CN102890092A - Characteristic angle cosine value method for detecting defects of honey peach brown rot - Google Patents

Characteristic angle cosine value method for detecting defects of honey peach brown rot Download PDF

Info

Publication number
CN102890092A
CN102890092A CN2012103866365A CN201210386636A CN102890092A CN 102890092 A CN102890092 A CN 102890092A CN 2012103866365 A CN2012103866365 A CN 2012103866365A CN 201210386636 A CN201210386636 A CN 201210386636A CN 102890092 A CN102890092 A CN 102890092A
Authority
CN
China
Prior art keywords
pixel
value
image
honey peach
brown rot
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.)
Granted
Application number
CN2012103866365A
Other languages
Chinese (zh)
Other versions
CN102890092B (en
Inventor
饶秀勤
陈思
应义斌
张若宇
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 DEFEILUO INTELLIGENT MACHINERY MANUFACTURING Co.,Ltd.
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 CN201210386636.5A priority Critical patent/CN102890092B/en
Publication of CN102890092A publication Critical patent/CN102890092A/en
Application granted granted Critical
Publication of CN102890092B publication Critical patent/CN102890092B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

The invention discloses a characteristic angle cosine value method for detecting defects of honey peach brown rot. The method comprises the following steps of: extracting images at the wave bands of 660nm, 680nm and 700nm from a honey peach hyperspectral image, and performing single threshold segmentation on the image at the wave band of 660nm to obtain a fruit area; performing 3*3 mean filtering on the fruit areas of the images at the three wave bands, and performing mean normalization on the spectrum; by taking a pixel point in the fruit area, taking a wavelength value as a horizontal coordinate, taking the spectrum normalization value as a vertical coordinate, taking an included angle ABC formed by three points of A (lambdaA, RA), B(lambdaB, RB) and C(lambdaC, RC) as a characteristic angle and taking a cosine value of the characteristic angle as a characteristic value, classifying pixels in the fruit areas of the images, and detecting the defects of the brown rot. The defects of honey peach brown rot are detected by using three wave bands only, the detection cost is reduced, the surface coloration interference of the honey peach is eliminated, and the method can be used for detecting the defects of apples and other fruits containing chlorophyll.

Description

The characteristic angle cosine value method that is used for honey peach brown rot defects detection
Technical field
The present invention relates to the fruit defects detection method, especially relate to a kind of characteristic angle cosine value method for honey peach brown rot defects detection.
Background technology
Honey peach is storage tolerance not, and various mechanical damages and infection process very easily occur in the transporting procedures, causes breakdown of fruit, gives orchard worker and consumer's economic loss.No matter eat peach, can peach or fruit juice peach raw, before coming into the market and processing, all to select and classification.Therefore, the Non-Destructive Testing of honey peach defective is necessary.
Because the honey peach surface color has background color and painted dividing, so that the difficulty of utilizing RGB color machines vision system to detect the honey peach defective strengthens, and the detected sample information of hyper-spectral image technique has merged image information and spectral information, and it can comprehensively react the situation such as external feature, surface imperfection of sample.Although it can not well satisfy the purpose of fast detecting, but it has an important effect is exactly the validity feature wavelength of determining the sample quality, for building of multispectral image system and Vision Builder for Automated Inspection provides theoretical foundation, online to realize, quick, Non-Destructive Testing quality of agricultural product purpose.
For example, utilize spectrum and image co-registration information, extract the spectral information of pixel in the image and set up the sorters such as LDA, QDA, structure cut apart bianry image detect the cotton magazine (Guo Junxian. based on the research .2011 of the moits detection method of high light spectrum image-forming technology, Zhejiang University .); Utilizing the sorting techniques such as k-NN, LDC, SVM that apple multi-band image pixel is classified cuts apart, realize apple defects detection (Unay D, Gosselin B.Automatic defect segmentation of ' Jonagold ' apples on multi-spectralimages:A comparative study.Postharvest Biology and Technology, 2006,42 (3): 271-279.)
At present, few as the high spectrum image research and comparison of research object with honey peach, great majority all are the high spectral investigations for other fruit surface defects, the domestic relevant report that yet there are no relevant honey peach defects detection.Gowen etc. utilize high spectral technique to detect white mushroom surface bruise, respectively high spectrum image cubic block and 600 spectrum (300 normal districts and 300 bruise districts) are done the PCA analysis, obtain major component component image and virtual image, through image segmentation, identify defect area.Xing etc. utilize high spectral technique to detect " Qiao Najin " apple surface bruise, cause that based on bruise the apple surface distortion makes PC1 image level line this principle that changes, and works out the bruise recognizer.Zwiggelaar etc. utilize spectroanalysis instrument and 3 kinds of sorting criterions to analyze the gained spectrum in conjunction with the bruise on spectral information and machine vision technique detection peach and apricot surface, obtain the optimal double band combination, 930nm and 970nm.Add filter plate by the CCD camera and obtain the band combination image, adopt flooding, be partitioned into defect area.But the involved wave band number of front two is too many, is unfavorable for the realization of multispectral on-line detecting system.
Summary of the invention
The object of the present invention is to provide a kind of characteristic angle cosine value method for honey peach brown rot defects detection, by average normalization the SPECTRAL DIVERSITY that is caused by the sphere effect to be revised, the cosine value of the characteristic angle that consists of according to the gray-scale value of 3 wave bands of pixel comes this pixel is classified, and realizes the honey peach defects detection.
The technical solution adopted for the present invention to solve the technical problems is:
The step of the method is as follows: extract 660nm, 680nm and 700nm band image in the honey peach high spectrum image, image is carried out after binary segmentation goes background, mean filter and average normalization, calculate the cosine value of the characteristic angle of 3 band image pixels formations, make up the LDA pixel classifier with this as eigenwert, realize honey peach brown rot defects detection.
Described extraction band image: take the black pallet as background, gather high spectrum image, and extract the honey peach image of 660nm, 680nm and 700nm wave band;
Describedly image is carried out binary segmentation go background: take the 680nm band image as object, carry out Threshold segmentation, namely adopting threshold value is cutting apart of image pixel maximum gradation value * 0.12, is the fruit zone greater than the zone of this threshold value, namely obtains the fruit zone;
Described mean filter: 3 * 3 mean filters are carried out in the fruit zone to 660nm, 680nm and 700nm band image, namely the pixel in 8 neighborhoods around each pixel and this pixel is averaged as the value at this pixel place, this 8 neighborhood travels through whole fruit zone successively with 3 * 3 mask Matrix covers;
Described average normalization: each pixel (x, y) in the fruit zone is handled as follows: the spectral value discrete series that pixel (x, y) is located is averaged, namely the spectral value summation at 660nm, 680nm and 700nm wave band place is averaged,
Figure BDA00002248426200021
Each band spectrum value realizes the average normalization of high spectrum image divided by this average; Computing formula is as follows:
R MeanN ( x , y , λ i ) = R MeanF ( x , y , λ i ) 1 N Σ i = 1 N R MeanF ( x , y , λ i ) , (i＝1,2,……，N)（1）
Wherein, N----is the wave band number, here N=3, i.e. λ 1=660nm, λ 2=680nm, λ 3=700nm; R MeanN(x, y, λ i)----be through the λ after the average normalization iThe band image at wavelength place; R MeanF(x, y, λ i)----be through the λ behind the mean filter iThe band image at wavelength place;
The cosine value of described characteristic angle calculates: choose the pixel (x, y) in the fruit zone, take wavelength value as horizontal ordinate, the spectrum normalized value is that ordinate is made spectrogram, gets respectively 3 points of this pixel spectrogram: A(λ A, R A), B(λ B, R B), C(λ C, R C), take angle ∠ ABC as characteristic angle, λ A=660nm, λ B=680nm, λ C=700nm, adopt formula (2) to calculate its cosine value:
cos θ = 0.002 + ( R A - R B ) ( R C - R B ) 0.0025 + ( R A - R B ) 2 0.0016 + ( R C - R B ) 2 - - - ( 2 )
The described sorter of setting up: take the characteristic angle cosine value cos θ of each pixel as eigenwert, get normal district and brown rot lesion pixel each 100, be divided into 0 class and 1 class: normally distinguishing pixel is 0 class, and brown rot lesion pixel is 1 class, as training set, make up the LDA pixel classifier;
Described honey peach brown rot defects detection: calculate the characteristic angle cosine value cos θ of each pixel in the fruit image fruit to be measured zone, if the value of cos θ distance 0 class cluster is nearer than 1 class cluster, then is judged to 0 class, otherwise is judged to 1 class; According to this principle, the LDA pixel classifier of utilizing step 6) to make up is classified to the pixel in the fruit zone, and brown rot defect area pixel value is 1, and the normal region is 0, realizes honey peach brown rot defects detection.
The useful effect that the present invention has is:
The present invention only realizes honey peach brown rot defects detection with 3 wave bands; can reduce the cost of detection system; eliminate the honey peach surface colour to the interference of defects detection, also can be used for containing such as apple etc. the defects detection of chlorophyllous fruit, or the identification of carpopodium calyx.
Description of drawings
Fig. 1 is the characteristic angle cosine value method realization flow figure of honey peach brown rot defects detection of the present invention.
Fig. 2 is image capturing system synoptic diagram of the present invention.
Fig. 3 is spectral curve before the pre-service of all band spectrum.
Fig. 4 is spectral curve after the pre-service of all band spectrum.
Fig. 5 is spectral curve before the 3 band spectrum pre-service.
Fig. 6 is spectral curve after the 3 band spectrum pre-service.
Fig. 7 is the characteristic angle synoptic diagram that the present invention's 3 wave bands consist of.
Fig. 8 be original image and through pre-service, do not pass through split image after the pre-service.
Fig. 9 is 630nm, 680nm, two class pixel characteristic angle cosine value scatter diagrams under the 720nm band combination.
Figure 10 is 550nm, 680nm, two class pixel characteristic angle cosine value scatter diagrams under the 720nm band combination.
Figure 11 is 550nm, 680nm, two class pixel characteristic angle cosine value scatter diagrams under the 900nm band combination.
Figure 12 is 630nm, 680nm, two class pixel characteristic angle cosine value scatter diagrams under the 900nm band combination.
Among the figure: 1, linear array CCD camera; 2, spectrometer; 3, sample; 4, objective table; 5, conveyer; 6, light source; 7, computing machine.
Embodiment
The invention will be further described below in conjunction with drawings and Examples.
Figure 1 shows that the characteristic angle cosine value method realization flow figure of honey peach brown rot defects detection.At first, the debugging Hyperspectral imager, image is as the criterion can gather out clearly, selects simultaneously the honey peach sample, is divided into normal fruit and brown rot fruit, then gathers their high spectrum image; Then be divided into two steps processing spectroscopic data: the one, manual extraction normal pixel and brown rot pixel through sequence of operations such as average normalized, calculated characteristics angle cosine values, make up the LDA pixel classifier; The 2nd, obtain all pixel spectra in fruit zone, carry out and identical before operation, make up the test set image.At last the test set data are input to the classification of LDA pixel wherein, realize honey peach brown rot defects detection.
As shown in Figure 2, Hyperspectral imager mainly comprises linear array CCD camera 1; Spectrometer 2; Sample 3; Objective table 4; Conveyer 5; Light source 6; Computing machine 7.Line of linear array CCD camera 1 every transversal scanning obtains in this sweep trace each pixel at the spectral information of whole wavelength band.Simultaneously, sample 3 vertically moves along doing perpendicular to the direction of sweep trace on objective table, and comprehensive transverse and longitudinal scanning information just can obtain the three-dimensional high spectrum image data of sample.
Light intensity at high spectroscopic system illumination system tends towards stability, the translational speed of objective table can not produce in the situation of deformation with the material object that the camera exposure time coordinates mutually, camera photographed, honey peach is placed camera vision area center, gather the honey peach high spectrum image.
The image of 660nm, 680nm and 700nm wave band in the extraction high spectrum image, and the 680nm band image carried out binary image segmentation, threshold value is made as 0.12 times of this image max pixel value, zone greater than this threshold value is the fruit zone, all the other are all background, consist of thus mask image and each band image multiplies each other, and remove to contain noisy background area, keep the fruit zone, subsequent treatment all operates for pixel in this fruit zone.
In this experimental system, since higher than the position intensity of illumination away from light source near the position intensity of illumination of light source, so that the spectral reflectance value is larger.Illumination intensity, reflected spectrum and blank and the distance between the camera (or light source) perpendicular to the reference white plate surface of spectrum camera are linear, and distance is nearer, and spectral value is larger.And the honey peach shape is the class sphere, causes the reflected spectrum of differentiated levels there are differences, so that the picture centre position that obtains is brighter than marginal position brightness.In order to overcome this impact, need spectrum is carried out pre-service, revise the SPECTRAL DIVERSITY that sphere causes.
The honey peach sample of experimental selection surface color uniformity (complete red or entirely green) extracts in the honey peach high spectrum image subject area, radially the averaged spectrum of 3 * 3 pixel size ROI of 8 on the radius.These spectrum are carried out the average normalized in all band.Fig. 3 is all band original spectrum curve, and Fig. 4 is the curve map of average normalization preprocess method after to spectral manipulation, can see the curve that originally relatively disperses, and it is more concentrated to become now.The method is effective equally for the spectrum of 3 wave bands, as shown in Figure 5 and Figure 6.
According to result shown in Figure 6, each pixel (x, y) in the fruit zone is handled as follows: the spectral value discrete series that pixel (x, y) is located is averaged, namely the spectral value summation at 660nm, 680nm and 700nm wave band place is averaged, Each band spectrum value realizes the average normalization of high spectrum image divided by this average.Computing formula is as follows:
R MeanN ( x , y , λ i ) = R MeanF ( x , y , λ i ) 1 N Σ i = 1 N R MeanF ( x , y , λ i ) , ( i = 1,2 , . . . . . . , N )
Wherein, N----is the wave band number, here N=3, i.e. λ 1=660nm, λ 2=680nm, λ 3=700nm; R MeanN(x, y, λ i)----be through the λ after the average normalization iThe band image at wavelength place; R MeanF(x, y, λ 1)----be through the λ behind the mean filter iThe band image at wavelength place.
Among Fig. 7, take wavelength as horizontal ordinate, the corresponding spectral value of this wavelength is ordinate, gets three wave band points in the spectrogram: A(λ A, R A), B(λ B, R B), C(λ C, R C), forming a triangle, wavelength is designated as respectively λ A=660nm, λ B=680nm, λ C=700nm, the spectral value at this wavelength place are R A, R B, R CLeg-of-mutton angle ∠ ABC=θ is characteristic angle, by following computing formula, and calculated characteristics cosine of an angle value:
cos θ = 0.002 + ( R A - R B ) ( R C - R B ) 0.0025 + ( R A - R B ) 2 0.0016 + ( R C - R B ) 2
As eigenwert, 100 normal district's pixel spectra of extraction and 100 brown rot defect area pixel spectra make up the LDA pixel classifier as training set with the cosine value cos θ of characteristic angle.Calculate the characteristic angle cosine value cos θ of each pixel in the fruit image fruit to be measured zone, if the value of cos θ distance 0 class cluster is nearer than 1 class cluster, then is judged to 0 class, otherwise is judged to 1 class.According to this principle, utilize the LDA pixel classifier to classifying with the pixel in the side fruit image fruit zone, brown rot defect area pixel value is 1, the normal region is 0, realizes honey peach brown rot defects detection.
Fig. 8 is λ 1=660nm, λ 2=680nm, λ 3Result after the characteristic angle cosine value of=700nm band combination is cut apart as eigenwert, the result shows, average method for normalizing can reduce the adverse effect that edge effect that sphere causes brings preferably, and 3 wave band characteristic angle cosine values can be partitioned into defect area preferably as eigenwert.
λ 1=660nm, λ 2=680nm, λ 3Determining of=700nm band combination: analyze according to PCA, 5 characteristic wave bands are arranged, 550nm, 630nm, 680nm, 720nm, 900nm.Wherein, wavelength 680nm is the chlorophyll absorption peak, is a special wave band.Determine λ 2=680nm immobilizes, λ 1=550nm or 630nm, λ 3Therefore=720nm or 900nm, have 4 kind of 3 band combination.From the brown rot defect sample, extract 100 normal district pixel wave spectrums, 100 brown rot defect area pixel wave spectrums, calculate normal district and brown rot defective institute's constitutive characteristic cosine of an angle value cos θ under these 4 kinds of band combinations according to said method, its average and standard deviation are as shown in table 1.
For selecting more excellent band combination, respectively with the characteristic angle cosine value of above-mentioned 4 kinds of band combinations as eigenwert, as test set, image is carried out Classification and Identification with all pixels in the fruit zone in the normal fruit of 15 defect sample and 10 sample, its recognition accuracy is as shown in table 2.
Fig. 9, Figure 10, Figure 11, Figure 12 show: in 4 kinds of band combinations, and center wave band λ 2=680nm remains unchanged, λ 1, λ 3More close to center wave band, normal district is just better with brown rot defect area separation property.Therefore, choose λ 1=660nm, λ 2=680nm, λ 3=700nm is that characteristic wave bands comes calculated characteristics angle cosine value.
Characteristic angle cosine value statistics under table 14 kind of the band combination
Annotate: combination 1={630nm, 680nm, 720nm}; Combination 2={550nm, 680nm, 720nm};
Combination 3={550nm, 680nm, 900nm}; Combination 4={630nm, 680nm, 900nm}.
Brown rot fruit and normal fruit discrimination under table 24 kind of the band combination
Figure BDA00002248426200062

Claims (2)

1. characteristic angle cosine value method that is used for honey peach brown rot defects detection, it is characterized in that, the step of the method is as follows: extract 660nm, 680nm and 700nm band image in the honey peach high spectrum image, image is carried out after binary segmentation goes background, mean filter and average normalization, calculate the cosine value of the characteristic angle of 3 band image pixels formations, make up the LDA pixel classifier with this as eigenwert, realize honey peach brown rot defects detection.
2. a kind of characteristic angle cosine value method for honey peach brown rot defects detection according to claim 1 is characterized in that:
1) described extraction band image: take the black pallet as background, gather high spectrum image, and extract the honey peach image of 660nm, 680nm and 700nm wave band;
2) describedly image is carried out binary segmentation go background: take the 680nm band image as object, carry out Threshold segmentation, namely adopting threshold value is cutting apart of image pixel maximum gradation value * 0.12, is the fruit zone greater than the zone of this threshold value;
3) described mean filter: 3 * 3 mean filters are carried out in the fruit zone to 660nm, 680nm and 700nm band image, namely the pixel in 8 neighborhoods around each pixel and this pixel is averaged as the value at this pixel place, this 8 neighborhood travels through whole fruit zone successively with 3 * 3 mask Matrix covers;
4) described average normalization: each pixel (x, y) in the fruit zone is handled as follows: the spectral value discrete series that pixel (x, y) is located is averaged, namely the spectral value summation at 660nm, 680nm and 700nm wave band place is averaged,
Figure FDA00002248426100011
Each band spectrum value realizes the average normalization of high spectrum image divided by this average; Computing formula is as follows:
R MeanN ( x , y , λ i ) = R MeanF ( x , y , λ i ) 1 N Σ i = 1 N R MeanF ( x , y , λ i ) , (i＝1,2,……，N)（1）
Wherein, N----is the wave band number, here N=3, i.e. λ 1=660nm, λ 2=680nm, λ 3=700nm; R MeanN(x, y, λ i)----be through the λ after the average normalization iThe band image at wavelength place; R MeanF(x, y, λ i)----be through the λ behind the mean filter iThe band image at wavelength place;
5) cosine value of described characteristic angle calculates: choose the pixel (x, y) in the fruit zone, take wavelength value as horizontal ordinate, the spectrum normalized value is that ordinate is made spectrogram, gets respectively 3 points of this pixel spectrogram: A(λ A, R A), B(λ B, R B), C(λ C, R C), take angle ∠ ABC as characteristic angle, λ A=660nm, λ B=680nm, λ C=700nm, adopt formula (2) to calculate its cosine value:
cos θ = 0.002 + ( R A - R B ) ( R C - R B ) 0.0025 + ( R A - R B ) 2 0.0016 + ( R C - R B ) 2 - - - ( 2 )
6) the described sorter of setting up: take the characteristic angle cosine value cos θ of each pixel as eigenwert, get normal district and brown rot lesion pixel each 100, be divided into 0 class and 1 class: normally distinguishing pixel is 0 class, and brown rot lesion pixel is 1 class, as training set, make up the LDA pixel classifier;
7) described honey peach brown rot defects detection: calculate the characteristic angle cosine value cos θ of each pixel in the fruit image fruit to be measured zone, if the value of cos θ distance 0 class cluster is nearer than 1 class cluster, then is judged to 0 class, otherwise is judged to 1 class; According to this principle, the LDA pixel classifier of utilizing step 6) to make up is classified to the pixel in the fruit zone, and brown rot defect area pixel value is 1, and the normal region is 0, realizes honey peach brown rot defects detection.
CN201210386636.5A 2012-10-12 2012-10-12 Characteristic angle cosine value method for detecting defects of honey peach brown rot Active CN102890092B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201210386636.5A CN102890092B (en) 2012-10-12 2012-10-12 Characteristic angle cosine value method for detecting defects of honey peach brown rot

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201210386636.5A CN102890092B (en) 2012-10-12 2012-10-12 Characteristic angle cosine value method for detecting defects of honey peach brown rot

Publications (2)

Publication Number Publication Date
CN102890092A true CN102890092A (en) 2013-01-23
CN102890092B CN102890092B (en) 2014-12-17

Family

ID=47533665

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201210386636.5A Active CN102890092B (en) 2012-10-12 2012-10-12 Characteristic angle cosine value method for detecting defects of honey peach brown rot

Country Status (1)

Country Link
CN (1) CN102890092B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103149214A (en) * 2013-02-27 2013-06-12 上海交通大学 Method for detecting flaw on surface of fruit
CN105891229A (en) * 2014-09-05 2016-08-24 熊菊莲 Method for determining characteristic wavelength for spectral image analysis and detection of surfaces of fruits
CN105975966A (en) * 2016-04-21 2016-09-28 南京农业大学 Rice grain mildew nondestructive test method
CN108399366A (en) * 2018-01-30 2018-08-14 何德珍 It is a kind of based on the remote sensing images scene classification extracting method classified pixel-by-pixel
CN109827908A (en) * 2018-12-24 2019-05-31 核工业北京地质研究院 A method of red fuji apple rotten level is differentiated using spectroscopic data
CN110377798A (en) * 2019-06-12 2019-10-25 成都理工大学 Outlier detection method based on angle entropy
CN111060471A (en) * 2019-11-28 2020-04-24 北京农业信息技术研究中心 Transmission imaging detection method for early rotting of citrus
CN111562273A (en) * 2020-06-05 2020-08-21 大连工业大学 Hyperspectrum-based fish water jet descaling slight damage visualization method
CN114359539A (en) * 2022-03-17 2022-04-15 广东省农业科学院农业质量标准与监测技术研究所 Intelligent identification method for high-spectrum image of parasite in sashimi
CN118071739A (en) * 2024-04-18 2024-05-24 山东北宏新材料科技有限公司 Masterbatch coloring visual detection method based on image enhancement
CN118071739B (en) * 2024-04-18 2024-06-28 山东北宏新材料科技有限公司 Masterbatch coloring visual detection method based on image enhancement

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN201041547Y (en) * 2007-05-29 2008-03-26 浙江大学 Device for analyzing and recognizing different types of green teas based on multi-spectrum image texture
CN101936902A (en) * 2010-08-02 2011-01-05 昆明理工大学 Method for identification and content determination of walnut oil

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN201041547Y (en) * 2007-05-29 2008-03-26 浙江大学 Device for analyzing and recognizing different types of green teas based on multi-spectrum image texture
CN101936902A (en) * 2010-08-02 2011-01-05 昆明理工大学 Method for identification and content determination of walnut oil

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
D.UNAY ET AL: "Automatic defect segmentation of‘Jonagold’ apples on multi-spectral images: A comparative study", 《POSTHARVEST BIOLOGY AND TECHNOLOGY》, vol. 42, 6 October 2006 (2006-10-06) *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103149214B (en) * 2013-02-27 2015-04-01 上海交通大学 Method for detecting flaw on surface of fruit
CN103149214A (en) * 2013-02-27 2013-06-12 上海交通大学 Method for detecting flaw on surface of fruit
CN105891229A (en) * 2014-09-05 2016-08-24 熊菊莲 Method for determining characteristic wavelength for spectral image analysis and detection of surfaces of fruits
CN105891230A (en) * 2014-09-05 2016-08-24 熊菊莲 Fruit appearance detection method based on spectral image analysis
CN105975966A (en) * 2016-04-21 2016-09-28 南京农业大学 Rice grain mildew nondestructive test method
CN108399366A (en) * 2018-01-30 2018-08-14 何德珍 It is a kind of based on the remote sensing images scene classification extracting method classified pixel-by-pixel
CN109827908B (en) * 2018-12-24 2021-04-13 核工业北京地质研究院 Method for judging rot degree of Fuji apples by using spectral data
CN109827908A (en) * 2018-12-24 2019-05-31 核工业北京地质研究院 A method of red fuji apple rotten level is differentiated using spectroscopic data
CN110377798B (en) * 2019-06-12 2022-10-21 成都理工大学 Outlier detection method based on angle entropy
CN110377798A (en) * 2019-06-12 2019-10-25 成都理工大学 Outlier detection method based on angle entropy
CN111060471A (en) * 2019-11-28 2020-04-24 北京农业信息技术研究中心 Transmission imaging detection method for early rotting of citrus
CN111562273A (en) * 2020-06-05 2020-08-21 大连工业大学 Hyperspectrum-based fish water jet descaling slight damage visualization method
CN114359539A (en) * 2022-03-17 2022-04-15 广东省农业科学院农业质量标准与监测技术研究所 Intelligent identification method for high-spectrum image of parasite in sashimi
US11727669B1 (en) 2022-03-17 2023-08-15 Institute of Facility Agriculture, Guangdong Academy of Agricultural Science Intelligent recognition method of hyperspectral image of parasites in raw fish
CN118071739A (en) * 2024-04-18 2024-05-24 山东北宏新材料科技有限公司 Masterbatch coloring visual detection method based on image enhancement
CN118071739B (en) * 2024-04-18 2024-06-28 山东北宏新材料科技有限公司 Masterbatch coloring visual detection method based on image enhancement

Also Published As

Publication number Publication date
CN102890092B (en) 2014-12-17

Similar Documents

Publication Publication Date Title
CN102890092B (en) Characteristic angle cosine value method for detecting defects of honey peach brown rot
Li et al. Multispectral detection of skin defects of bi-colored peaches based on vis–NIR hyperspectral imaging
US11079334B2 (en) Food inspection apparatus, method of inspecting food, and method of learning in identification unit of food inspection apparatus
Sahu et al. Defect identification and maturity detection of mango fruits using image analysis
CN101832941B (en) Fruit quality evaluation device based on multispectral image
Arjenaki et al. Online tomato sorting based on shape, maturity, size, and surface defects using machine vision
Makky et al. Development of an automatic grading machine for oil palm fresh fruits bunches (FFBs) based on machine vision
US9527115B2 (en) Computer vision and machine learning software for grading and sorting plants
CN102288613B (en) Surface defect detecting method for fusing grey and depth information
Bennedsen et al. Performance of a system for apple surface defect identification in near-infrared images
CN103914708B (en) Food kind detection method based on machine vision and system
Sahu et al. Identification and classification of mango fruits using image processing
CN107610114A (en) Optical satellite remote sensing image cloud snow mist detection method based on SVMs
US10596603B2 (en) Method of producing gluten free oats
CN106442561A (en) Online image acquisition system for detecting citrus skin defects and method
CN107505325B (en) Omnibearing quality detection method for winter jujube fruits
Rozario et al. Segmentation of the region of defects in fruits and vegetables
CN109115719A (en) A kind of Citrus Huanglongbing pathogen Band fusion rapid detection method based on high light spectrum image-forming technology
CN103528967A (en) Hyperspectral image based overripe Lonicera edulis fruit identification method
CN105891229A (en) Method for determining characteristic wavelength for spectral image analysis and detection of surfaces of fruits
CN111122590A (en) Ceramic surface defect detection device and detection method
CN106568784A (en) Multispectral imaging system and implementation method which are used for fruit and vegetable surface defect on-line detection
CN109883967A (en) A kind of Eriocheir sinensis quality grade method of discrimination based on information fusion
CN105300895B (en) A kind of method using characteristic point tangent line angle early warning potato sprouting defect
Sola-Guirado et al. A smart system for the automatic evaluation of green olives visual quality in the field

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
C41 Transfer of patent application or patent right or utility model
TR01 Transfer of patent right

Effective date of registration: 20160721

Address after: 321300 Zhejiang city of Jinhua province Yongkang City West West Industrial Zone Lantian Road No. 16

Patentee after: Xinduo Group Co.,Ltd.

Address before: 310027 Hangzhou, Zhejiang Province, Xihu District, Zhejiang Road, No. 38, No.

Patentee before: Zhejiang University

TR01 Transfer of patent right

Effective date of registration: 20200717

Address after: 321300 No. 199 Huacheng East Road, Xicheng Street, Yongkang City, Jinhua City, Zhejiang Province

Patentee after: ZHEJIANG DEFEILUO INTELLIGENT MACHINERY MANUFACTURING Co.,Ltd.

Address before: 321300, Lantian Road, Xicheng Industrial Zone, Xicheng County, Jinhua, Zhejiang, Yongkang 16

Patentee before: SIMTO GROUP Co.,Ltd.

TR01 Transfer of patent right
PE01 Entry into force of the registration of the contract for pledge of patent right

Denomination of invention: Method for detecting brown rot defect of honey peach by cosine value of characteristic angle

Effective date of registration: 20220413

Granted publication date: 20141217

Pledgee: Yongkang branch of the Industrial Commercial Bank of China Ltd.

Pledgor: ZHEJIANG DEFEILUO INTELLIGENT MACHINERY MANUFACTURING Co.,Ltd.

Registration number: Y2022330000503

PE01 Entry into force of the registration of the contract for pledge of patent right