CN212515898U - Pistachio head and tail recognition device based on multi-angle optical coding and deep learning - Google Patents
Pistachio head and tail recognition device based on multi-angle optical coding and deep learning Download PDFInfo
- Publication number
- CN212515898U CN212515898U CN202020618517.8U CN202020618517U CN212515898U CN 212515898 U CN212515898 U CN 212515898U CN 202020618517 U CN202020618517 U CN 202020618517U CN 212515898 U CN212515898 U CN 212515898U
- Authority
- CN
- China
- Prior art keywords
- pistachio
- light source
- head
- tail
- rgb
- 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
- 240000006711 Pistacia vera Species 0.000 title claims abstract description 73
- 235000003447 Pistacia vera Nutrition 0.000 title claims abstract description 69
- 235000020233 pistachio Nutrition 0.000 title claims abstract description 41
- 238000013135 deep learning Methods 0.000 title claims description 27
- 230000003287 optical effect Effects 0.000 title description 6
- 239000003086 colorant Substances 0.000 claims abstract description 9
- 230000007704 transition Effects 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims description 12
- 235000014571 nuts Nutrition 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims 1
- 230000007723 transport mechanism Effects 0.000 abstract description 9
- 235000013399 edible fruits Nutrition 0.000 abstract description 8
- 230000015572 biosynthetic process Effects 0.000 abstract description 5
- 238000001514 detection method Methods 0.000 abstract description 5
- 230000008859 change Effects 0.000 abstract description 3
- 238000012549 training Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 5
- 238000012937 correction Methods 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 2
- 230000001788 irregular Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 235000011201 Ginkgo Nutrition 0.000 description 1
- 235000008100 Ginkgo biloba Nutrition 0.000 description 1
- 244000194101 Ginkgo biloba Species 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 230000009955 peripheral mechanism Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The utility model discloses a fruit of pistachio head and tail recognition device based on multi-angle light coding and degree of depth study, the device includes and is provided with transport mechanism and support frame on the supporting platform, has placed the fruit of pistachio on the transport mechanism, is connected with the industry camera on the support frame, and the industry camera is just to transport mechanism, still is connected with RGB three-colour structure light source on the support frame, and it has the coaxial light source still to centre gripping between RGB three-colour structure light source and the industry camera. The utility model discloses a detection device adopts three-colour RGB light and coaxial light, carry out the multi-angle formation of image to the object, because there is obvious difference in the head and the tail transition curved surface of pistachio nut and the change angle of profile skeleton, based on coaxial white light source, the light source of four kinds of different colours of RGB three-colour structure light source, form images the pistachio nut tip from the angle of difference, the transition curved surface of profile that the pistachio nut is different just demonstrates different colours, become the colour coding of two-dimentional after camera formation of image, the discrimination of pistachio nut head and the tail has been strengthened, the recognition rate of network has been improved.
Description
Technical Field
The utility model relates to an automatic optical detection technical field especially relates to a happy fruit head and tail recognition device based on multi-angle light coding and degree of depth study.
Background
In the market, the quality price of the natural opening of the pistachio, and the opening of the machine, is not the same. In order to maximize the benefits of the enterprise, it is desirable to initially distinguish the natural opening from the open-end of the machine at the factory. The yield of pistachio nuts is huge every year, and if the pistachio nuts are detected by artificial naked eyes, the efficiency is low and the cost is high. The needs of enterprises cannot be met. Therefore, there is an urgent need for an automatic and efficient intelligent product to replace manual work for the opening recognition of pistachios. The identification of the opening is needed to be realized firstly.
SUMMERY OF THE UTILITY MODEL
The embodiment of the utility model provides a happy fruit head and tail recognition device based on multi-angle light coding and degree of depth study can discern the head and the tail of happy. Because the shape of the pistachio nuts is naturally formed, is multifaceted and complex, and is mostly a plurality of relatively complex and irregular curved surfaces, enterprises handling the pistachio nuts want to recognize the heads and the tails of the pistachio nuts so as to open the heads of the pistachio nuts, but the enterprises cannot start from the beginning to the end. The utility model provides an imaging and detection device of the pistachio nut head and tail of constituteing by industry camera, white coaxial light, RGB three-colour light source etc. carries out the illumination of multi-angle to the pistachio nut physique to encode, carry out the network training with unique degree of deep learning algorithm to information at last, thereby the head and the tail of automatic identification pistachio nut. The problem of pistachio nut food processing enterprise carry out quick accurate head and the tail automatic identification to pistachio nut, overcome the head opening, equipment is difficult to the location is solved.
The utility model discloses the mechanical structure technical scheme who adopts as follows:
pistachio nut head and tail recognition device based on multi-angle light coding and deep learning comprises: supporting platform, the last transport mechanism V type groove that is provided with of supporting platform, be the open heart fruit that the flat was lain and is placed in the V type groove, supporting platform on have a support frame, be connected with the industry camera on the support frame, the industry camera is just right transport mechanism, still install RGB three-colour structured light on the support frame, structured light source is located under the industry camera, still installs the coaxial light source between RGB structured light and the industry camera.
Further, the supporting clamp is connected with a clamp, and the clamp is used for clamping the industrial camera.
Further, the fixture is also used for clamping the RGB three-color structured light source and the coaxial light source, and the coaxial light source is positioned between the RGB structured light and the industrial camera.
Further, the conveying mechanism comprises a V-shaped conveying guide rail for conveying the pistachios.
Furthermore, the support frame is fastened on the support platform by bolts.
The utility model has the advantages that: the intelligent degree that present detection device exists is low, artifical excessively take in, production efficiency, the complicated scheduling problem of structure have been solved to this device, and this device has characteristics such as camera lens dismantlement is simple, work efficiency height, can realize the quick accurate detection of the head and the tail direction of pistachio nut. The pistachio nuts are relatively complex and irregular curved surfaces in shape. It is difficult to perform feature extraction and classification on its surface. The utility model discloses a pistachio nut head and tail recognition device based on multi-angle light coding and degree of depth study has adopted RGB three-colour light and white coaxial light to form images to the pistachio nut, R, G, B encodes the curved surface of different angles with the photochromic light of four kinds of different colours that white constitutes, white coaxial light encodes the planar part, R, G, B's coding order is from high angle to low angle, the transition process that corresponds pistachio nut surface transition curved surface by flat area to precipitous area. Because there is obvious difference in the head and the tail transition curved surface of pistachio nut and the change angle of outline skeleton, the utility model discloses the device adopts coaxial white light source, and RGB three-colour structure light source detects pistachio nut, based on the light source of above-mentioned four kinds of different colours, images the pistachio nut from the angle of difference, and the transition curved surface of outline that the pistachio nut is different just demonstrates different colours, becomes the colour code of two-dimentional after camera formation of image. The adoption of four colors of light for imaging at different angles enhances the discrimination of the head and the tail of the pistachio nuts, improves the recognition rate of a network, reduces the rejection rate in the pistachio nut processing and improves the production benefit.
Drawings
Fig. 1 is a schematic view of the mechanical structure assembly of a device for recognizing the head and tail of an pistachio based on multi-angle optical coding and deep learning of the present invention;
FIG. 2 is a schematic diagram of an optical path structure for optical imaging;
FIG. 3 is a flow chart of a method of image processing;
FIG. 4 is a diagram of a deep learning network architecture;
FIG. 5 is a schematic top elevational view of a pistachio nut
FIG. 6 is a schematic view of the head of a pistachio nut;
fig. 7 is a schematic view of the tail of a pistachio.
Description of reference numerals: the device comprises a supporting platform 1, a supporting frame 2, an industrial camera 3, a clamp 4, a coaxial light source 5, an RGB three-color structure light source 6 and a pistachio nut 7; a transport mechanism 8.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments in the present invention, all other embodiments obtained by a person skilled in the art without creative efforts belong to the protection scope of the present invention.
The utility model provides a pistachio nut head and tail recognition device structure assembly sketch map based on multi-angle light coding and degree of depth study, it is shown in figure 1, including supporting platform 1, be provided with transport mechanism 8 on supporting platform 1, pistachio nut 7 is located transport mechanism 8, and supporting platform 1 is connected with support frame 2, is connected with industrial camera 3 on the support frame 2, and industrial camera 3 is just to transport mechanism, still is connected with RGB three-colour structure light source 5 on the support frame, and RGB three-colour structure light source 6 is located 3 belows of industrial camera. A coaxial light source 5 is clamped between the RGB three-color structure light source 6 and the industrial camera 3. The on-axis light source 5 may comprise a 45 ° beam splitter parallel to the support platform 1 and a white light source illuminating the 45 ° beam splitter in a direction parallel to the support platform 1. The RGB three-color structure light source 6 can provide a good imaging environment for a vision system to realize RGB coding, and the coaxial light source 5 is used as an auxiliary structure light source to achieve a better imaging environment to realize W coding. Adopt three-colour RGB structured light to encode the curved surface of different angles, coaxial light encodes the plane part, realizes that the multi-angle is stable even formation of image.
The support frame 2 is connected with a clamp 4, and the clamp 4 is respectively used for clamping an industrial camera 3, an RGB three-color structure light source 6 and a coaxial light source 5. The working principle of the clamp is as follows: the clamp is mainly divided into two parts which are connected through the threaded holes communicated with each other at the two sides of the clamp, and preferably, the clamp can also be made of a material with certain deformation strength, so that the light source environment can be better adjusted. Since the position adjustment structure is a common knowledge in the art, the present invention is not described herein, and the addition and deletion of components and functions can be performed according to actual needs.
The support frame 2 is clamped on the support platform 1 through a clamp, and the position of the support frame 2 is convenient to replace.
The utility model discloses a light path mechanism schematic diagram based on multi-angle light coding and degree of depth study pistachio head and tail recognition device, as shown in fig. 2, mainly by industry camera 3, coaxial light source 5, RGB three-colour structure light source 6 constitutes, because pistachio exhibits more complicated curved surface, during the formation of image, more special pistachio is difficult to extract shape characteristic a bit. The utility model discloses a fruit of pistachio head and tail recognition device based on multi-angle light coding and degree of depth study has adopted three-colour RGB structured light to encode the fruit of pistachio curved surface of different angles, and is shown by the figure, and the characteristic of different angles is 0 degree respectively < theta 1< theta 2< theta 3< 90 degrees, and theta 4 is 90 degrees, as shown in formula 1. The coaxial light source encodes the flat plane part, multi-angle color coding of the pistachio (the specific form of the pistachio is shown in the attached figures 5-7) is realized, and then the head and the tail of the pistachio are identified by applying a deep learning algorithm.
The invention aims at a method flow chart of a method for recognizing the beginning and the end of a happy fruit based on multi-angle optical coding and deep learning, and the method is shown in figure 3. The flow chart mainly consists of three major parts: image preprocessing, target inclination correction and a deep learning network. The image preprocessing comprises the following steps: acquiring an image (S104), analyzing the image (S105), and extracting ROI (S106); the target tilt correction includes: edge extraction (S107), ellipse fitting (S108), ellipse major axis inclination angle (S109), affine change (S110); the deep learning network comprises: normalization processing (S111), inputting a deep learning network (S112), and outputting a result (S113). The method comprises the following specific steps:
the method comprises the following steps: firstly, an industrial camera is used for collecting images of pistachio nuts in multiple angles with RGB structure light and coaxial light (S104);
step two: and (S105) carrying out image analysis on the image acquired in the step one (S105), converting the image into a gray-scale image, then carrying out denoising processing on the image by using a low-pass filter smoothing image (9 x 9 kernel), and carrying out binarization on the image to extract an interested ROI (S106).
Step three: and (5) performing edge extraction on the ROI processed in the step two by using a Canny method (S107), performing least square ellipse fitting on the image (S108), finding a target ellipse fitting long axis inclination angle (S109), and performing inclination correction on the image.
Step four: affine transformation S110 is performed on the image subjected to the inclination correction in step three, and the image is subjected to normalization processing S111.
Step five: inputting the preprocessed image into deep learning network (LeNet, RCNN, VGG16, M0Bilenet, etc.) S112, and determining the head and tail of the pistachio S114
Step six: the result output by the deep learning network is output to the peripheral mechanism driving mechanism, and the state of the pistachio is adjusted S115.
The invention is directed to a deep learning flow chart of a ginkgo head and tail recognition method based on optical coding and deep learning, and the deep learning flow chart is shown in fig. 4. The method comprises the following specific steps:
the method comprises the following steps: before deep learning is started, a data set needing to be learned by a machine is prepared, wherein a data set of pistachio nuts is prepared (S116);
step two: inputting the prepared data into a deep learning network, and manually marking the head and the tail of the pistachio during training (manually marking S117);
step three: after the samples are labeled, the augmented data set S118 can be performed based on the labeled samples, the images can be turned over or randomly cut, the training sample set can be added, and only under the condition that the training sample set is large and correct, the designed deep learning network can more accurately identify the test sample set.
Step four: after a sample set is prepared, dividing the prepared sample into two parts, wherein one part is a training set S119, giving priority to learning of a machine, and judging affairs according to the weight of the occurrence times of data features; some of the tests are performed in step S120 to check the result of deep learning. Wherein, the proportion design of training and test set sample quantity is 9: 1;
step five: after the data sets are divided, the machine firstly learns the training samples S121, and the weight of the characteristics of the pistachio is automatically mastered;
step six: when training is complete, the ten percent untrained test samples are tested for training outcomes using the trained "pistachio" feature weights S122. And calculating the accuracy of the whole data of the test sample. When the accuracy of the data meets the requirement, the current weight can be used to automatically identify the next new pistachio sample image S123.
The above examples are merely illustrative for clearly illustrating the present invention and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (8)
1. The utility model provides a pistachio nut head and tail recognition device based on multi-angle light coding and deep learning which characterized in that includes:
the supporting platform is provided with a conveying mechanism; the conveying mechanism comprises a V-shaped groove for placing pistachio nuts;
the supporting platform is provided with a supporting frame, and the supporting frame is connected with the industrial camera; the industrial camera is opposite to the conveying mechanism;
still be connected with RGB three-colour structure light source on the support frame, RGB three-colour structure light source is located industry camera below, RGB three-colour structure light source with still be fixed with coaxial light source between the industry camera.
2. The pistachio nut head and tail recognition device based on multi-angle light coding and deep learning as claimed in claim 1, wherein a clamp is connected to the support frame, and the clamp is used for clamping the industrial camera.
3. The device for pistachio nut head and tail recognition based on multi-angle light coding and deep learning of claim 2, wherein the clamp comprises two clamping pieces connected by threads.
4. The device as claimed in claim 3, wherein the fixture is used for holding the RGB three-color structured light source and the coaxial light source, and the coaxial light source is located between the RGB three-color structured light and an industrial camera.
5. The device for identifying the head and tail of the pistachio nut based on the multi-angle light coding and the deep learning as claimed in claim 1, wherein the conveying mechanism further comprises a V-shaped conveying guide rail for conveying the pistachio nut.
6. The device for recognizing the head and tail of pistachios based on multi-angle light coding and deep learning as claimed in claim 5, wherein the V-shaped conveying rails are used for conveying lying pistachios.
7. The pistachio nut head and tail recognition device based on multi-angle light coding and deep learning of claim 1, wherein the support frame is fixed on the support platform through screws.
8. The device for identifying the head and the tail of the pistachio based on the multi-angle light coding and the deep learning as claimed in claim 7, wherein the pistachio is imaged by adopting RGB three-color structured light and white coaxial light; r, G, B, four lights with different colors and white color are used for coding the pistachio nut curved surface with different angles; white coaxial light encodes the plane part of the pistachio, and the encoding sequence of R, G, B is from high angle to low angle, and corresponds to the conversion process of the transition surface of the pistachio from a flat area to a steep area; the transition curved surfaces with different outlines of the pistachio nuts present different colors and are changed into two-dimensional color codes after being imaged by a camera; because the four colors of light form images at different angles, the discrimination of the head and the tail of the pistachio is enhanced.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202020618517.8U CN212515898U (en) | 2020-04-22 | 2020-04-22 | Pistachio head and tail recognition device based on multi-angle optical coding and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202020618517.8U CN212515898U (en) | 2020-04-22 | 2020-04-22 | Pistachio head and tail recognition device based on multi-angle optical coding and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN212515898U true CN212515898U (en) | 2021-02-09 |
Family
ID=74435558
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202020618517.8U Expired - Fee Related CN212515898U (en) | 2020-04-22 | 2020-04-22 | Pistachio head and tail recognition device based on multi-angle optical coding and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN212515898U (en) |
-
2020
- 2020-04-22 CN CN202020618517.8U patent/CN212515898U/en not_active Expired - Fee Related
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107941808B (en) | 3D printing forming quality detection system and method based on machine vision | |
| CN107944504B (en) | Board recognition and machine learning method and device for board recognition and electronic equipment | |
| CN108805862B (en) | A Tag Identification Method Based on Improved Structural Similarity | |
| CN108491788A (en) | A kind of intelligent extract method and device for financial statement cell | |
| CN110108712A (en) | Multifunctional visual defect detection system | |
| US20180059162A1 (en) | Multi-fiber identification using jacket color | |
| CN111611953B (en) | Target feature training-based oil pumping unit identification method and system | |
| CN114926410A (en) | Method for detecting appearance defects of brake disc | |
| CN112304957A (en) | Machine vision-based intelligent detection method and system for appearance defects | |
| CN118014933B (en) | A defect detection and identification method and device based on image detection | |
| CN106709529B (en) | A visual inspection method for color difference classification of photovoltaic cells | |
| CN117147580A (en) | Visual detection device and method for silicon wafer chamfering edge breakage defect | |
| CN110648366A (en) | Orange detection system and orange sectioning method based on deep learning | |
| CN212515898U (en) | Pistachio head and tail recognition device based on multi-angle optical coding and deep learning | |
| CN109492544B (en) | Method for classifying animal fibers through enhanced optical microscope | |
| CN102609699A (en) | Device and method for recognizing number of cast workpiece scanned by laser | |
| CN114120318B (en) | Dark field image target point accurate extraction method based on integrated decision tree | |
| CN119049054B (en) | Product irregular deformation character detection method and device based on deep learning | |
| CN111583195A (en) | Pistachio nut head and tail recognition device and method based on optical coding and deep learning | |
| Radovan et al. | An approach for automated inspection of wood boards | |
| CN118279652A (en) | Glass sorting method and device based on combination of image color recognition and deep learning | |
| CN114594102B (en) | Machine vision-based data line interface automatic detection method | |
| CN112348772A (en) | Industrial visual detection method based on deep learning | |
| CN112067625A (en) | Device and method for detecting surface defects of rubber ring parting line | |
| CN112215149B (en) | Fitting sorting system and method based on visual detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210209 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |