CN201051074Y - Multi-spectrum flesh verdure artificial intelligence measurement system - Google Patents

Multi-spectrum flesh verdure artificial intelligence measurement system Download PDF

Info

Publication number
CN201051074Y
CN201051074Y CNU2007201094586U CN200720109458U CN201051074Y CN 201051074 Y CN201051074 Y CN 201051074Y CN U2007201094586 U CNU2007201094586 U CN U2007201094586U CN 200720109458 U CN200720109458 U CN 200720109458U CN 201051074 Y CN201051074 Y CN 201051074Y
Authority
CN
China
Prior art keywords
meat
artificial intelligence
freshness
image
spectrum
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CNU2007201094586U
Other languages
Chinese (zh)
Inventor
鲍一丹
童晓星
何勇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CNU2007201094586U priority Critical patent/CN201051074Y/en
Application granted granted Critical
Publication of CN201051074Y publication Critical patent/CN201051074Y/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Abstract

The utility model discloses a multispectral meat freshness artificial intelligence measurement system, and is characterized in that an adjustable parameter light source sends out special beams on a meat-placing platform, and a 3 CCD multispectral camera receives the meat reflection light and sends the signals to an image collection card and a computer process system. Different meat varieties are selected for various pre-processes according to meat database selection in a computer, wave ranges with various characteristics are extracted, and then a pixel set is selected as a research object of which the image is further studied, the fresh meat, secondary fresh meat and putrid meat are identified in three different modes through artificial intelligence judgment method, and the results are finally output and displayed on the computer. The utility model utilizes the combination of techniques such as machine vision, image process, artificial intelligence and the like to evaluate meat freshness rapidly, nondestructively and accurately.

Description

The multiple spectrum meat freshness artificial intelligence measuring system
Technical field
The utility model relates to and utilizes optical instrument to come analysis of material, especially relates to a kind of multiple spectrum meat freshness artificial intelligence measuring system.
Background technology
The evaluation metrics of freshness of meat is a lot, and traditional evaluating method should carry out Comprehensive Assessment from each side such as the outward appearance of meat itself, smell, chemical constitutions, requires very high to the reviewer.Existing detection technique mainly contains following several: the 1) test of the total alkali nitrogen-TVB-N of volatility.No matter be to adopt Kjeldahl, still other decide the nitrogen method, and all unavoidable operation is loaded down with trivial details, and the test duration is long, and is not Non-Destructive Testing, is difficult to satisfy current requirement for the detection of batch samples quick nondestructive.2) detect surperficial yellowish pink and reflect freshness.This method adopts the illumination of certain wavelength to the reference substance surface, the light of returning by the receiver detection of reflected, the illumination of the same wavelength of light emitted is to meat surface then, the light that the receiver detection of reflected is returned, calculate the emission absorbance, and with this foundation of value employing artificial neural network and the corresponding mathematical model of yellowish pink standard, it is yellowish pink to be used for detection, and then the supposition freshness.But the wave band of the light that this method detects is too single, and the light-wave information amount that acceptance obtains is less, and the data that obtain in this method are directly perceived inadequately and lack the post-processed in early stage, are subject to external interference.
Visitor in the invention
The purpose of this utility model is to provide a kind of multiple spectrum meat freshness artificial intelligence measuring system, adopts combining of technology such as machine vision, Flame Image Process and artificial intelligence, fast, can't harm, evaluate exactly the freshness of meat.
The technical scheme that its technical matters that solves the utility model adopts is:
Comprise the customized parameter light source, meat placement platform, the multispectral video camera of 3CCD; Image pick-up card and computer processing system and database; Customized parameter light emitted particular beam is shone on the meat placement platform, and the multispectral video camera of 3CCD receives the meat reflected light, and signal is transferred to image pick-up card, is transferred to computer processing system by capture card and handles.
The principle of this method is:
1) by the irradiation of fixed light source, the multispectral video camera of 3CCD is accepted the meat reflective information;
2) reflective information is transferred to computing machine by image pick-up card, obtains 3 original width of cloth images, is respectively the monochrome image of 550nm, 650nm, three waveband channels of 800nm;
3) by wavelet de-noising, reduce picture noise, because the background and the meat of near infrared monochrome image differ greatly, can make meat and background separation, and finish rim detection, obtain the image information of independent meat;
4) compressed image, the pixel collection of the image co-registration Cheng Xin of original 10*10 pixel, the mean flow rate, brightness variance, the brightness that select this pixel collection are differential, as the proper vector of this pixel collection;
5) choose in advance by detecting several samples that the total alkali nitrogen-TVB-N of volatility determines freshness of meat, with these different freshnesss as output, by above 1) to 4) proper vector extracted of step is as input, sets up the database model of different classes of meat by the artificial intelligence approach of support vector machine or artificial neural network;
6) take the multispectral image of meat to be measured, by above 1) to 4) proper vector extracted of step and as input, with 5) model set up of step is the differentiation process, judges the fresh classification of meat of each pixel collection;
7) comprehensively judge with the freshness judged result proportion of each pixel collection, take not have weight here, determine the freshness of monoblock meat, and the result is exported with maximum number pixel collection types.
The beneficial effect that the utlity model has is:
1. powerful, can realize quick, accurate, nondestructive diagnosis to freshness of meat.
2. simple in structure, entire measuring device only is made up of a 3CCD multi-optical spectrum imaging system, a computing machine.
3. easy to use, as long as each building block in the measurement mechanism is coupled together as requested, just can measure.
4. has good economic benefit.Traditional measurement means need expend great amount of manpower and material resources at aspects such as sampling, mensuration, data analyses, and weak effect.This measurement mechanism can be measured the multispectral image information of meat quickly and accurately because of simple in structure, easy to make, analyzes the freshness that obtains meat, thereby realizes real-time, Non-Destructive Testing freshness of meat.
Description of drawings
Fig. 1 is the utility model system chart.
Fig. 2 is a workflow diagram of the present utility model.
Embodiment
As shown in Figure 1, the utility model comprises the customized parameter light source, meat placement platform, the multispectral video camera of 3CCD, image pick-up card and computer processing system and database.Customized parameter light emitted particular beam is shone on the meat placement platform, and the multispectral video camera of 3CCD receives the meat reflected light, and signal is transferred to image pick-up card, is transferred to computer processing system by capture card and handles.
Described multispectral video camera is the multispectral camera system of the MS of Duncan company series 3CCD.
Described image pick-up card is National Instrument PCI of company 1424 or PCI 1428.
The multispectral camera system of MS series of Duncan company will be taken in image after filtration, be separated into green (550nm) in real time, red (650nm), the monochrome image of three waveband channels of near infrared (800nm), the image pick-up card by PCI 1424 or 1428 (National Instrument company) is connected to computing machine.
As shown in Figure 1, concrete workflow of the present utility model is as follows:
1. unify the meat of light source irradiation on platform, multispectral video camera clap reflected light, collect by image pick-up card, import computing machine into, obtain independently achromatic sheet of three pairs, the gained image by wavelet de-noising, reduces picture noise through a series of anticipating.Because the background and the meat of near infrared monochrome image differ greatly, can make meat and background separation, be usually used in the isolation of research object and background, and finish rim detection, obtain the image information of independent meat, near-infrared band wherein has bigger discrimination to rim detection.
2. (g, r's vector of each pixel on the meat that computing machine obtains and a three-dimensional NIR) are associated, and wherein g, r, NIR are respectively the monochromatic brightness of green (550nm), red (650nm), three passages of near infrared (800nm).By further extraction to image information, carry out compression of images, every 10*10 pixel is formed a pixel collection, select the mean flow rate of this pixel collection, brightness variance, the differential parameter of brightness, because it is 3*3=9 that independently image of three width of cloth, the proper vector of gained are arranged, as the proper vector of this pixel collection.
3. the demarcation of freshness of meat.When estimating the various method of inspection, most scholars think that the total alkali nitrogen-TVB-N of volatility is in the metamorphic process of meat, can reflect the variation of freshness of meat regularly, this index has been included into national standard, is most important physical and chemical index during present meat product freshness detects.By standard GB 2722, the freshness of meat grade such as the table 1 of general volatile alkali nitrogen content correspondence.
Data in the database are used for setting up master pattern, data in the database are under certain certain light intensity, the image of a certain concrete meat that shooting obtains, and measured total alkali nitrogen value, and data are divided into green meat, inferior fresh meat, corrupt meat according to different freshness of meat standards.More than 9 proper vectors obtaining as input, the grade of green meat, inferior fresh meat, corrupt meat is as output, utilization Support Vector Machine (support vector machine) sets up the artificial intelligence model of known sample as the theory of artificial intelligence.More than for setting up the data base procedure of master pattern.Sample to be judged then needn't be judged freshness by measuring total alkali nitrogen value, take by above-mentioned multispectral video camera, by image pick-up card input computing machine, and obtain after the proper vector of pixel collection, the proper vector of this pixel collection as input, output terminal is the result of artificial intelligent decision, model is the existing model of a certain concrete meat in the database, sort out judgement, judge the freshness attribute of each meat pixel collection, the shared quantity of freshness judged result with each pixel collection comprehensively judges, takes not have weight here, determines the freshness of monoblock meat with maximum number pixel collection types. and the result exported..
The corresponding relation of table 1 freshness of meat grade and TVB-N content

Claims (3)

1. a multiple spectrum meat freshness artificial intelligence measuring system is characterized in that: comprise the customized parameter light source, meat placement platform, the multispectral video camera of 3CCD; Image pick-up card and computer processing system and database; Customized parameter light emitted particular beam is shone on the meat placement platform, and the multispectral video camera of 3CCD receives the meat reflected light, and signal is transferred to image pick-up card, is transferred to computer processing system by capture card and handles.
2. a kind of multiple spectrum meat freshness artificial intelligence measuring system according to claim 1 is characterized in that: described multispectral video camera is the multispectral camera system of the MS of Duncan company series 3CCD.
3. a kind of multiple spectrum meat freshness artificial intelligence measuring system according to claim 1 is characterized in that: described image pick-up card is National Instrument PCI of company 1424 or PCI1428.
CNU2007201094586U 2007-05-22 2007-05-22 Multi-spectrum flesh verdure artificial intelligence measurement system Expired - Fee Related CN201051074Y (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNU2007201094586U CN201051074Y (en) 2007-05-22 2007-05-22 Multi-spectrum flesh verdure artificial intelligence measurement system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNU2007201094586U CN201051074Y (en) 2007-05-22 2007-05-22 Multi-spectrum flesh verdure artificial intelligence measurement system

Publications (1)

Publication Number Publication Date
CN201051074Y true CN201051074Y (en) 2008-04-23

Family

ID=39337565

Family Applications (1)

Application Number Title Priority Date Filing Date
CNU2007201094586U Expired - Fee Related CN201051074Y (en) 2007-05-22 2007-05-22 Multi-spectrum flesh verdure artificial intelligence measurement system

Country Status (1)

Country Link
CN (1) CN201051074Y (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104049068A (en) * 2014-06-06 2014-09-17 中国肉类食品综合研究中心 Nondestructive determining device and determining method for freshness of fresh livestock meat
CN109313126A (en) * 2017-05-17 2019-02-05 Eba日本株式会社 Information search system and method and information search program

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104049068A (en) * 2014-06-06 2014-09-17 中国肉类食品综合研究中心 Nondestructive determining device and determining method for freshness of fresh livestock meat
CN104049068B (en) * 2014-06-06 2015-11-18 中国肉类食品综合研究中心 The non-destructive determination device of fresh poultry meat freshness and assay method
CN109313126A (en) * 2017-05-17 2019-02-05 Eba日本株式会社 Information search system and method and information search program

Similar Documents

Publication Publication Date Title
CN100480680C (en) Multiple spectrum meat freshness artificial intelligence measurement method and system
CN100357725C (en) Method and device for rapidly detecting tenderness of beef utilizing near infrared technology
CN103257118A (en) Fish tenderness hyperspectral detection method based on characteristic wave band
CN101692052B (en) Hyperspectrum image technique-based method and hyperspectrum image technique-based device for identifying true and false famous tea
CN101419166A (en) Tea quality nondestructive detecting method and device based on near-infrared spectrum and machine vision technology
CN101308086B (en) Fruit internal quality on-line checking apparatus based on near infrared spectra technology
CN102147402A (en) Machine vision technology based method for rapidly detecting egg freshness
CN101144780A (en) Pork freshness intelligent detection device
CN101881726A (en) Nondestructive detection method for comprehensive character living bodies of plant seedlings
CN101832941A (en) Fruit quality evaluation device based on multispectral image
CN101059452A (en) Fruit quality damage-free detection method and system based on multiple spectral imaging technique
CN104359838A (en) Non-destructive detection method for hardness of pear
Cluff et al. Optical scattering with hyperspectral imaging to classify longissimus dorsi muscle based on beef tenderness using multivariate modeling
CN104256882B (en) Based on reconstituted tobacco ratio measuring method in the pipe tobacco of computer vision
CN105092579A (en) Mango quality non-destructive testing device
CN105466921A (en) Simultaneous detection method of many samples
CN104198325B (en) Stem ratio measuring method in pipe tobacco based on computer vision
CN201041553Y (en) Fruit quality non-damage detection system base on multi-spectrum imaging technology
CN105548124A (en) Method and device for detecting citrus canker
CN108956604A (en) A method of Eriocheir sinensis quality is identified based on hyper-spectral image technique
CN101021488A (en) Method and apparatus for detecting rice damage
CN104198324A (en) Computer vision-based method for measuring proportion of cut leaves in cut tobacco
CN109100323B (en) Nondestructive quantitative evaluation method for transmission spectrum of apple water core disease
CN201051074Y (en) Multi-spectrum flesh verdure artificial intelligence measurement system
Ismail et al. Hue optical properties to model oil palm fresh fruit bunches maturity index

Legal Events

Date Code Title Description
C14 Grant of patent or utility model
GR01 Patent grant
C17 Cessation of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20080423

Termination date: 20100522