CN202614694U - Recognition device for digital edible mushroom impurities - Google Patents
Recognition device for digital edible mushroom impurities Download PDFInfo
- Publication number
- CN202614694U CN202614694U CN 201120515732 CN201120515732U CN202614694U CN 202614694 U CN202614694 U CN 202614694U CN 201120515732 CN201120515732 CN 201120515732 CN 201120515732 U CN201120515732 U CN 201120515732U CN 202614694 U CN202614694 U CN 202614694U
- Authority
- CN
- China
- Prior art keywords
- edible fungi
- digital
- image
- impurities
- edible mushroom
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
The utility model provides a recognition device for digital edible mushroom impurities. The recognition device for the digital edible mushroom impurities comprises an edible mushroom cleaning production line, wherein a lighting device and a digital camera are arranged in the edible mushroom cleaning production line, the recognition device of the digital edible mushroom impurities also comprises an image collecting card and an industrial personal computer, and the digital camera delivers shot image into the industrial personal computer through the image collecting card for image processing and impurities recognizing. Recognition method adopts machine vision technology, so that various impurities in edible mushroom such as hair, silk wadding shell, woven bag silk, and mulch fragment are recognized and detected.
Description
Technical field
The application relates to a kind of agricultural product impurity ONLINE RECOGNITION device, particularly a kind of digital edible fungi impurities identification device that has adopted machine vision technique.
Background technology
In farming industry, from raw-material classification, screen occasions such as baking and banking up with earth the process, impurity monitoring, finished product portioning, end product quality detection, the application of machine vision technique is arranged all.Agricultural product are of a great variety, differ from vegetables, fruit to specifications such as meats, ever-changing, with machine vision technique to this series products classify automatically, detection, analysis and processing be challenging.Tao etc. (1995) have succeeded in developing Merling high-speed high frequency machine vision fruit grading system, and this system has adopted technology such as frequency spectrum enhancing, coloured image analysis, noise filtering and conversion.The processing power of this equipment is 44t/h, has been widely used in the classification of fruit such as apple, orange, peach, tomato at present.According to reports, there is apple more than 50% every year in the U.S. through this device processes, and has been generalized to other countries such as Canada.Zayas (1985) separates them according to the shape facility of soft wheat of the winter of 2 kinds of different cultivars hard wheat, winter from 8 kinds of other article grow wheats.Miller etc. (1989) obtain the surperficial gray level image of peach with colour TV camera in the peach classification research, propose new algorithm and confirm damaged area.Shearer etc. (1990) will justify green pepper and place in the transparent tube, from six its color digital images of angle shot, then rgb value will be converted into three parameters of HSI with colourful CCD video camera, and according to the colour type of H value and frequency distribution judgement circle green pepper thereof, accuracy reaches 96%.(2000) tobacco leaf less to profile such as lift a sail utilizes the neuroid The technique of extracting of tobacco leaf area, girth and qualitative character value such as damaged also to obtain ideal results.The machine vision product that successfully has been used at present farming industry has apparatus of selecting rice color, fruit sorter etc.Apparatus of selecting rice color is mainly distinguished according to the color of selected thing, and the selected beyond the region of objective existence shape of fruit sorter is big, colouring discrimination is also bigger, so technical difficulty is not high.And other also be in conceptual phase basically, genuine productsization still rare.
In the impurity context of detection; The different fiber scavenger of cotton is the NI Vision Builder for Automated Inspection that is successfully applied to textile industry; Its principle mainly is through color differentiating cotton and different fiber; Common video camera can not be distinguished the different fiber approaching with the cotton color, if use the ultraviolet video camera, can also detect the different fiber that contains fluorescence.Chen Wentao (2003) has studied the online foreign body eliminating system of tobacco; This system adopts the HSI color model, is the principal character amount with colourity, and setting threshold is differentiated the pixel of tobacco and foreign matter; The ratio of foreign matter pixel in the statistic unit is carried out secondary discrimination more then.Above-mentioned foreign matter detecting method all is based on color.
More above-mentioned agricultural product hierarchical detection, the detection task of impurity, the technical difficulty that the online foreign matter of edible fungi detects is bigger.The identification difficulty mainly comes from four aspects: 1, profile is irregular, size differs, the position is at random unordered as the edible fungi (mainly containing Asparagus, straw mushroom, dried mushroom etc.) of image background; 2, the shape of exemplary impurity hair and color be prone to the edible fungi image in thalline shade or border obscure; 3, other impurity such as silk floss shell, woven bag silk, mulch film fragment etc. all do not have fixed shape and size basically; 4, edible fungi semi-manufacture water percentage is higher, and is reflective serious.In order to address these problems, in light source design, image acquisition and processing and identification, all need more careful consideration, in order to adapt to the requirement of ONLINE RECOGNITION, need balance to consider between the complexity of recognizer and the recognition speed.Adopt the online foreign matter detection system of edible fungi of Computer Recognition Technology still not have relevant report and product at present both at home and abroad, abroad we do not find identical or close product yet.
The utility model content
The purpose of the utility model is to provide a kind of digital edible fungi impurities identification device; Adopt machine vision technique, can be with the plurality of impurities in the edible fungi, impurities identification such as hair, silk floss shell, woven bag silk, mulch film fragment and detecting for example; Thereby improve edible fungus production efficiency and quality; Reduce labour intensity and cost of labor, reduce Export Risk, help the international image and the market core competitiveness that promote China edible fungi export enterprise.
The utility model provides a kind of digital edible fungi impurities identification device; This device comprises the edible fungi cleaning product line; It is characterized in that: on described edible fungi cleaning product line, lighting device, digital camera are installed; Said device also comprises image pick-up card and industrial computer, and wherein said lighting device is positioned at the top of edible fungi to be identified, and said lighting device can weaken the even interference that brings to image segmentation and identification of shade, reflective, uneven illumination; Digital camera comprises a plurality of cameras; Said digital camera connects image pick-up card through data connecting line, thus with the image of taking through image pick-up card, send into industrial computer and carry out Flame Image Process and impurities identification.
Further, said lighting device is selected the even diffused illumination mode of no shadow, adopts strip light, circline or dome illumination, and installs diffusing panel additional, and adopt the filter of different wave length.
Further; According to the translational speed of the width and the production line of said edible fungi cleaning product line, adjust the focal length and the aperture size of said digital camera, set the distance that distance of camera lens detects target; And the brightness and the light angle of regulating lighting device, so that obtain distinct image;
Further, said digital camera adopts 2/3, and " colored area array CCD is joined 5 mega pixel camera lenses; distance of camera lens detects target 20cm～30cm, and the visual field is 50mm * 50mm, with the speed photographic images of per second 9 frames; every two field picture is 2452 * 2056 pixels, every pixel 8bit.
Further, the movement velocity of said edible fungi cleaning product line is about 24 meters/minute.
Further, industrial computer also is responsible for management, network communication, control function and the printout function of man-machine interface, historical data and view data.
Description of drawings
The Asparagus image that contain hair impurity of Fig. 1 for taking under the fluorescent lighting condition.
Fig. 2 is digital edible fungi impurities identification apparatus structure synoptic diagram.
Fig. 3 is the user interface capabilities figure of digital edible fungi impurities identification device.
Fig. 4 is the recognizer process flow diagram of digital edible fungi impurities identification device.
Fig. 5 gives an example for the recognition result that contains hair impurity in the Asparagus.
Wherein: 1-edible fungi cleaning product line, 2-lighting device, 3-digital camera, 4-image pick-up card, 5-industrial computer.
Embodiment
Edible fungi cleaning product line 1 is sent edible fungi (Asparagus, straw mushroom, flat mushroom etc.) the process ultrasonic cleaning earlier of different size into the digital edible fungi impurities identification device of the utility model after ozone sterilization, the decolouring bleaching.Edible fungi lies in a horizontal plane on the production line, and the production line movement velocity is about 24 meters/minute.
As shown in Figure 2, digital edible fungi impurities identification device also comprises lighting device 2, digital camera 3, image pick-up card 4, industrial computer 5.Wherein lighting device 2 adopts Flame Image Process dedicated illumination light source, and adopts the even diffused illumination mode of no shadow.Digital camera 3 sensors adopt 2/3, and " colored area array CCD is joined 5 mega pixel camera lenses, and distance of camera lens detects target 20cm～30cm, and the visual field is 50mm * 50mm.Can select 2～4 road cameras on demand for use according to the actual production line width.The ccd video camera camera is with the speed photographic images of per second 9 frames, and every two field picture is 2452 * 2056 pixels, and every pixel 8bit, this image send into industrial computer 5 and handle and discern through image pick-up card 4.Industrial computer 5 also is responsible for management, network communication, control function and the printout function of man-machine interface, historical data and view data.
As shown in Figure 3, industrial computer 5 softwares are realized user interactions, system's input and output control, Flame Image Process, analysis, identification, historical data management, report output and printing, network communication function.
Illumination has critical role in machine vision.Design lighting device 2 at first will be selected suitable light source.The purpose of lighting device 2 design be make the illumination of measurement face evenly, intensity stabilization, and reduce shade and the reflective effect in the image as far as possible.Should select not have the even diffused illumination mode of shadow on the general direction, in the practical implementation, can adopt strip light, circline, dome illumination etc., and install diffusing panel additional, and adopt the filter of different wave length.
The utility model has used advanced image processing techniques, and has studied the recognizer to the impurity that often has in the edible fungi.
The step of carrying out impurities identification of the digital edible fungi impurities identification device of the utility model comprises:
(a) weaken shade in the images, reflective, disturbing factor such as uneven illumination is even through lighting device 2;
(b) use digital camera 3 to obtain the image of impure edible fungi;
(c) carry out color space transformation, figure image intensifying, image filtering etc. and help the image pre-service handling and discern;
(d) applying edge detection and movable contour model carry out the Region Segmentation of foreign matter in the edible fungi image;
(e) utilize area informations such as geometric configuration, marginal information, color, texture, extract the characteristic of edible fungi and exemplary impurity;
(f) adopt above-mentioned steps (a)-(e) to carry out the off-line sample training, set up the mixed Gauss model of representing impurity, with information such as the edge feature of obtaining impurity, provincial characteristicss;
(g) to above-mentioned off-line sample, the neuroid contrast model of structure edible fungi shadow region makes system have the separating capacity to shade and foreign matter;
(f) adopt above-mentioned steps (a)-(e) to analyzing through the online sample of the edible fungi of on edible fungi cleaning product line 1, moving after ultrasonic cleaning, ozone sterilization, the decolouring bleaching with certain speed; The mixed Gauss model that information such as the edge feature of the online sample that obtains, provincial characteristics and above-mentioned off-line sample training are obtained matees; The shade probability that the matching probability that obtains combines above-mentioned neuroid to obtain adopts the judgement of many characteristics fuzzy reasoning that inclusion-free is arranged at last.
The recognizer process flow diagram of the digital edible fungi impurities identification device of the utility model is as shown in Figure 4.At first be image processing process, the original color image that ccd video camera obtains belongs to rgb color space, does not meet human eye vision hue distinguishes characteristic, can it be transformed into HSI or YUV color space, extracts to make things convenient for color characteristic.Directly adopt color image, because dimension is high, data volume is big, makes troubles for analysis and search, earlier coloured image is converted into gray level image and carries out pre-service usually.The image pre-service comprises that gray scale normalization processing, linear filtering or nonlinear filtering carry out the figure image intensifying, to remove useless details (noise) and to strengthen the details that contains characteristic information.Also can use one of them component in the color space, carry out the Shape Feature Extraction experiment like the U component of yuv space.
Marginal information is one type of important bottom layer image characteristic information.The utility model adopts the multi-scale edge method for distilling to extract marginal information, and this method can adopt instruments such as edge detection operator, wavelet analysis, from shade of gray information, extracts edge strength, direction and edge type information.For example, adopt the wavelet analysis instrument, because the multiple dimensioned characteristic that wavelet transformation has; The wavelet transformation of each yardstick of image all provides certain marginal information, and when yardstick hour, the edge of image detailed information is than horn of plenty; Edge precision is higher, but is subject to interference of noise; During large scale, edge of image is stable, and antinoise is good, but bearing accuracy is poor, and the result of the edge image of each yardstick is integrated, and brings into play the advantage of big small scale, just can obtain accurate edge image information.For example for impurity such as hairs; Because hair has faciola shape characteristic; And hair is thinner than edible fungi, and hair diameter is approximately 0.07 millimeter, according to two nearer relatively parallel edges of distance; The axis of hair impurity can be located soon, curvature information, the smoothness of axis can be obtained in view of the above.The utilization of this heuristic information makes the hunting zone of hair impurity reduce greatly, for the identification of follow-up hair provides advantage.
According to actual impurity image library, comprehensively go out a cover impurity template, adopt matched filtering again, tentatively definite possibly be the center in the zone of impurity.Use movable contour model (snake model) then, obtain the TP of impurity.After having obtained to be the image-region of impurity; Can further extract based on this regional advanced features, like shape facility, color characteristic, textural characteristics etc., for example the extraction of color characteristic is to the impurity of colour; Like mulch film, the identification of glass rope etc. has vital role.
Then, to the various impurity that often have in the edible fungi, for example; Hair, silk floss shell, woven bag silk, mulch film fragment etc.; Adopt above-mentioned image processing techniques to carry out the off-line sample training, set up the mixed Gauss model of representing impurity, wherein comprise information such as edge feature, provincial characteristics.In addition; For mushroom edge, particulate contamination and the mushroom shade of distinguishing hair and wire; The utility model has trained a neural network model as the shade contrast model with the mushroom shadow image; This neural network model output valve is high more, shows that then alternative area is that the probability of edible fungi image shade is high more.Further; For online sample; On the basis of above-mentioned Flame Image Process, based on information such as region shape characteristic, field color characteristic and the middle paraxial curvature of the online sample that obtains, smoothnesss, the mixed Gauss model that these information and the above-mentioned off-line sample training of online sample obtained matees; The shade probability that the matching probability that obtains combines neuroid to obtain adopts the judgement of many characteristics fuzzy reasoning that inclusion-free is arranged at last.
The present discrimination that in Asparagus, contains under the black hair impurity situation according to the digital edible fungi impurities identification device of the utility model has reached 95%, and Fig. 5 is a routine recognition result, and its pairing original image is Fig. 1.
China's Edible Fungi spreads all over the country, and Fujian, Zhejiang, Hubei, Shandong, Henan, Sichuan and Yunnan etc. are economized and are the main producing region.China also is the first in the world canned mushroom producing country and exported country, and according to Chinese customs statistics, before 2007 10 months, China's canned mushroom was exported goods and earned foreign currency 3.7 hundred million dollars; And 2008 and edible fungi of china export trade in 2009 encounter setbacks, and wherein one of reason is that food safety requirements has been improved in the international market.Therefore Edible Fungi presses for the production facility and the robotization quality detection apparatus of robotization.Under so urgent situation, foreign matter ONLINE RECOGNITION device has good industrial prospect and social benefit in the edible fungi process in the utility model.
Utilize the digital edible fungi impurities identification device of the employing machine vision technique of the utility model; Solved this difficult problem of foreign matter ONLINE RECOGNITION in the Edible Fungi process; Further target is the application of development machines vision technique in the quality monitoring of processing of farm products industry, and promotes.Can carry out combination, reduction and the optimization of hardware and software of system to concrete the application, being applicable to various application such as food service industry, agricultural product detection, so the technical application of the utility model and industrialization prospect are boundless.
Claims (6)
1. digital edible fungi impurities identification device, this device comprises the edible fungi cleaning product line, it is characterized in that:
On described edible fungi cleaning product line, lighting device, digital camera are installed; Said device also comprises image pick-up card and industrial computer; Wherein said lighting device is positioned at the top of edible fungi to be identified; Said lighting device can weaken shade, the even interference of bringing for image segmentation and identification of reflective, uneven illumination, and digital camera comprises a plurality of cameras, and said digital camera is through data connecting line connection image pick-up card; Thereby the image of taking through image pick-up card, is sent into industrial computer and carried out Flame Image Process and impurities identification.
2. digital edible fungi impurities identification device as claimed in claim 1; It is characterized in that: said lighting device is selected the even diffused illumination mode of no shadow; Adopt strip light, circline or dome illumination, and install diffusing panel additional, and adopt the filter of different wave length.
3. digital edible fungi impurities identification device as claimed in claim 1; It is characterized in that translational speed according to the width and the production line of said edible fungi cleaning product line; Adjust the focal length and the aperture size of said digital camera; Set the distance that distance of camera lens detects target, and regulate the brightness and the light angle of lighting device, so that obtain distinct image.
4. digital edible fungi impurities identification device as claimed in claim 1; " colored area array CCD is joined 5 mega pixel camera lenses, distance of camera lens detection target 20cm～30cm; the visual field is 50mm * 50mm; with the speed photographic images of per second 9 frames, and every two field picture is 2452 * 2056 pixels, every pixel 8bit to it is characterized in that said digital camera adopts 2/3.
5. digital edible fungi impurities identification device as claimed in claim 1, the movement velocity that it is characterized in that said edible fungi cleaning product line is about 24 meters/minute.
6. a digital edible fungi impurities identification device as claimed in claim 1 is characterized in that industrial computer also management, network communication, control function and the printout function of responsible man-machine interface, historical data and view data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201120515732 CN202614694U (en) | 2011-12-06 | 2011-12-06 | Recognition device for digital edible mushroom impurities |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201120515732 CN202614694U (en) | 2011-12-06 | 2011-12-06 | Recognition device for digital edible mushroom impurities |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN202614694U true CN202614694U (en) | 2012-12-19 |
Family
ID=47348383
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201120515732 Expired - Fee Related CN202614694U (en) | 2011-12-06 | 2011-12-06 | Recognition device for digital edible mushroom impurities |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN202614694U (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10902577B2 (en) | 2017-06-19 | 2021-01-26 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify object |
| US10902581B2 (en) | 2017-06-19 | 2021-01-26 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify foreign object |
-
2011
- 2011-12-06 CN CN 201120515732 patent/CN202614694U/en not_active Expired - Fee Related
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10902577B2 (en) | 2017-06-19 | 2021-01-26 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify object |
| US10902581B2 (en) | 2017-06-19 | 2021-01-26 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify foreign object |
| US11410295B2 (en) | 2017-06-19 | 2022-08-09 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify foreign object |
| US11443417B2 (en) | 2017-06-19 | 2022-09-13 | Apeel Technology, Inc. | System and method for hyperspectral image processing to identify object |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102495067B (en) | System for identifying impurities of edible funguses on line | |
| CN107486415A (en) | Thin bamboo strip defect on-line detecting system and detection method based on machine vision | |
| CN105181714B (en) | A kind of bottling liquid Impurity Detection System and its detection method based on image procossing | |
| CN102135416B (en) | Online image detecting system and method for bottle covers | |
| CN102519971A (en) | On-line identification apparatus and method for impurities in edible fungi | |
| CN102982350B (en) | A kind of station caption detection method based on color and histogram of gradients | |
| CN103034838B (en) | A kind of special vehicle instrument type identification based on characteristics of image and scaling method | |
| CN110276386A (en) | A kind of apple grading method and system based on machine vision | |
| Sahu et al. | Identification and classification of mango fruits using image processing | |
| CN102521564A (en) | Method for identifying tea leaves based on colors and shapes | |
| CN101984346A (en) | Method of detecting fruit surface defect based on low pass filter | |
| CN207238542U (en) | A kind of thin bamboo strip defect on-line detecting system based on machine vision | |
| CN102601063B (en) | Automatic identifying and grading method for bamboo chips | |
| CN103440035A (en) | Gesture recognition system in three-dimensional space and recognition method thereof | |
| CN104048966B (en) | The detection of a kind of fabric defect based on big law and sorting technique | |
| CN106370667A (en) | Visual detection apparatus and method for quality of corn kernel | |
| CN108154195A (en) | Tealeaves recognition methods and the tealeaves sorting equipment using this method | |
| CN112044808A (en) | Household garbage recognition system | |
| CN109509173A (en) | A kind of method of counting and counting device of bacterial clump | |
| CN103543107A (en) | Intelligent classification system and method for tobacco leaves based on machine vision and hyperspectral technology | |
| CN102788806B (en) | Fruit peel defect detection method based on spheroidic brightness transformation | |
| CN103528967A (en) | Hyperspectral image based overripe Lonicera edulis fruit identification method | |
| CN102507008B (en) | Multi-template automatic optical color detection method | |
| CN115862004A (en) | Corn ear surface defect detection method and device | |
| CN202614694U (en) | Recognition device for digital edible mushroom impurities |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121219 Termination date: 20131206 |