CN102323267A - System and method used for rapidly evaluating freshness of raw meat products - Google Patents

Abstract

The invention belongs to the technical field of meat product quality safety testing, and provides a system and a method used for rapidly evaluating the freshness of raw meat products. The system is composed of an imaging spectrometer, a CCD digital camera, a tungsten halogen lamp direct current point light source, a light source stabilizing device, a camera controller, a computer and an image collecting chip. The method provided by the invention comprises steps that: the CCD digital camera, the imaging spectrometer and the camera controller are used for collecting hyper-spectral images of the surfaces of pork; evaluation and grading are carried out according to measured physical and chemical parameters; the hyper-spectral images are analyzed by using mathematical techniques, and a forecasting model used for evaluating pork freshness is established. With the system and the method provided by the invention, rapid and lossless detections can be carried out upon raw meat freshness parameters such as volatile basic nitrogen, pH value and color.

Description

The system and method for the fresh meat freshness of a kind of quick evaluation

Technical field

The invention belongs to meat quality safety detection technology field, be specifically related to the system and method for the fresh meat freshness of a kind of quick evaluation.

Background technology

Fresh meat is the main source that constitutes China's meat food; Along with improving constantly of living standard, consumption demand growing, the relative variation of diet structure and to the growing interest of health; People become more diversified the demand of fresh meat, and the meat quality safe requirement is also improved constantly.Because pork itself is rich in nutriment and moisture, very easily corrupt, add and butcher the good sanitary condition of shortage in processing and the sales process, make the freshness situation of meat become intricate along with the variation of standing time during sale.The meat freshness is meant the freshness of meat, is the integrated status of standard local flavor, flavour, color and luster, quality, mouthfeel and the qualified hygienic standard of microorganism of refering in particular to a certain type animal food.Fresh meat freshness is to weigh the objective standard whether pork meets edible demand, can synthetically reflect the degree of reliability of trophism, security and the hobby property of meat.The quality of fresh meat matter; Directly have influence on people's quality of life and healthy and safe; Also affect simultaneously the development of whole meat industry; It is the big event of meat hygiene and various fresh-keeping, process technology research that fresh meat freshness detects, and therefore, sets up the freshness of meat experimental technique that quick nondestructive can practicality and has important practical significance and practical value.

Conventional at present pork freshness detection method has organoleptic detection, physics and chemistry to detect and microorganism detection, and detection efficiency is low, required time long, product destroys problems such as big but the above-mentioned method of inspection all exists.Modern meat quality safety detection technology require towards fast, harmless, accurately, can practical direction develop.Wherein, the high light spectrum image-forming technology is a kind of not damaged detection technique, and collection sensor, precision optics machinery, Detection of Weak Signals, computing machine and the information processing technology are one.With respect to other spectral techniques, the sample message that the high light spectrum image-forming technology can obtain is more comprehensive.The image information that the incident light that pointolite in the detection system sends obtains in the pork internal divergence can either characterize the chemical property and the physical property of internal composition, also can reflect the sample external feature comprehensively.

Summary of the invention

The object of the present invention is to provide the system of the fresh meat freshness of a kind of quick evaluation.

The present invention also aims to provide the method for the fresh meat freshness of a kind of quick evaluation.

The system of the fresh meat freshness of a kind of quick evaluation, this system is made up of CCD digital camera 1, imaging spectrometer 2, halogen tungsten lamp dc point light source 4, light stability device 5, phase machine controller 6, computing machine 7 and image pick-up card 8; CCD digital camera 1 is connected with imaging spectrometer 2, and halogen tungsten lamp dc point light source 4 is connected with light stability device 5, and phase machine controller 6 is connected with image pick-up card 8 with computing machine 7 successively; The spectral range of this system acquisition is 400-1100nm, and spectrum rate respectively is 2.8nm.

The method of the fresh meat freshness of a kind of quick evaluation comprises following steps:

A, utilize CCD digital camera 1, imaging spectrometer 2 and machine controller 6 mutually, gather the high spectrum image on pork surface;

B, to representing index VBN, pH value, the color of pork freshness parameter, carry out physical and chemical determination, as reference point; According to the physicochemical data that experiment obtains, be leading indicator with the VBN, formulate the evaluation criteria of pork freshness, be divided into one-level freshness, secondary freshness and rotten meat;

C, the image that step a is obtained carry out atlas analysis; Analyze the diffuse information that spectrum forms on the pork surface; Adopt the spatial diffusion curve of each wavelength of Lorentz distribution function match pork sample, with the Lorentz VBN, pH value and the color parameter that obtain as the high spectrum characteristics spectral information of pork;

D, the characteristic spectrum information of utilizing step c to obtain, adopt stepwise regression method obtain can representative sample freshness information optimal wavelength and corresponding Lorentz parameter;

The forecast model of e, foundation evaluation pork freshness, as follows:

$y=k_1{x}_1+k_2{x}_2+k_3\underset{i=1}{\overset{3}{\mathrm{\Σ}}}{j}_i{z}_i$

Wherein, y represents the forecast model of meat freshness, and x1 represents the forecast model of VBN, and x2 represents the forecast model of pH value, zi representative color parameter L ^*, a ^*, b ^*Forecast model, color adopts the CIE_Lab color space, the CIE_Lab color space is with L ^*The lightness of value representation color, a ^*The green red value of value representation color, b ^*The blue yellow value of value representation color, k1 is the weighting coefficient of VBN forecast model, and k2 is the weighting coefficient of pH value prediction model, and k3 is the weighting coefficient of color prediction model, ji representative color parameter L ^*, a ^*, b ^*Weighting coefficient in color prediction model.

Beneficial effect of the present invention: the present invention can be to the detection of freshness index quick nondestructives such as fresh meat freshness parameter VBN, pH value, color.

Description of drawings

Fig. 1 is an employed high light spectrum image-forming detection system synoptic diagram in the embodiment of the invention.

Among the figure, 1-CCD digital camera, 2-imaging spectrometer, 3-pork sample, 4-halogen tungsten lamp dc point light source, 5-light stability device, 6-phase machine controller, 7-computing machine are formed, the 8-image pick-up card.

Fig. 2 is the pork high spectrum image that collects in the embodiment of the invention.

Fig. 3 is the spatial diffusion curve of pork sample under the embodiment of the invention different wave length.

The related coefficient of each wavelength Lorentzian match of Figure 44 00-1100nm.

Embodiment

Below in conjunction with accompanying drawing and specific embodiment the present invention is further specified.

Embodiment 1

The meat appearance of present embodiment is chosen the longissimus dorsi muscle meat of cold fresh pork.Meat appearance through acid discharge in 24 hours, is together downcut together with connective tissue and fat on every side during sampling after butchering from The Big Red Gate meat product company limited.Pork is evenly cut into the thick cube meat of 2.5cm; Meat appearance wrapped with fresh-keeping plastic bag to be placed in 4 ℃ of refrigerator-freezers storage to be measured; In 12 days, accomplish hyper-spectral data gathering and freshness index physics and chemistry pH-value determination pH to the pork sample.The storage environment of meat appearance and experimental implementation environment are consistent in the implementation process of present embodiment, promptly keep identical temperature and humidity.

The system of the fresh meat freshness of employed quick evaluation is as shown in Figure 1 in the method for present embodiment.This system is by CCD digital camera (Sencicam QE; Germany) 1, imaging spectrometer (ImSpector V10E; Spectral Imaging Ltd.; Finland) 2, halogen tungsten lamp dc point light source (Oriel Instruments, USA) 4, light stability device 5, phase machine controller 6, computing machine 7 and image pick-up card 8 form; The spectral range of this system acquisition is 400-1100nm, and spectrum rate respectively is 2.8nm, and the resolution of CCD digital camera is 1376 * 1040; Light source is output as the pointolite of diameter 5mm.

As shown in Figure 1, utilize the step of the fresh meat freshness of quick evaluation of above-mentioned sample and system implementation following:

A, gather the high spectrum image of pork sample: every at a distance from 6 hours, from refrigerator, take out 2 samples, remove the sample external packing, in air, expose 30 minutes, surface moisture is evaporated after, begin to gather high spectrum image.Place pork sample 3 on objective table, regulate sample position simultaneously, the position that makes pork sample 3 place CCD digital camera 2 field range interscan lines to set, and guarantee that sample surface is identical with the distance that timing is provided with.Open phase machine controller 6, observe greyish white dynamic real-time image, observe whether tangible diffusion phenomena are arranged, other abnormal occurrencies such as image disruption spectrum whether occur.Gather dark current image, dark current is gathered button in the software of click phase machine controller 6, and resulting dark current image is preserved.Each sample is gathered 5 times at parallel diverse location, and image cuts dark current automatically with 16 bit byte tif format during storage.The high spectrum image of the pork sample of gathering is as shown in Figure 2.

Pork sample freshness physics and chemistry reference point after b, mensuration are gathered through high spectrum image: the pork sample is carrying out the later physics and chemistry value detection of at once same sample being carried out meat freshness parameter index VBN, pH value, color of high spectral scan.Carry out color (L earlier ^*, a ^*, b ^*) mensuration, adopt portable precision standard color difference meter (HP-200, ShangHai HanPu opto-electrical Science Co., Ltd), each sample is different muscle position horizontal survey 6 times, mean value is as the ultimate criterion reference point of this sample.The mensuration of pH value adopts GB/T 9695.5 2008 " meat and meat products pH measure ", adopts FE20K laboratory pH meter to measure.VBN takes the semimicro diffusion method of GB GB/T5009.44-2003 to measure, and each meat appearance is surveyed three times altogether and averaged.

Formulate the rating of pork freshness.With the VBN is leading indicator, in conjunction with resulting freshness physicochemical data in the embodiment of the invention, pork freshness is divided into one-level freshness, secondary freshness, rotten meat Three Estate, as shown in table 1 below:

The evaluation criteria of table 1 pork freshness grade

C, the high spectral space diffusion profile of match pork are resolved the high spectrum image of obtaining pork sample, obtain the spatial diffusion information of sample at the different wave length place.Utilize spatial diffusion curve (Fig. 3) with each wavelength of Lorentz distribution function match pork sample:

$I_{w_i}=a_{\mathrm{wi}}+\frac{c_{w_i}}{1+{(x/b_{w_i})}^2}$

Wherein, any reflection strength (CCD gray-scale value) of any on the I-scattering curve; The distance of x-diffusion profile decentering point, mm; The asymptotic value of a-diffusion profile; The b-diffusion profile is at the half-wave bandwidth at 1/2 place of peak value, mm; The c-diffusion profile is at the peak value at sweep trace central point x=0 place; Subscript wi-scope is the number of wavelengths of 400～1100nm, i=1, and 2,3 ..., N, N are total number of wavelengths.

Utilize the fitting coefficient of Lorentzian match pork sample space diffusion profile as shown in Figure 4; Can see that fitting correlation coefficient is higher in the 470-1000nm scope; More than 0.99, the effective spectroscopic data of the spectroscopic data of therefore selecting this wave band during as modeling.

D, obtain can representative sample freshness information optimal wavelength and corresponding Lorentz parameter: sample is divided into two groups at random, and calibration set accounts for 3/4 of population sample, and forecast set accounts for 1/4 of population sample.To pass through Lorentz three parameter a, b, the c that the Lorentzian match obtains and make up, and can reflect the spatial diffusion information of high spectrum in the pork sample more comprehensively, thereby reflect the quality information of pork sample more accurately through spectral information.Use Lorentz three parameter combinations [abc] to set up the PLSR forecast model.If three parameters matrix separately of Lorentzian is a (n * m), b (n * m), c (n * m); Wherein n is a sample number, and m is effective wave number, and then the matrix of Lorentz three parameter combinations [abc] is a n * 3m combinatorial matrix; Wherein, Preceding m row are Lorentz parameter a, middle m row formula Lorentz parameter b, back m row formula Lorentz parameter c.Represent the characteristic spectrum information of pork sample to participate in the foundation of pork freshness parametric prediction model Lorentz three parameter combinations [abc].The detailed process of setting up each parametric prediction model of freshness is: utilize the Return Law progressively to choose the characteristic wavelength of three parameter indexs of pork freshness in the 470-1000nm scope respectively, utilize the data of corresponding Lorentz three parameter combinations in each selected characteristic wavelength place to set up the multiple linear regression forecast model.

E, set up Comprehensive Assessment pork freshness forecast model: with the pork freshness parameter VBN, pH value, the color (L that obtain in the above-mentioned steps ^*, a ^*, b ^*) utilize weighted method will represent three parameter models of pork freshness comprehensively to arrive together, set up the forecast model that is used to evaluate pork freshness.

$y=k_1{x}_1+k_2{x}_2+k_3\underset{i=1}{\overset{3}{\mathrm{\Σ}}}{j}_i{z}_i$

Wherein, y represents the forecast model of meat freshness, represents x1 to represent the forecast model of VBN (TVB-N), and x2 represents the forecast model of pH value, zi representative color parameter L ^*, a ^*, b ^*Forecast model, color adopts the CIE_Lab color space, the CIE_Lab color space is with L ^*The lightness of value representation color, a ^*The green red value of value representation color, b ^*The blue yellow value of value representation color, k1 is the weighting coefficient of TVB-N forecast model, and k2 is the weighting coefficient of pH value prediction model, and k3 is the weighting coefficient of color prediction model, ji representative color parameter L ^*, a ^*, b ^*Weighting coefficient in color prediction model.

Claims (2)

1. system that estimates fast fresh meat freshness; It is characterized in that this system is made up of CCD digital camera (1), imaging spectrometer (2), halogen tungsten lamp dc point light source (4), light stability device (5), phase machine controller (6), computing machine (7) and image pick-up card (8); CCD digital camera (1) is connected with imaging spectrometer (2), and halogen tungsten lamp dc point light source (4) is connected with light stability device (5), and phase machine controller (6) is connected with image pick-up card (8) with computing machine (7) successively; The spectral range of this system acquisition is 400-1100nm, and spectrum rate respectively is 2.8nm.

2. a method of estimating fresh meat freshness fast is characterized in that, comprises following steps:

A, utilize CCD digital camera 1, imaging spectrometer 2 and machine controller 6 mutually, gather the high spectrum image on pork surface;

B, to representing index VBN, pH value, the color of pork freshness parameter, carry out physical and chemical determination, as reference point; According to the physicochemical data that experiment obtains, be leading indicator with the VBN, formulate the evaluation criteria of pork freshness, be divided into one-level freshness, secondary freshness and rotten meat;

C, the image that step a is obtained carry out atlas analysis; Analyze the diffuse information that spectrum forms on the pork surface; Adopt the spatial diffusion curve of each wavelength of Lorentz distribution function match pork sample, with the Lorentz VBN, pH value and the color parameter that obtain as the high spectrum characteristics spectral information of pork;

D, the characteristic spectrum information of utilizing step c to obtain, adopt stepwise regression method obtain can representative sample freshness information optimal wavelength and corresponding Lorentz parameter;

The forecast model of e, foundation evaluation pork freshness, as follows:

$y=k_1{x}_1+k_2{x}_2+k_3\underset{i=1}{\overset{3}{\mathrm{\Σ}}}{j}_i{z}_i$

Wherein, y represents the forecast model of meat freshness, and x1 represents the forecast model of VBN, and x2 represents the forecast model of pH value, zi representative color parameter L ^*, a ^*, b ^*Forecast model, color adopts the CIE_Lab color space, the CIE_Lab color space is with L ^*The lightness of value representation color, a ^*The green red value of value representation color, b ^*The blue yellow value of value representation color, k1 is the weighting coefficient of VBN forecast model, and k2 is the weighting coefficient of pH value prediction model, and k3 is the weighting coefficient of color prediction model, ji representative color parameter L ^*, a ^*, b ^*Weighting coefficient in color prediction model.

Images