dimensional spectral database, the dual-band light source comprises a near infrared spectrum and a Raman spectrum, the nutritional ingredient prediction fitting module is used for performing nutritional ingredient prediction fitting based on the nutritional ingredient defect three-dimensional spectral database to establish a nutritional ingredient defect prediction channel, the visual nutritional ingredient assessment module is used for acquiring a visual nutritional ingredient image set of the multi-type non-fried instant noodles, the visual nutritional ingredient image set is subjected to feature recognition and defect assessment to obtain a visual nutritional ingredient assessment channel, the multi-mode channel establishment module is used for combining the nutritional ingredient defect prediction channel and the visual nutritional ingredient assessment channel to establish a nutritional ingredient defect detection mode channel, and the rapid defect detection module is used for determining the defect detection result of the nutritional ingredient defect detection module based on the nutritional ingredient three-dimensional spectral database. One or more technical schemes provided by the application have at least the following technical effects or advantages: The application provides a rapid detection method and a rapid detection system for nutritional ingredient defects of non-fried instant noodles, which relate to the technical field of food detection, synchronously