CN111260609A - Cigarette appearance defect detection method based on deep learning - Google Patents
Cigarette appearance defect detection method based on deep learning Download PDFInfo
- Publication number
- CN111260609A CN111260609A CN202010018188.8A CN202010018188A CN111260609A CN 111260609 A CN111260609 A CN 111260609A CN 202010018188 A CN202010018188 A CN 202010018188A CN 111260609 A CN111260609 A CN 111260609A
- Authority
- CN
- China
- Prior art keywords
- image
- cigarette
- defect
- deep learning
- images
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a cigarette appearance defect detection method based on deep learning, which comprises the following steps: s, acquiring a front image and a back image of the cigarette through a camera assembly; carrying out histogram equalization processing on the acquired image; carrying out median filtering processing on the processed image; cutting the processed image in proportion, establishing a gray template based on an image pyramid, and correcting the image position according to the image position and the direction; combining the corrected upper image and the corrected lower image into one image; cutting the processed image into three images; respectively generating masks in background areas of the cigarette holder segment image, the LOGO segment image and the cigarette body segment image, and respectively taking a non-mask area as an interested area 1-3; establishing a model data set by using the region of interest sample set: and establishing and optimizing a deep learning algorithm model through the model data set, and detecting the defects of the cigarette image by using the deep learning model. The deep learning model used by the invention can accurately classify the defects, and can provide information for manufacturers to remove mechanical faults and improve production schemes.
Description
Technical Field
The invention relates to cigarette detection, in particular to a cigarette appearance defect detection method based on deep learning.
Background
1. The existing cigarette appearance online detection system based on the machine vision technology uses an area array CCD camera, the acquired cigarette end face image is two-dimensional plane information, and the defect detection of end face cavity depth (three-dimensional information) cannot be realized.
2. The existing cigarette appearance online detection system based on the machine vision technology can only reduce the interference of the droplets to the droplet removal method, but cannot identify the droplets and remove the interference. The existing method for eliminating the interference of the droplets is to install two cameras with different shooting angles on an optical module, collect two images of the same circumference of the cigarette, and compare the defect positions of the two images to eliminate the interference of the droplets. However, when the droplets are in a position where the fields of view of the two cameras overlap, the droplets may appear on both images simultaneously, causing erroneous judgment.
3. The existing cigarette appearance online detection system based on the machine vision technology uses an image comparison algorithm to detect defects, the position of each cigarette at a roller is different, the position of the collected image with logo or pattern and other patterns can also change, and the cigarette with the pattern detected by using the comparison algorithm is easy to cause misjudgment, so the existing comparison algorithm can not realize the defect detection of the cigarette area with the pattern.
Disclosure of Invention
The invention aims to solve the technical problems on the market that: provides a cigarette appearance defect detection method based on deep learning.
The technical scheme adopted by the invention for solving the technical problems is as follows: a cigarette appearance defect detection method based on deep learning comprises the following steps:
s1: when the cigarette passes through the phase camera, a front image and a back image of the cigarette are acquired by camera assemblies which are oppositely arranged on two sides of the cigarette making machine, the front image or the back image is composed of two side images, and the intersection of the two side images comprises a 180-degree circumferential area of the cigarette; the intersection of the front image and the back image comprises a 360-degree circumferential surface area of the cigarette;
s2: performing histogram equalization processing on the image acquired in the step S1;
s3: performing median filtering processing on the image processed in the step S2;
s4: the image processed in S3 is displayed at a height of 1: scale-out of 1 into an upper image and a lower image, and performing S5 and S6 processing on the two images, respectively;
s5: establishing a gray level template based on the image pyramid, and acquiring the position and the direction of the cigarette image by utilizing the template matching based on the gray level template;
s6: carrying out affine transformation according to the image position and the direction of S5 to correct the image position;
s7: combining the corrected upper image and the corrected lower image into one image;
s8: cutting the image processed by the S7 into three images according to the non-character cigarette holder area, the character cigarette holder area and the cigarette body area, wherein the three images are respectively a cigarette holder section image, a LOGO section image and a cigarette body section image;
s9: respectively generating masks in background areas of the cigarette holder segment image, the LOGO segment image and the cigarette body segment image, wherein a non-mask area of the cigarette holder segment image is used as an interested area 1, a non-mask area of the LOGO segment image is used as an interested area 2, and a non-mask area of the cigarette body segment image is used as an interested area 3;
s10: establishing a model data set by using an interested area sample set, wherein the interested area sample set is divided into a training set and a prediction set, and the training set and the prediction set jointly form the model data set; wherein the sample set of the region of interest 1 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 2 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 3 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image, an entrainment defect image;
s11: a deep learning algorithm model is established and optimized through the model data set,
s12: and detecting the defects of the cigarette image by using the deep learning model.
Further, the deep learning algorithm model in S11 is VGG16, and the sample sets of the three regions of interest are respectively brought into the model for training.
Further, the training process of the VGG16 model is as follows:
①, taking a sample (Xp, Yp) from the sample set, inputting the Xp into the network, wherein Xp is the p-th input sample, and Yp is the ideal output result;
in this process, the network performs the calculation of Op ═ Fn (… (F2(F1(XpW (1)) W (2)) …) W (n)), where Fn is a calculation function of the nth layer network and W (n) is a weight matrix of the nth layer.
③ calculating the difference between the actual output Op and the corresponding ideal output Yp;
④ the weight matrix is adjusted by back propagation with minimized error.
Further, the defect detection performed in S12 is to take the cigarette image acquired by the machine component in real time into the deep learning algorithm model of iv to obtain an identification result, specifically, a non-defective cigarette image, a droplet interference image, a roundness defect image, a damage defect image, a stain defect image, and a droplet defect image, if the identification result of the deep learning algorithm model is the non-defective cigarette image or the droplet interference image, the image processing control system determines that the cigarette is a good product, otherwise, the image processing control system rejects the cigarette as a bad product.
Further, image processing control module group, including image processing server, collection rejection controller, coding disc, solenoid valve and display, image processing server rejects controller and display electricity with the collection respectively and is connected, gathers and rejects the controller and is connected with coding disc and solenoid valve electricity respectively, still includes image card and IO card in the image processing server, and the image card is connected with image acquisition unit and collection rejection controller electricity respectively, and the IO card is connected with collection rejection controller electricity, the image acquisition unit is four camera subassemblies, and every camera subassembly of group is connected with image card electricity alone.
Further, the image processing control module has the following steps in the image processing process:
step 1: image acquisition: the coding disc inputs the position information of each cigarette to the acquisition and rejection controller, the acquisition and rejection controller converts the position information into a frame signal and sends the frame signal to the image card, the image card sends an acquisition signal to the image acquisition unit, and the image acquisition unit acquires images of two camera assemblies of the same 1 cigarette and then sends the images back to the image processing server to finish image acquisition;
step 2: image processing: the image processing server processes the images of the two camera assemblies to obtain a judgment result;
and step 3: and (3) judging and processing: if the image discrimination results in the step 2 are good products, the image processing control module sends good product statistical information to the display;
if the image discrimination result in the step 2 is a bad product, the image processing server executes two operations: firstly, the I/O card sends information of specifying cigarettes as bad products to the acquisition and rejection controller, the acquisition and rejection controller sends rejection information to the electromagnetic valve to achieve rejection processing of the defective cigarettes, and then the image processing server sends information of bad product statistics, defect types and defect images to the display
The cigarette appearance online detection method has the beneficial effects that the cigarette appearance online detection method aims at solving the problem that the existing cigarette appearance detection system cannot realize end face cavity depth (three-dimensional information) defect detection and the defect that the detection function is single. According to the invention, the on-line detection of the cavity depth of the cigarette holder is realized by using the triangulation type 3D camera and the method for detecting the cavity depth of the cigarette holder by using the three-dimensional information image, and the consistency of the feeling of the finished cigarette at the mouth is ensured. The invention realizes the multi-index appearance defect detection of the depth, the end face and the circumferential face of the cigarette cavity.
Aiming at the problem that the existing on-line cigarette appearance detection system based on the machine vision technology needs to acquire two images by a plurality of cameras or one camera acquires two images, the method reduces the interference of droplets by using a comparison algorithm. The invention uses the camera component to collect the primary image, and accurately identifies the droplets based on an artificial intelligence algorithm, wherein the identification rate is more than 95%. The invention not only reduces the requirement of image acquisition, but also improves the detection accuracy.
Aiming at the defect detection that the existing cigarette appearance online detection system can not realize the defect detection of the cigarette area with the pattern, the invention divides the cigarette image into a cigarette mouth section, a LOGO section and a cigarette body section, and models based on a deep learning model by using three different pattern areas, thereby realizing the appearance defect detection of LOGO section damage, oil stain, foam inclusion and the like. The invention can realize the detection of the appearance defects of cigarettes with complex patterns.
In addition, the deep learning model used by the invention can accurately classify the defects, and can provide information for manufacturers to remove mechanical faults and improve production schemes.
Drawings
FIG. 1 is a functional block diagram of the present invention;
FIG. 2 is an original image of a cigarette according to an embodiment of the present invention;
FIG. 3 is a corrected image of a cigarette according to an embodiment of the present invention;
FIG. 4 is an image of a region of interest of a cigarette in accordance with an embodiment of the present invention.
Detailed Description
The invention will now be described in further detail with reference to the drawings and preferred embodiments. These drawings are simplified schematic views illustrating only the basic structure of the present invention in a schematic manner, and thus show only the constitution related to the present invention.
As shown in fig. 1-4, a method for detecting cigarette appearance defects based on deep learning includes the following steps:
s1: when the cigarette passes through the phase camera, a front image and a back image of the cigarette are acquired by camera assemblies which are oppositely arranged on two sides of the cigarette making machine, the front image or the back image is composed of two side images, and the intersection of the two side images comprises a 180-degree circumferential area of the cigarette; the intersection of the front image and the back image comprises a 360-degree circumferential surface area of the cigarette;
s2: performing histogram equalization processing on the image acquired in the step S1;
s3: performing median filtering processing on the image processed in the step S2;
s4: the image processed in S3 is displayed at a height of 1: scale-out of 1 into an upper image and a lower image, and performing S5 and S6 processing on the two images, respectively;
s5: establishing a gray level template based on the image pyramid, and acquiring the position and the direction of the cigarette image by utilizing the template matching based on the gray level template;
s6: carrying out affine transformation according to the image position and the direction of S5 to correct the image position;
s7: combining the corrected upper image and the corrected lower image into one image;
s8: cutting the image processed by the S7 into three images according to the non-character cigarette holder area, the character cigarette holder area and the cigarette body area, wherein the three images are respectively a cigarette holder section image, a LOGO section image and a cigarette body section image;
s9: respectively generating masks in background areas of the cigarette holder segment image, the LOGO segment image and the cigarette body segment image, wherein a non-mask area of the cigarette holder segment image is used as an interested area 1, a non-mask area of the LOGO segment image is used as an interested area 2, and a non-mask area of the cigarette body segment image is used as an interested area 3;
wherein:
the sample set of ROIs 1 contains: defect-free images, droplet interference defect-free images, breakage defect images, stain defect images;
the sample set of ROIs 2 contains: defect-free images, droplet interference defect-free images, breakage defect images, stain defect images;
the sample set of ROIs 3 contains: defect free images, spray interference defect free images, breakage defect images, stain defect images, entrapment defect images.
S10: establishing a model data set by using an interested area sample set, wherein the interested area sample set is divided into a training set and a prediction set, and the training set and the prediction set jointly form the model data set; wherein the sample set of the region of interest 1 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 2 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 3 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image, an entrainment defect image;
s11: a deep learning algorithm model is established and optimized through the model data set,
s12: and detecting the defects of the cigarette image by using the deep learning model.
Further, the deep learning algorithm model in S11 is VGG16, and the sample sets of the three regions of interest are respectively brought into the model for training.
Further, the training process of the VGG16 model is as follows:
①, taking a sample (Xp, Yp) from the sample set, inputting the Xp into the network, wherein Xp is the p-th input sample, and Yp is the ideal output result;
in this process, the network performs the calculation of Op ═ Fn (… (F2(F1(XpW (1)) W (2)) …) W (n)), where Fn is a calculation function of the nth layer network and W (n) is a weight matrix of the nth layer.
③ calculating the difference between the actual output Op and the corresponding ideal output Yp;
④ the weight matrix is adjusted by back propagation with minimized error.
VGG16 comprises in combination: 13 Convolutional layers (Convolutional Layer), represented by conv3-XXX, 3 Fully connected layers (full connected Layer), represented by FC-XXXX, and 5 pooling layers (Pool Layer), represented by maxpool. Here, the convolutional layer and the all-connected layer have a weight coefficient and are also referred to as weight layers, and the total number is 13+3 to 16.
Further, the defect detection performed in S12 is to take the cigarette image acquired by the machine component in real time into the deep learning algorithm model of iv to obtain an identification result, specifically, a non-defective cigarette image, a droplet interference image, a roundness defect image, a damage defect image, a stain defect image, and a droplet defect image, if the identification result of the deep learning algorithm model is the non-defective cigarette image or the droplet interference image, the image processing control system determines that the cigarette is a good product, otherwise, the image processing control system rejects the cigarette as a bad product.
Further, image processing control module group, including image processing server, collection rejection controller, coding disc, solenoid valve and display, image processing server rejects controller and display electricity with the collection respectively and is connected, gathers and rejects the controller and is connected with coding disc and solenoid valve electricity respectively, still includes image card and IO card in the image processing server, and the image card is connected with image acquisition unit and collection rejection controller electricity respectively, and the IO card is connected with collection rejection controller electricity, the image acquisition unit is four camera subassemblies, and every camera subassembly of group is connected with image card electricity alone.
Further, the image processing control module has the following steps in the image processing process:
step 1: image acquisition: the coding disc inputs the position information of each cigarette to the acquisition and rejection controller, the acquisition and rejection controller converts the position information into a frame signal and sends the frame signal to the image card, the image card sends an acquisition signal to the image acquisition unit, and the image acquisition unit acquires images of two camera assemblies of the same 1 cigarette and then sends the images back to the image processing server to finish image acquisition;
step 2: image processing: the image processing server processes the images of the two camera assemblies to obtain a judgment result;
and step 3: and (3) judging and processing: if the image discrimination results in the step 2 are good products, the image processing control module sends good product statistical information to the display;
if the image discrimination result in the step 2 is a bad product, the image processing server executes two operations: firstly, the I/O card sends information of the specified cigarettes as bad cigarettes to the acquisition and rejection controller, the acquisition and rejection controller sends rejection information to the electromagnetic valve to achieve rejection processing of the defective cigarettes, and then the image processing server sends the bad cigarette statistics, defect types and defect image information to the display.
While particular embodiments of the present invention have been described in the foregoing specification, various modifications and alterations to the previously described embodiments will become apparent to those skilled in the art from this description without departing from the spirit and scope of the invention.
Claims (6)
1. A cigarette appearance defect detection method based on deep learning is characterized by comprising the following steps: the method comprises the following steps:
s1: when the cigarette passes through the phase camera, a front image and a back image of the cigarette are acquired by camera assemblies which are oppositely arranged on two sides of the cigarette making machine, the front image or the back image is composed of two side images, and the intersection of the two side images comprises a 180-degree circumferential area of the cigarette; the intersection of the front image and the back image comprises a 360-degree circumferential surface area of the cigarette;
s2: performing histogram equalization processing on the image acquired in the step S1;
s3: performing median filtering processing on the image processed in the step S2;
s4: the image processed in S3 is displayed at a height of 1: scale-out of 1 into an upper image and a lower image, and performing S5 and S6 processing on the two images, respectively;
s5: establishing a gray level template based on the image pyramid, and acquiring the position and the direction of the cigarette image by utilizing the template matching based on the gray level template;
s6: carrying out affine transformation according to the image position and the direction of S5 to correct the image position;
s7: combining the corrected upper image and the corrected lower image into one image;
s8: cutting the image processed by the S7 into three images according to the non-character cigarette holder area, the character cigarette holder area and the cigarette body area, wherein the three images are respectively a cigarette holder section image, a LOGO section image and a cigarette body section image;
s9: respectively generating masks in background areas of the cigarette holder segment image, the LOGO segment image and the cigarette body segment image, wherein a non-mask area of the cigarette holder segment image is used as an interested area 1, a non-mask area of the LOGO segment image is used as an interested area 2, and a non-mask area of the cigarette body segment image is used as an interested area 3;
s10: establishing a model data set by using an interested area sample set, wherein the interested area sample set is divided into a training set and a prediction set, and the training set and the prediction set jointly form the model data set; wherein the sample set of the region of interest 1 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 2 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image; wherein the sample set of the region of interest 3 comprises a defect-free image, a droplet interference defect-free image, a breakage defect image, a stain defect image, an entrainment defect image;
s11: a deep learning algorithm model is established and optimized through the model data set,
s12: and detecting the defects of the cigarette image by using the deep learning model.
2. The cigarette appearance defect detection method based on deep learning of claim 1, characterized in that: the deep learning algorithm model in the step S11 is VGG16, and the sample sets of the three regions of interest are respectively brought into the model for training.
3. The cigarette appearance defect detection method based on deep learning as claimed in claim 2, characterized in that: the VGG16 model training process is as follows:
①, taking a sample (Xp, Yp) from the sample set, inputting the Xp into the network, wherein Xp is the p-th input sample, and Yp is the ideal output result;
in this process, the network performs the calculation of Op ═ Fn (… (F2(F1(XpW (1)) W (2)) …) W (n)), where Fn is a calculation function of the nth layer network and W (n) is a weight matrix of the nth layer.
③ calculating the difference between the actual output Op and the corresponding ideal output Yp;
④ the weight matrix is adjusted by back propagation with minimized error.
4. The cigarette appearance defect detection method based on deep learning of claim 3, characterized in that: and the defect detection in the step S12 is specifically to take the cigarette image acquired by the machine component in real time into the deep learning algorithm model of iv to obtain an identification result, specifically, a non-defective cigarette image, a spray interference image, a roundness defect image, a damage defect image, a stain defect image and a foam inclusion defect image, if the identification result of the deep learning algorithm model is the non-defective cigarette image or the spray interference image, the image processing control system judges the cigarette to be a good product, otherwise, the image processing control system removes the cigarette as a bad product.
5. The cigarette appearance defect detection method based on deep learning of claim 1, characterized in that: image processing control module group, including image processing server, collection rejection controller, coding disc, solenoid valve and display, image processing server rejects controller and display electricity with the collection respectively and is connected, gathers and rejects the controller and is connected with coding disc and solenoid valve electricity respectively, still includes image card and IO card in the image processing server, and the image card rejects the controller electricity with image acquisition unit and collection respectively and is connected, and the IO card is rejected the controller electricity with gathering and is connected, the image acquisition unit is four camera subassemblies, and every camera subassembly of group is connected with image card electricity alone.
6. The cigarette appearance defect detection method based on deep learning as claimed in claim 1 or 5, characterized in that: the image processing control module comprises the following steps in the image processing process:
step 1: image acquisition: the coding disc inputs the position information of each cigarette to the acquisition and rejection controller, the acquisition and rejection controller converts the position information into a frame signal and sends the frame signal to the image card, the image card sends an acquisition signal to the image acquisition unit, and the image acquisition unit acquires images of two camera assemblies of the same 1 cigarette and then sends the images back to the image processing server to finish image acquisition;
step 2: image processing: the image processing server processes the images of the two camera assemblies to obtain a judgment result;
and step 3: and (3) judging and processing: if the image discrimination results in the step 2 are good products, the image processing control module sends good product statistical information to the display;
if the image discrimination result in the step 2 is a bad product, the image processing server executes two operations: firstly, the I/O card sends information of the specified cigarettes as bad cigarettes to the acquisition and rejection controller, the acquisition and rejection controller sends rejection information to the electromagnetic valve to achieve rejection processing of the defective cigarettes, and then the image processing server sends the bad cigarette statistics, defect types and defect image information to the display.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010018188.8A CN111260609B (en) | 2020-01-08 | 2020-01-08 | Cigarette appearance defect detection method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010018188.8A CN111260609B (en) | 2020-01-08 | 2020-01-08 | Cigarette appearance defect detection method based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111260609A true CN111260609A (en) | 2020-06-09 |
CN111260609B CN111260609B (en) | 2023-04-18 |
Family
ID=70946829
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010018188.8A Active CN111260609B (en) | 2020-01-08 | 2020-01-08 | Cigarette appearance defect detection method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111260609B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112184648A (en) * | 2020-09-22 | 2021-01-05 | 苏州中科全象智能科技有限公司 | Piston surface defect detection method and system based on deep learning |
CN113333329A (en) * | 2021-08-04 | 2021-09-03 | 南京创智出彩科技有限公司 | Cigarette defect detection system based on deep learning |
CN113421235A (en) * | 2021-06-17 | 2021-09-21 | 中国电子科技集团公司第四十一研究所 | Cigarette positioning device and method based on deep learning |
CN114965462A (en) * | 2022-04-18 | 2022-08-30 | 中烟摩迪(江门)纸业有限公司 | Real-time tracking detection method for cigarette ash holding force |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103504472A (en) * | 2012-06-14 | 2014-01-15 | 豪尼机械制造股份公司 | Device and method for examining end surface of transversally conveyed rod-shaped product of tobacco processing industry |
CN108613987A (en) * | 2018-05-15 | 2018-10-02 | 南京焦耳科技有限责任公司 | Cigarette appearance real-time online detects device for eliminating and method |
GB201900672D0 (en) * | 2019-01-17 | 2019-03-06 | Smiths Heimann Sas | Classifier using data generation |
CN109858539A (en) * | 2019-01-24 | 2019-06-07 | 武汉精立电子技术有限公司 | A kind of ROI region extracting method based on deep learning image, semantic parted pattern |
CN209300267U (en) * | 2018-09-06 | 2019-08-27 | 云南远足科技有限公司 | A kind of cigarette appearance images processing control system and its on-line measuring device |
CN110403232A (en) * | 2019-07-24 | 2019-11-05 | 浙江中烟工业有限责任公司 | A kind of cigarette quality detection method based on second level algorithm |
CN110451035A (en) * | 2019-07-02 | 2019-11-15 | 中国电子科技集团公司第四十一研究所 | A kind of electronic cigarette cigarette detection device and method based on machine vision technique |
-
2020
- 2020-01-08 CN CN202010018188.8A patent/CN111260609B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103504472A (en) * | 2012-06-14 | 2014-01-15 | 豪尼机械制造股份公司 | Device and method for examining end surface of transversally conveyed rod-shaped product of tobacco processing industry |
CN108613987A (en) * | 2018-05-15 | 2018-10-02 | 南京焦耳科技有限责任公司 | Cigarette appearance real-time online detects device for eliminating and method |
CN209300267U (en) * | 2018-09-06 | 2019-08-27 | 云南远足科技有限公司 | A kind of cigarette appearance images processing control system and its on-line measuring device |
GB201900672D0 (en) * | 2019-01-17 | 2019-03-06 | Smiths Heimann Sas | Classifier using data generation |
CN109858539A (en) * | 2019-01-24 | 2019-06-07 | 武汉精立电子技术有限公司 | A kind of ROI region extracting method based on deep learning image, semantic parted pattern |
CN110451035A (en) * | 2019-07-02 | 2019-11-15 | 中国电子科技集团公司第四十一研究所 | A kind of electronic cigarette cigarette detection device and method based on machine vision technique |
CN110403232A (en) * | 2019-07-24 | 2019-11-05 | 浙江中烟工业有限责任公司 | A kind of cigarette quality detection method based on second level algorithm |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112184648A (en) * | 2020-09-22 | 2021-01-05 | 苏州中科全象智能科技有限公司 | Piston surface defect detection method and system based on deep learning |
CN113421235A (en) * | 2021-06-17 | 2021-09-21 | 中国电子科技集团公司第四十一研究所 | Cigarette positioning device and method based on deep learning |
CN113333329A (en) * | 2021-08-04 | 2021-09-03 | 南京创智出彩科技有限公司 | Cigarette defect detection system based on deep learning |
CN114965462A (en) * | 2022-04-18 | 2022-08-30 | 中烟摩迪(江门)纸业有限公司 | Real-time tracking detection method for cigarette ash holding force |
Also Published As
Publication number | Publication date |
---|---|
CN111260609B (en) | 2023-04-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111189837B (en) | Cigarette appearance online detection method and device | |
CN111260609B (en) | Cigarette appearance defect detection method based on deep learning | |
CN110570393B (en) | Mobile phone glass cover plate window area defect detection method based on machine vision | |
CN105069790B (en) | A kind of gear open defect fast image detection method | |
CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
US10803573B2 (en) | Method for automated detection of defects in cast wheel products | |
CN104008370B (en) | A kind of video face identification method | |
CN111223093A (en) | AOI defect detection method | |
CN109636772A (en) | The defect inspection method on the irregular shape intermetallic composite coating surface based on deep learning | |
CN105389586B (en) | A method of detecting shrimp body integrality automatically based on computer vision | |
CN107230203A (en) | Casting defect recognition methods based on human eye vision attention mechanism | |
CN110473178A (en) | A kind of open defect detection method and system based on multiple light courcess fusion | |
CN110232389A (en) | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance | |
CN111582294A (en) | Method for constructing convolutional neural network model for surface defect detection and application thereof | |
CN108985337A (en) | A kind of product surface scratch detection method based on picture depth study | |
CN103927534A (en) | Sprayed character online visual detection method based on convolutional neural network | |
CN109840900A (en) | A kind of line detection system for failure and detection method applied to intelligence manufacture workshop | |
CN109540925B (en) | Complex ceramic tile surface defect detection method based on difference method and local variance measurement operator | |
CN110097596A (en) | A kind of object detection system based on opencv | |
CN108389180A (en) | A kind of fabric defect detection method based on deep learning | |
CN108830831A (en) | One kind is based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF | |
CN110276276A (en) | The determination method and system of examinee's face direction of visual lines in a kind of Driving Test | |
CN113920142A (en) | Sorting manipulator multi-object sorting method based on deep learning | |
CN106570877A (en) | Coal mining machine pose positioning system and method based on coal mining machine virtual prototype and real image registration | |
CN113706496B (en) | Aircraft structure crack detection method based on deep learning model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |