CN105548029A - Meat product freshness detection method based on spectral imaging technology - Google Patents
Meat product freshness detection method based on spectral imaging technology Download PDFInfo
- Publication number
- CN105548029A CN105548029A CN201510925619.8A CN201510925619A CN105548029A CN 105548029 A CN105548029 A CN 105548029A CN 201510925619 A CN201510925619 A CN 201510925619A CN 105548029 A CN105548029 A CN 105548029A
- Authority
- CN
- China
- Prior art keywords
- meat products
- freshness
- image
- sample
- area
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention relates to a meat product freshness detection method based on a spectral imaging technology; the detection method includes the steps: acquiring a hyperspectral image of a meat product sample; detecting the content of a freshness component corresponding to the meat product sample; extracting a meat product sample-corresponding interest region in the hyperspectral image, and extracting spectral characteristics and image texture characteristics of the meat product sample; according to the content of the freshness component of the meat product sample and the spectral characteristics and the image texture characteristics of the meat product sample-corresponding interest region, building a prediction relational model of the content of the freshness component of the meat product sample; through an acquired hyperspectral image of a to-be-tested meat product, adopting the pre-built prediction relational model to invert to obtain the content of the freshness component of the to-be-tested meat product. The method can realize objective and quantitative detection of the freshness of the meat product, can simplify the detection process of the freshness of the meat product, shortens the detection period, and reduces the detection cost.
Description
Technical field
The present invention relates to meat products detection field, particularly relate to a kind of meat products Noninvasive Measuring Method of Freshness based on spectral imaging technology.
Background technology
Meat product contains rich in protein, fat and mineral matter, again because it has the advantage such as delicious flavour, instant, and the dark favor by consumers in general.The problem such as corrupt deterioration degree and the authenticity of frame delivery date of meat product in storage is the focus that people pay close attention to always.Freshness is the direct reflection of meat quality, is also one of important measurement index of fresh meat and meat products quality simultaneously.Therefore, in order to clear and definite Freshness evaluation standard, judge the storage period that meat perish goes bad and the freshness of meat products is detected, meet consumer to the demand that meat products class quality is growing while, also to realization scientifically instruct meat products process and storage have important practical significance and using value.
Meat is putrid and deteriorated is a complicated change procedure, and kind and the degradation thereof of metamorphic grade and stigma of degeneracy and contaminating microorganisms are closely related.Tradition identifying meat freshness method mainly contains sensory evaluation, physical and chemical index detection and microbe colony detection etc.Wherein, sensory evaluation subjectivity quantizes by force, not easily, and the personnel suggestion of evaluation is difficult to unanimously; Physical and chemical index and microorganism detection exist that sample preparation is loaded down with trivial details, chemical reagent consumption amount is large, sense cycle is long and high in cost of production problem, cannot meet meat products class freshness is carried out fast, the requirement of accurate, stable detection.Along with spectral technique, developing rapidly of digital image processing techniques and artificial intelligence technology, near-infrared spectrum technique and computer vision technique are applied among the detection of freshness of meat by existing increasing researchist.Such as, use multispectral technology to carry out detecting etc. to fresh spiced beef freshness many index (total volatile basic nitrogen TVB-N, pH value, total number of bacteria, yellowish pink) in prior art simultaneously.But these class methods do not consider the spatial image characteristic information that meat sample is abundant, cause accuracy of detection can be subject to the impact of meat surface unevenness feature distribution.
Therefore, how a kind of method that accuracy of detection is high, speed also can detect meat products freshness soon is quantitatively provided to become one of technical matters being badly in need of at present solving for above-mentioned defect.
Summary of the invention
For solving the problems of the technologies described above, one aspect of the present invention proposes a kind of meat products Noninvasive Measuring Method of Freshness based on spectral imaging technology, and the method comprises:
Obtain the high spectrum image of the first meat products sample under the different default storage time;
Freshness component concentration corresponding under detecting described first meat products sample default storage time described in each;
Utilize image processing techniques to carry out Iamge Segmentation to the high spectrum image of the first meat products sample obtained, extract the area-of-interest that the first meat products sample in described high spectrum image is corresponding;
Extract spectral signature and the image texture characteristic of area-of-interest corresponding to described first meat products sample;
The freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic is built according to the spectral signature of the freshness component concentration of the described first meat products sample area-of-interest corresponding with described first meat products sample and image texture characteristic;
Obtain the high spectrum image of meat products to be measured, the high spectrum image of image processing techniques to the meat products to be measured obtained is utilized to carry out Iamge Segmentation, and extract the area-of-interest of the high spectrum image of described meat products to be measured, to extract spectral signature to be measured and the testing image textural characteristics of this area-of-interest;
According to spectral signature to be measured and testing image textural characteristics, the projected relationship model inversion set up in advance is adopted to obtain the freshness component concentration of described meat products to be measured.
Preferably, described freshness component is total volatile basic nitrogen TVB-N.
Preferably, the spectral signature of the area-of-interest that the described first meat products sample of described extraction is corresponding, comprising:
Utilize the algorithm RF that leapfrogs at random to extract and can characterize the characteristic wavelength of described first meat products sample freshness change in storage process as spectral signature.
Preferably, the described characteristic wavelength that can characterize the freshness change in storage process of described first meat products sample comprises 660nm, 720nm, 790nm, 820nm, 880nm, 910nm.
Preferably, the described high spectrum image of image processing techniques to the first meat products sample obtained that utilize carries out Iamge Segmentation, comprising:
According to described characteristic wavelength, to the smoothing denoising of area-of-interest corresponding to described first meat products sample, the process of mask gray processing, and utilizing the image partition method that Otsu ' s thresholding method combines with dilation erosion algorithm, the area-of-interest corresponding to the first meat products sample through smoothing denoising, the process of mask gray processing carries out background removal.
Preferably, the image texture characteristic of the area-of-interest that the described first meat products sample of described extraction is corresponding, comprising:
The feature extraction algorithm utilizing local binary pattern LBP and Tamura textural characteristics method to combine extracts the image texture characteristic of area-of-interest corresponding to described first meat products sample.
Preferably, spectral signature and the image texture characteristic of the described area-of-interest corresponding with described first meat products sample according to the freshness component concentration of described first meat products sample build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic, comprising:
Principal component analysis (PCA) PCA method is utilized to merge, the spectral signature of area-of-interest corresponding for described first meat products sample and image texture characteristic according to this fusion results using the input variable of PCA score as modeling;
According to described input variable, artificial fish-swarm extreme learning machine AF-ELM algorithm is utilized to build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic.
Preferably, after the spectral signature of the described area-of-interest corresponding with described first meat products sample according to the freshness component concentration of described first meat products sample and image texture characteristic build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic, the method also comprises:
Evaluate accuracy and the robustness of described projected relationship model.
Preferably, the described image processing techniques that utilizes carries out Iamge Segmentation to the high spectrum image of the first meat products sample obtained, and before extracting area-of-interest corresponding to the first meat products sample in described high spectrum image, the method also comprises:
Black and white correction is carried out to the high spectrum image of the first meat products sample obtained.
Preferably, the described image processing techniques that utilizes carries out Iamge Segmentation to the high spectrum image of the first meat products sample obtained, and before extracting area-of-interest corresponding to the first meat products sample in described high spectrum image, the method also comprises:
Gaussian filtering process and color enhancement conversion process are carried out to the high spectrum image of the first meat products sample corrected through black and white.
Meat products Noninvasive Measuring Method of Freshness based on spectral imaging technology of the present invention has the advantage that accuracy of detection is high, detection speed is fast, can realize objective, meat products freshness detection quantitatively; Meanwhile, can realize effectively simplifying meat products freshness testing process, shorten sense cycle, and the problem that chemical reagent consumption amount is large, testing cost is high can be avoided.
Accompanying drawing explanation
In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.
Fig. 1 shows the meat products Noninvasive Measuring Method of Freshness process flow diagram based on spectral imaging technology of one embodiment of the invention;
Fig. 2 shows the structural representation of the high spectrum image acquisition system of one embodiment of the invention;
Fig. 3 shows the meat products freshness quantitative forecast relational model construction method process flow diagram of one embodiment of the invention;
Fig. 4-1 to Fig. 4-3 shows the spiced beef gray-scale map corresponding under 660nm, 720nm, 820nm characteristic wavelength of one embodiment of the invention;
Fig. 5 show one embodiment of the invention after threshold division and dilation erosion area-of-interest schematic diagram corresponding to the first meat products sample;
Fig. 6 shows the evaluation result schematic diagram of the projected relationship model of one embodiment of the invention.
Embodiment
For making the object of the embodiment of the present invention, technical scheme and advantage clearly, below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly described, obviously, described embodiment is the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
Fig. 1 shows the meat products Noninvasive Measuring Method of Freshness process flow diagram based on spectral imaging technology of one embodiment of the invention.As shown in Figure 1, the method comprises:
S1: obtain the high spectrum image of the first meat products sample under the different default storage time;
S2: freshness component concentration corresponding under detecting described first meat products sample default storage time described in each;
S3: utilize image processing techniques to carry out Iamge Segmentation to the high spectrum image of the first meat products sample obtained, extracts the area-of-interest that the first meat products sample in described high spectrum image is corresponding;
S4: the spectral signature and the image texture characteristic that extract area-of-interest corresponding to described first meat products sample;
S5: build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic according to the spectral signature of the freshness component concentration of the described first meat products sample area-of-interest corresponding with described first meat products sample and image texture characteristic;
S6: the high spectrum image obtaining meat products to be measured, the high spectrum image of image processing techniques to the meat products to be measured obtained is utilized to carry out Iamge Segmentation, and extract the area-of-interest of the high spectrum image of described meat products to be measured, to extract spectral signature to be measured and the testing image textural characteristics of this area-of-interest;
S7: according to spectral signature to be measured and testing image textural characteristics, adopts the projected relationship model inversion set up in advance to obtain the freshness component concentration of described meat products to be measured.
The meat products Noninvasive Measuring Method of Freshness accuracy of detection based on spectral imaging technology of the present embodiment is high, and detection speed is fast, can carry out the detection of meat products freshness objective, quantitatively; Further, effectively can simplify testing process, shorten sense cycle; Chemical reagent consumption amount can be reduced simultaneously, reduce testing cost.
Preferred as the present embodiment, above-mentioned freshness component can be total volatile basic nitrogen TVB-N.TVB-N refers to the effect of animal food due to enzyme and bacterium, in decay process, make breaks down proteins and produce the alkaline nitrogen substance such as ammonia and amine, this type of material has volatility, its content is higher, show that amino acid is destroyed more, particularly methionine and tyrosine, therefore nutritive value is greatly affected.
Further, extract the spectral signature of area-of-interest corresponding to described first meat products sample in step S4, preferably include:
Utilize the algorithm RF that leapfrogs at random to extract and can characterize the characteristic wavelength of described first meat products sample freshness change in storage process as spectral signature.
On this basis, the above-mentioned characteristic wavelength that can characterize the freshness change in storage process of described first meat products sample comprises 660nm, 720nm, 790nm, 820nm, 880nm, 910nm.
Meanwhile, utilize the high spectrum image of image processing techniques to the first meat products sample obtained to carry out Iamge Segmentation in step S3, preferably include:
According to described characteristic wavelength, to the smoothing denoising of area-of-interest corresponding to described first meat products sample, the process of mask gray processing, and utilizing the image partition method that Otsu ' s thresholding method combines with dilation erosion algorithm, the area-of-interest corresponding to the first meat products sample through smoothing denoising, the process of mask gray processing carries out background removal.
On this basis, extract the image texture characteristic of area-of-interest corresponding to described first meat products sample in step S4, preferably include:
The feature extraction algorithm utilizing local binary pattern LBP and Tamura textural characteristics method to combine extracts the image texture characteristic of area-of-interest corresponding to described first meat products sample.
Preferred as the present embodiment, step S5 can comprise:
Principal component analysis (PCA) PCA method is utilized to merge, the spectral signature of area-of-interest corresponding for described first meat products sample and image texture characteristic according to this fusion results using the input variable of PCA score as modeling;
According to described input variable, artificial fish-swarm extreme learning machine AF-ELM algorithm is utilized to build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic.
Further, after step S5, the method also can comprise the accuracy and robustness of evaluating described projected relationship model.
Such as: forecast set meat sample is used for building projected relationship model in step S5, with root-mean-square error (RMSE) and the coefficient of determination (R
2) be evaluation criterion, model prediction of output value and the TVB-N value measured by chemical experiment are carried out paired t-test, the accuracy of further evaluation model and robustness simultaneously, determine best meat product freshness projected relationship model.
In addition, before step S3, the method also can comprise carries out black and white correction to the high spectrum image of the first meat products sample obtained.
On this basis, before step S3, the method also can comprise further carries out gaussian filtering process and color enhancement conversion process to the high spectrum image of the first meat products sample corrected through black and white.
Below by citing, the Whole Work Flow of described method is described.
Fig. 2 shows the structural representation of the high spectrum image acquisition system of one embodiment of the invention; As shown in Figure 2, spiced beef sample 1-1 is placed on objective table 1-2, and adjust the vertical range of objective table 1-2 and amasthenic lens 1-5 and normal bit consistent; After instrument starts, control article carrying platform 1-2 by precision stepper motor 1-3 and move horizontally, guarantee that meat sample 1-1 can be scanned by gross area; A set of 150W stablizes output halogen lamp light source 1-4 and sends incident light, after the effects such as meat sample 1-1 absorption, scattering, diffuse reflection, meat surface reflected light line focus camera lens 1-5 focuses on, then by after spectrometer 1-6 light splitting, imaging is carried out again by face battle array EMCCD camera 1-7, finally by CCD controller, the hyperspectral image data of the meat sample collected is sent in computing machine 1-8, to carry out follow-up analysis and treament.Except computing machine 1-8, whole system is all placed in the middle of a camera bellows 1-9, to avoid the interference of external stray light.
Use this system to gather high spectrum image that SPECTRAL REGION is 325 ~ 1100nm, spectral resolution is 2.8nm, and the spatial resolution of the image of acquisition is 1004 × 1002.The crack width of incident light is 30 μm, and the camera exposure time is 22ms, and the translational speed of objective table is 0.85mm/s, and the distance of meat sample and amasthenic lens is 480mm.
Fig. 3 shows the meat products freshness quantitative forecast relational model construction method process flow diagram of one embodiment of the invention.As shown in Figure 3, the method concrete steps are as follows:
A1: by the meat products sample obtained, such as spiced beef, carry out removal corner angle, the irregular process in edge, then default size is divided into (such as equably, regularly, 50mm × 35mm × 10mm), prepare 100, similar sample, and clean with the oil stain of oil-Absorbing Sheets by spiced beef surface, then load freshness protection package and carry out 0 under the condition of 0-4 DEG C, 3,5,7, the stored under refrigeration of 9 days, can from fresh to the change procedure of corruption in storage number of days to realize spiced beef.
Random for sample is divided into 5 groups, often organize 20, select 7 samples at random from every group, totally 35 samples are as forecast set, for evaluating accuracy and the robustness of the projected relationship model of foundation, remaining 65 samples are as modeling collection, and as the meat products sample setting up projected relationship model, the collection of high spectrum image is carried out in regularly sampling;
A2: after each collection high spectrum image, spiced beef sample is carried out to the mensuration of freshness index, the semimicro nitriding in GB/T5009.44-2003 (the Physicochemical test method of national Specification) is utilized to determine total volatile basic nitrogen (TVB-N) content in sample, as the component concentration characterizing meat products freshness, i.e. the reference value of quantitative test.
A3: the high spectrum image of spiced beef sample is analyzed:
A31: black and white correction is carried out to the hyperspectral image data of the spiced beef sample collected;
Particularly, at timing, keep image capturing system parameter constant, be first that 99% reference white correction plate is sampled to reflectivity, obtain entirely white uncalibrated image R
white, then close lens cap, and close light source, collect entirely black uncalibrated image R
dark, the high spectrum image after utilizing following formulae discovery to correct:
In formula, R
corrfor the high spectrum image after correction; R
imgfor the original high spectrum image gathered; DN is the maximal value of brightness, preferably, is taken as 4096.
A32: gaussian filtering process and color enhancement conversion are carried out to the meat products sample high spectrum image corrected through black and white, area-of-interest (ROI) is carried out to the sample image after process and divides, extract the average spectral data in meat sample ROI.
A33: utilize wavelet transformation to carry out denoising to all band average spectral data extracted, then adopt (RF) algorithm that leapfrogs at random to optimize to characterize 6 characteristic wavelengths of spiced beef freshness change: 660nm, 720nm, 790nm, 820nm, 880nm and 910nm.
A34: to the smoothing denoising of area-of-interest, the process of mask gray processing of the meat products sample high spectrum image under above-mentioned 6 characteristic wavelengths, utilize Otsu ' s thresholding method to carry out background removal in conjunction with dilation erosion algorithm to the characteristic image of meat sample on this basis, retain the image feature information of meat sample.
Particularly, Fig. 4-1 to Fig. 4-3 shows the spiced beef gray-scale map (spiced beef gray-scale map corresponding under 790nm, 880nm and 910nm characteristic wavelength is not shown) corresponding under 660nm, 720nm, 820nm characteristic wavelength of one embodiment of the invention, present the texture information of spiced beef in figure clearly, can be used as the master sample image of follow-up texture feature extraction.Fig. 5 show one embodiment of the invention after threshold division and dilation erosion area-of-interest schematic diagram corresponding to the first meat products sample.
A35: the characteristic image after splitting steps A 34, utilizes local binary pattern (LBP) to extract the characteristic parameter that can characterize meat sampled images texture variations in conjunction with Tamura Texture Segmentation Algorithm.LBP is from the textural characteristics of partial descriptions image, obtain the LPB eigenwert of ROI region, Tamura arthmetic statement textural characteristics mode is proceeded from the situation as a whole, extract roughness, contrast, direction degree, the linearity, rule degree and rough degree 6 textural characteristics values, totally 7 characteristic variables after data combine, can characterize the texture information of meat surface image better.
A4: the foundation of spiced beef freshness quantitative forecast relational model:
A41: extract modeling and concentrate 6 spectral variables of each sample under characteristic wavelength, and the 6 width characteristic images corresponding with these 6 spectral variables, and extract 42 textural characteristics variablees (see steps A 35, every width image zooming-out 7 textural characteristics variablees) of this 6 width characteristic image; Then principal component analysis (PCA) (PCA) is utilized to be merged by 48 variablees, according to fusion results, using the input variable of PCA score as modeling.
A42: the characteristic variable after merging is combined, as the training sample of extreme learning machine (ELM) with the corresponding TVB-N value determined in steps A 2.
A43:ELM is a kind of Novel learning algorithm of Single hidden layer feedforward neural networks, calculates and can resolve the output weights obtaining learning network, greatly improve generalization ability and the pace of learning of network by means of only a step.The mathematic(al) representation of ELM network model can be expressed as:
Known training sample is { x
j, t
j, 1≤j≤M, x
j=[x
j1, x
j2..., x
jm]
t, wherein M is the variable number after merging, t
jfor the TVB-N value of corresponding sample.L is node in hidden layer, f (w
ix
j+ b
i) be the excitation function of hidden layer neuron, w
ifor connecting the input weights of input neuron and i-th hidden layer node, b
ibe the threshold value of i-th hidden layer node, β
ifor connecting the output weights of i-th hidden node and output neuron, o
jfor the output valve of a jth input amendment, the cost function E of ELM can be expressed as:
S=(w in formula
i, b
i, i=1,2 ..., L), the training objective of extreme learning machine is exactly seek optimum (S, β), make model output valve and measured value error minimum, namely realize min (E (S, β)).
A44: utilize artificial fish-swarm (AF) algorithm to be minimised as criterion (S, β) parameter to ELM with cost function E (S, β) and carry out optimizing, set up the meat product freshness projected relationship model based on AF-ELM.Model is utilized to obtain the discreet value of meat sample TVB-N content.
A5: utilize forecast set sample to verify the model set up in steps A 4, obtain optimum freshness projected relationship model:
Fig. 6 shows the evaluation result schematic diagram of the projected relationship model of one embodiment of the invention.As shown in Figure 6, in this example, the modeling collection of institute's established model and the root-mean-square error (RMSE) of forecast set are respectively 0.093 and 0.109, the coefficient of determination (R
2) be respectively 0.977 and 0.964.As shown be forecast sample collection model evaluation result.Model predication value and the TVB-N content measured value utilizing semimicro nitriding to determine are carried out paired t-test, and testing result is p>0.05, there was no significant difference.Visible, it is feasible for utilizing the meat product freshness quantitative detecting method that the present invention is based on high light spectrum image-forming technology to detect for meat products class freshness.
High light spectrum image-forming technology is the integration technology of a kind of image and spectrum, space and the spectral information of experimental subjects can be obtained simultaneously, the present invention takes full advantage of the characteristic of this technology collection of illustrative plates unification, high light spectrum image-forming technology is used for meat products class freshness quantitatively to detect, with solve existing meat products Noninvasive Measuring Method of Freshness subjectivity strong, cannot quantize to detect, and testing process is loaded down with trivial details, sense cycle long and detect the problems such as chemical reagent consumption amount is large, testing cost is high.
The present invention takes full advantage of the characteristic of the collection of illustrative plates unification in high light spectrum image-forming technology, the spectral information of meat sample and image feature information are effectively merged, utilize local feature excavation to excavate with global characteristics the mode degree of depth combined and excavate the texture feature information of image, overcome the impact of sample inequality on model robustness, further Optimization Prediction relational model, makes final testing result more comprehensively, reliably.Compared with prior art, the precision of the quantitative forecast relational model that the present invention sets up and robustness have obvious lifting, for meat products quality intellectualized detection provides technical support, to guarantee food quality safety, safeguard that consumer health has direct realistic meaning.
Above embodiment only for illustration of technical scheme of the present invention, is not intended to limit; Although with reference to previous embodiment to invention has been detailed description, those of ordinary skill in the art is to be understood that: it still can be modified to the technical scheme described in foregoing embodiments, or carries out equivalent replacement to wherein portion of techniques feature; And these amendments or replacement, do not make the essence of appropriate technical solution depart from the spirit and scope of various embodiments of the present invention technical scheme.
Claims (10)
1., based on a meat products Noninvasive Measuring Method of Freshness for spectral imaging technology, it is characterized in that, comprising:
Obtain the high spectrum image of the first meat products sample under the different default storage time;
Freshness component concentration corresponding under detecting described first meat products sample default storage time described in each;
Utilize image processing techniques to carry out Iamge Segmentation to the high spectrum image of the first meat products sample obtained, extract the area-of-interest that the first meat products sample in described high spectrum image is corresponding;
Extract spectral signature and the image texture characteristic of area-of-interest corresponding to described first meat products sample;
The freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic is built according to the spectral signature of the freshness component concentration of the described first meat products sample area-of-interest corresponding with described first meat products sample and image texture characteristic;
Obtain the high spectrum image of meat products to be measured, the high spectrum image of image processing techniques to the meat products to be measured obtained is utilized to carry out Iamge Segmentation, and extract the area-of-interest of the high spectrum image of described meat products to be measured, to extract spectral signature to be measured and the testing image textural characteristics of this area-of-interest;
According to spectral signature to be measured and testing image textural characteristics, the projected relationship model inversion set up in advance is adopted to obtain the freshness component concentration of described meat products to be measured.
2., as claimed in claim 1 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, described freshness component is total volatile basic nitrogen TVB-N.
3., as claimed in claim 1 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the spectral signature of the area-of-interest that the described first meat products sample of described extraction is corresponding, comprising:
Utilize the algorithm RF that leapfrogs at random to extract and can characterize the characteristic wavelength of described first meat products sample freshness change in storage process as spectral signature.
4. as claimed in claim 3 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the described characteristic wavelength that can characterize the freshness change in storage process of described first meat products sample comprises 660nm, 720nm, 790nm, 820nm, 880nm, 910nm.
5. as claimed in claim 3 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the described high spectrum image of image processing techniques to the first meat products sample obtained that utilize carries out Iamge Segmentation, comprising:
According to described characteristic wavelength, to the smoothing denoising of area-of-interest corresponding to described first meat products sample, the process of mask gray processing, and utilizing the image partition method that Otsu ' s thresholding method combines with dilation erosion algorithm, the area-of-interest corresponding to the first meat products sample through smoothing denoising, the process of mask gray processing carries out background removal.
6., as claimed in claim 5 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the image texture characteristic of the area-of-interest that the described first meat products sample of described extraction is corresponding, comprising:
The feature extraction algorithm utilizing local binary pattern LBP and Tamura textural characteristics method to combine extracts the image texture characteristic of area-of-interest corresponding to described first meat products sample.
7. as claimed in claim 1 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, spectral signature and the image texture characteristic of the described area-of-interest corresponding with described first meat products sample according to the freshness component concentration of described first meat products sample build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic, comprising:
Principal component analysis (PCA) PCA method is utilized to merge, the spectral signature of area-of-interest corresponding for described first meat products sample and image texture characteristic according to this fusion results using the input variable of PCA score as modeling;
According to described input variable, artificial fish-swarm extreme learning machine AF-ELM algorithm is utilized to build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic.
8. as claimed in claim 1 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, after the spectral signature of the described area-of-interest corresponding with described first meat products sample according to the freshness component concentration of described first meat products sample and image texture characteristic build the freshness component concentration of meat products to be measured and the projected relationship model of described spectral signature and image texture characteristic, the method also comprises:
Evaluate accuracy and the robustness of described projected relationship model.
9. as claimed in claim 1 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the described high spectrum image of image processing techniques to the first meat products sample obtained that utilize carries out Iamge Segmentation, before extracting area-of-interest corresponding to the first meat products sample in described high spectrum image, the method also comprises:
Black and white correction is carried out to the high spectrum image of the first meat products sample obtained.
10. as claimed in claim 9 based on the meat products Noninvasive Measuring Method of Freshness of spectral imaging technology, it is characterized in that, the described high spectrum image of image processing techniques to the first meat products sample obtained that utilize carries out Iamge Segmentation, before extracting area-of-interest corresponding to the first meat products sample in described high spectrum image, the method also comprises:
Gaussian filtering process and color enhancement conversion process are carried out to the high spectrum image of the first meat products sample corrected through black and white.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510925619.8A CN105548029A (en) | 2015-12-14 | 2015-12-14 | Meat product freshness detection method based on spectral imaging technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510925619.8A CN105548029A (en) | 2015-12-14 | 2015-12-14 | Meat product freshness detection method based on spectral imaging technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105548029A true CN105548029A (en) | 2016-05-04 |
Family
ID=55827370
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510925619.8A Pending CN105548029A (en) | 2015-12-14 | 2015-12-14 | Meat product freshness detection method based on spectral imaging technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105548029A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106404692A (en) * | 2016-11-09 | 2017-02-15 | 大连工业大学 | Method for detecting freshness grade of instant sea cucumber by using hyperspectral imaging technology |
| CN106442339A (en) * | 2016-09-19 | 2017-02-22 | 北京农业质量标准与检测技术研究中心 | Visual detection method for refrigeration time of cooked meat and device |
| CN106570855A (en) * | 2016-07-22 | 2017-04-19 | 北京农业信息技术研究中心 | Method and system for quickly judging pork freshness |
| CN106841052A (en) * | 2016-12-29 | 2017-06-13 | 中国科学院遥感与数字地球研究所 | A kind of method and system of quick detection meat freshness |
| CN106960091A (en) * | 2017-03-20 | 2017-07-18 | 南京林业大学 | A kind of THz optical spectrum rapid nondestructive detection methods of fresh meat freshness K values |
| CN107271375A (en) * | 2017-07-21 | 2017-10-20 | 石河子大学 | A kind of high spectral image detecting method of quality of mutton index |
| CN108362652A (en) * | 2018-03-02 | 2018-08-03 | 江南大学 | A kind of object freshness lossless detection method based on evidence theory |
| CN108414476A (en) * | 2018-03-14 | 2018-08-17 | 中南民族大学 | A method of freshness of meat is detected by superweak bio-photon imaging system |
| CN108692815A (en) * | 2017-04-04 | 2018-10-23 | 手持产品公司 | Use the multispectral imaging of longitudinal chromatic aberration |
| CN109060675A (en) * | 2018-09-05 | 2018-12-21 | 东北大学 | One kind is for iron content detection method and device in iron ore |
| CN109612950A (en) * | 2018-12-29 | 2019-04-12 | 中国肉类食品综合研究中心 | A kind of freezing poultry meat storage period forecasting device and method |
| CN109856055A (en) * | 2016-11-15 | 2019-06-07 | 青岛农业大学 | Meat gruel based on multispectral imaging adulterates quick detection device |
| CN109883966A (en) * | 2019-02-26 | 2019-06-14 | 江苏大学 | A method of Eriocheir sinensis amount of cure is detected based on multispectral image technology |
| CN111929257A (en) * | 2020-08-31 | 2020-11-13 | 怀化市明友食品有限责任公司 | Meat product freshness detection method based on spectral imaging technology |
| CN112161937A (en) * | 2020-11-04 | 2021-01-01 | 安徽大学 | Wheat flour gluten degree detection method based on cascade forest and convolutional neural network |
| CN112229808A (en) * | 2020-09-21 | 2021-01-15 | 佛山国防科技工业技术成果产业化应用推广中心 | Food microorganism detection device and detection method based on multispectral technology |
| CN113588571A (en) * | 2021-09-29 | 2021-11-02 | 广东省农业科学院动物科学研究所 | Hyperspectral imaging-based method and system for identifying fishy smell of aquatic product |
| CN114252318A (en) * | 2021-12-27 | 2022-03-29 | 浙江大学 | Method and system for detecting staphylococcus aureus in chicken |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344577A (en) * | 2013-07-12 | 2013-10-09 | 中国农业大学 | Non-destructive detection method for freshness of livestock meat based on multispectral imaging technology |
| CN103439285A (en) * | 2013-08-19 | 2013-12-11 | 华南理工大学 | Fish fillet freshness detection method based on hyperspectral imaging |
| CN104655586A (en) * | 2015-02-28 | 2015-05-27 | 华南理工大学 | Hyperspectral-data-fusio-based fast non-contact fish fat oxidation monitoring method |
-
2015
- 2015-12-14 CN CN201510925619.8A patent/CN105548029A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344577A (en) * | 2013-07-12 | 2013-10-09 | 中国农业大学 | Non-destructive detection method for freshness of livestock meat based on multispectral imaging technology |
| CN103439285A (en) * | 2013-08-19 | 2013-12-11 | 华南理工大学 | Fish fillet freshness detection method based on hyperspectral imaging |
| CN104655586A (en) * | 2015-02-28 | 2015-05-27 | 华南理工大学 | Hyperspectral-data-fusio-based fast non-contact fish fat oxidation monitoring method |
Non-Patent Citations (2)
| Title |
|---|
| 王家云等: "基于NIR高光谱成像技术的滩羊肉内部品质无损检测", 《现代食品科技》 * |
| 祝志慧等: "基于高光谱信息融合和相关向量机的种蛋无损检测", 《农 业 工 程 学 报》 * |
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570855A (en) * | 2016-07-22 | 2017-04-19 | 北京农业信息技术研究中心 | Method and system for quickly judging pork freshness |
| CN106442339A (en) * | 2016-09-19 | 2017-02-22 | 北京农业质量标准与检测技术研究中心 | Visual detection method for refrigeration time of cooked meat and device |
| CN106442339B (en) * | 2016-09-19 | 2019-03-15 | 北京农业质量标准与检测技术研究中心 | Cold cuts cold preservation time visible detection method and device |
| CN106404692A (en) * | 2016-11-09 | 2017-02-15 | 大连工业大学 | Method for detecting freshness grade of instant sea cucumber by using hyperspectral imaging technology |
| CN109856055A (en) * | 2016-11-15 | 2019-06-07 | 青岛农业大学 | Meat gruel based on multispectral imaging adulterates quick detection device |
| CN106841052A (en) * | 2016-12-29 | 2017-06-13 | 中国科学院遥感与数字地球研究所 | A kind of method and system of quick detection meat freshness |
| CN106841052B (en) * | 2016-12-29 | 2019-07-12 | 中国科学院遥感与数字地球研究所 | A kind of method and system of quick detection meat freshness |
| CN106960091B (en) * | 2017-03-20 | 2020-05-05 | 南京林业大学 | THz spectrum rapid nondestructive testing method for fresh meat freshness K value |
| CN106960091A (en) * | 2017-03-20 | 2017-07-18 | 南京林业大学 | A kind of THz optical spectrum rapid nondestructive detection methods of fresh meat freshness K values |
| CN108692815B (en) * | 2017-04-04 | 2021-08-31 | 手持产品公司 | Multispectral imaging using longitudinal chromatic aberration |
| CN108692815A (en) * | 2017-04-04 | 2018-10-23 | 手持产品公司 | Use the multispectral imaging of longitudinal chromatic aberration |
| CN107271375A (en) * | 2017-07-21 | 2017-10-20 | 石河子大学 | A kind of high spectral image detecting method of quality of mutton index |
| CN107271375B (en) * | 2017-07-21 | 2019-10-01 | 石河子大学 | A kind of high spectral image detecting method of quality of mutton index |
| CN108362652A (en) * | 2018-03-02 | 2018-08-03 | 江南大学 | A kind of object freshness lossless detection method based on evidence theory |
| CN108362652B (en) * | 2018-03-02 | 2020-06-09 | 江南大学 | Object freshness nondestructive testing method based on evidence theory |
| CN108414476A (en) * | 2018-03-14 | 2018-08-17 | 中南民族大学 | A method of freshness of meat is detected by superweak bio-photon imaging system |
| CN109060675A (en) * | 2018-09-05 | 2018-12-21 | 东北大学 | One kind is for iron content detection method and device in iron ore |
| CN109612950A (en) * | 2018-12-29 | 2019-04-12 | 中国肉类食品综合研究中心 | A kind of freezing poultry meat storage period forecasting device and method |
| CN109883966B (en) * | 2019-02-26 | 2021-09-10 | 江苏大学 | Method for detecting aging degree of eriocheir sinensis based on multispectral image technology |
| CN109883966A (en) * | 2019-02-26 | 2019-06-14 | 江苏大学 | A method of Eriocheir sinensis amount of cure is detected based on multispectral image technology |
| CN111929257A (en) * | 2020-08-31 | 2020-11-13 | 怀化市明友食品有限责任公司 | Meat product freshness detection method based on spectral imaging technology |
| CN112229808A (en) * | 2020-09-21 | 2021-01-15 | 佛山国防科技工业技术成果产业化应用推广中心 | Food microorganism detection device and detection method based on multispectral technology |
| CN112161937A (en) * | 2020-11-04 | 2021-01-01 | 安徽大学 | Wheat flour gluten degree detection method based on cascade forest and convolutional neural network |
| CN113588571A (en) * | 2021-09-29 | 2021-11-02 | 广东省农业科学院动物科学研究所 | Hyperspectral imaging-based method and system for identifying fishy smell of aquatic product |
| CN113588571B (en) * | 2021-09-29 | 2021-12-03 | 广东省农业科学院动物科学研究所 | Hyperspectral imaging-based method and system for identifying fishy smell of aquatic product |
| CN114252318A (en) * | 2021-12-27 | 2022-03-29 | 浙江大学 | Method and system for detecting staphylococcus aureus in chicken |
| CN114252318B (en) * | 2021-12-27 | 2023-11-17 | 浙江大学 | Method and system for detecting staphylococcus aureus in chicken |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105548029A (en) | Meat product freshness detection method based on spectral imaging technology | |
| Khulal et al. | Intelligent evaluation of total volatile basic nitrogen (TVB-N) content in chicken meat by an improved multiple level data fusion model | |
| CN106546581B (en) | Test paper detection card intelligent detection system and test paper detection card intelligent analysis method | |
| Li et al. | Nondestructive detection of total volatile basic nitrogen (TVB-N) content in pork meat by integrating hyperspectral imaging and colorimetric sensor combined with a nonlinear data fusion | |
| CN103278464B (en) | Flesh of fish detection method and device | |
| CN103900972B (en) | Multi-feature fusion-based meat freshness hyperspectral image visual detection | |
| CN106023158B (en) | The fresh water pipless pearl pearly layer defect identification method of SD-OCT images | |
| CN106596416A (en) | Chilled fresh meat quality non-destructive testing method based on hyperspectral imaging technology | |
| TWI684997B (en) | Establishing method of bone age assessment and height prediction model, bone age assessment and height prediction system, and bone age assessment and height prediction method | |
| CN101806703B (en) | Non-destructive inspection method of total amount of meat bacteria | |
| Paliwal et al. | Classification of cereal grains using a flatbed scanner | |
| CN109490306B (en) | Pork freshness detection method based on color and smell data fusion | |
| CN108956604A (en) | A method of Eriocheir sinensis quality is identified based on hyper-spectral image technique | |
| CN103839283A (en) | Area and circumference nondestructive measurement method of small irregular object | |
| CN104198695A (en) | Method for analyzing developing result of colloidal gold test strip | |
| CN115170572A (en) | BOPP composite film surface gluing quality monitoring method | |
| Chen et al. | Evaluating fabric pilling with light-projected image analysis | |
| CN110161233A (en) | A kind of fast quantitative measurement method for detecting of immune chromatography test card | |
| EP3688182B1 (en) | Methods and systems for automated assessment of antibiotic sensitivity | |
| CN107590816B (en) | Water body information fitting method based on remote sensing image | |
| CN114419311B (en) | Multi-source information-based passion fruit maturity nondestructive testing method and device | |
| Feng et al. | Detection of blood spots in eggs by hyperspectral transmittance imaging | |
| US20100266191A1 (en) | Method for quantifying an underlying property of a multitude of samples | |
| CN108240806A (en) | A kind of method of EO-1 hyperion Depth extraction waveband selection | |
| CN108645798A (en) | The method of on-line quick detection chicken content of lactic acid bacteria |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160504 |
|
| RJ01 | Rejection of invention patent application after publication |