CN109214222A - Based on Embedded 32 cigarette laser code identifying systems and its recognition methods - Google Patents
Based on Embedded 32 cigarette laser code identifying systems and its recognition methods Download PDFInfo
- Publication number
- CN109214222A CN109214222A CN201811009972.1A CN201811009972A CN109214222A CN 109214222 A CN109214222 A CN 109214222A CN 201811009972 A CN201811009972 A CN 201811009972A CN 109214222 A CN109214222 A CN 109214222A
- Authority
- CN
- China
- Prior art keywords
- recognizer
- module
- laser code
- imaging
- image
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/12—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using a selected wavelength, e.g. to sense red marks and ignore blue marks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/0004—Hybrid readers
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Electromagnetism (AREA)
- General Health & Medical Sciences (AREA)
- Toxicology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses be based on Embedded 32 cigarette laser code identifying systems and its recognition methods, identifying system includes light-source system, image capturing system, algoritic module, and recognizer selecting module and at least one recognizer module are provided in algoritic module；Recognition methods includes obtaining initial imaging picture, and the texture information of present image is calculated by image capturing system；According to preliminary imaging texture information effect, the light source light spectrum of adaptively selected second of imaging and the exposure value for determining selected light spectrum carry out secondary imaging；The texture information of second of imaging is calculated again；According to the texture information of second of imaging, adaptively selected algorithm carries out high-accuracy and identifies a cigarette laser code, and shows；Based under unlike material carton cigarette, light source type and brightness are automatically selected and adjust, it under any circumstance, can blur-free imaging；A variety of laser code recognizers can be merged automatically as a result, being optimal recognition effect according to laser code image quality.
Description
Technical field
The present invention relates to the fields such as industrial camera ccd image technology, optical character identification OCR technique, are bases specifically
In Embedded 32 cigarette laser code identifying systems and its recognition methods.
Background technique
Industrial camera is commonly called as video camera again, and for traditional civil camera (video camera), it has high image
Stability, high-transmission ability and high anti-jamming capacity etc., industrial camera is based on CCD (Charge Coupled mostly on the market
Device) the camera of chip.
Industrial camera (CCD), full name in English: Charge-coupled Device, Chinese name: charge coupled cell.It can
With referred to as ccd image sensor, image controller is also.CCD is a kind of semiconductor devices, optical image can be converted into electricity
Signal.The small photoactive substance being implanted on CCD is referred to as pixel (Pixel).The pixel number for including on one piece of CCD is more, provides
Screen resolution it is also higher.CCD to act like film the same, but it is that optical signal is converted into charge signal.On CCD
There are many marshalling photodiode, light can be incuded, and convert optical signals into electric signal, through external sampling amplification and
Analog to digital conversion circuit is converted into data image signal.
OCR (Optical Character Recognition, optical character identification) refers to that electronic equipment (such as scans
Instrument or digital camera) check the character printed on paper, its shape is determined by the mode for detecting dark, bright, then uses character recognition
Shape is translated into the process of computword by method；That is, it is directed to printed character, it will be in paper document using optical mode
Text conversion become the image file of black and white lattice, and by identification software by the text conversion in image at text formatting,
The technology further edited and processed for word processor.It is OCR how except mistake or using auxiliary information raising recognition correct rate
Therefore most important project, the noun of ICR (Intelligent Character Recognition) also generate.Measure one
A OCR system performance quality refers mainly to indicate: reject rate, misclassification rate, recognition speed, the friendly of user interface, product
Stability, ease for use and feasibility etc..
Tobacco business is the industry of one high tax revenue high profit, and under the temptation of huge interests, some criminals are quickly
And take a risk, make and sell without restraint personation cigarette seek exorbitant profit.In order to effectively hit the illegal activities such as personation, smuggling, string goods cigarette, maintenance
National fundamental interests and the health for ensureing consumer, the sale monopoly of national team's tobacco entity.Legal tobacco leaf production enterprise and warp
Pin trade company stamped on cigarette case comprising the coding date, sources of supply, the laser of retail customer property, the supply of material information such as object compiles
Code.
However, the outer packing of cigarette, the material of use is different, the degree of laser code calcination, in different light sources and difference
Under brightness irradiation, the clarity of laser code imaging is different, meanwhile, laser code character is in different-effect, in single identification
Under algorithm, discrimination is also different.
The product of laser code identifying system is also fewer both at home and abroad at present, even if a small number of several families, the product provided are all
Under single light source, fixed light source brightness, single algorithm is identified, in the case where not having to material carton cigarette, laser code recognition effect is not
It is ideal.
Summary of the invention
It is an object of the invention to design to be based on Embedded 32 cigarette laser code identifying systems and its recognition methods,
The identifying system can be realized laser code blur-free imaging, thus the recognition effect being optimal；The recognition methods can be real
The identification of existing laser code is handled in identification process using at least one recognizer, to effectively raise entire
Identify the adaptive capacity of data.
The present invention is achieved through the following technical solutions: being based on Embedded 32 cigarettes laser code identifying system, including light
Source system, image capturing system, algoritic module, the light-source system are connected with image capturing system, and image capturing system connects
Connect algoritic module；It is provided with recognizer selecting module and at least one recognizer module in the algoritic module, and schemes
As acquisition system connection recognizer selecting module, recognizer selecting module is connected at least one recognizer module.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved, it is special
Cai Yong following set-up modes: there are three types of the recognizer module settings, and is respectively to be connected with recognizer selecting module
Recognizer modules A, recognizer module B and the recognizer module C connect, and be additionally provided with respectively with recognizer modules A,
The result display module that recognizer module B is connected with recognizer module C.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved, it is special
Cai Yong following set-up modes: described image acquisition system includes industrial camera and image processor interconnected, and image
Processor connects recognizer selecting module；The light-source system includes control circuit for light source and light source interconnected, and light
Source control circuit connects image processor.
Based on the recognition methods of Embedded 32 cigarettes laser code identifying system, following set-up modes are especially used: packet
Include step in detail below:
1) initial imaging picture is obtained, the texture information of present image is calculated by image capturing system；
2) it according to preliminary imaging texture information effect, the light source light spectrum of adaptively selected second of imaging and determines selected
The exposure value of light source light spectrum carries out secondary imaging；
3) texture information of second of imaging is calculated again；
4) according to the texture information of second of imaging, adaptively selected algorithm carries out high-accuracy and identifies a cigarette laser code,
And it shows.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods, especially use following set-up modes: the texture information includes the mean value of image and the mean square deviation of image；Image
Mean value uses formula:It calculates；The mean square deviation of image uses formula:It calculates.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods especially uses following set-up modes: the light source light spectrum of adaptively selected second of imaging uses following manner:
(1) when mean square deviation is being greater than 4096, white light source is selected, while exposure value takes range: 100-130, it is preferred that white
The exposure value of radiant is 128；
(2) when equal sides is looked into greater than 1024 and when less than 4096, select blue light source, while exposure value takes range:
150-180, it is preferred that the exposure value of blue light source is 160；
(3) when mean square deviation is less than 1024, red-light source is selected, while exposure value takes range: 70-90, it is preferred that light source
Exposure value is 98.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods, especially use following set-up modes: algorithm includes algorithm A, algorithm B and algorithm C in the step 4)；
4.1) when mean square deviation be greater than 4096, selection algorithm A；
4.2) when mean square deviation is greater than 1024 and when less than 4096, selection algorithm B；
4.3) when mean square deviation is less than 1024, selection algorithm C.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods especially uses following set-up modes: the algorithm A is the OCR recognizer based on SVM.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods especially uses following set-up modes: the algorithm B is the OCR recognizer based on Adboost.
It is further of the present invention based on Embedded 32 cigarettes laser code identifying system to be better achieved
Recognition methods especially uses following set-up modes: the algorithm C is convolutional neural networks recognizer.
Compared with prior art, the present invention have the following advantages that and the utility model has the advantages that
(1) 32 cigarettes laser code identifying system provided by the present invention, in progress cigarette laser code identification, based on not
Under same material carton cigarette, light source type and brightness are automatically selected and adjust, it under any circumstance, can blur-free imaging；And it can
According to laser code image quality, a variety of laser code recognizers are merged automatically as a result, being optimal recognition effect.
(2) recognition methods of the present invention can be realized the identification of laser code, in identification process, using at least one
Recognizer is handled, to effectively raise the adaptive capacity of entire identification data.
(3) present invention comprehensive utilization SVM based on OCR recognizer, the OCR recognizer based on Adboost with
And convolutional neural networks recognizer, accuracy of identification and recognition performance can be effectively improved, to effectively improve recognition effect.
(4) one picture of initial acquisition of the present invention, according to picture textural characteristics, adaptively selected light source type and light source
Brightness, exposure value carry out second of acquisition picture, second of imaging are enable to reach optimal imaging effect according to the material of cigarette；
The textural characteristics of the further image according to second of acquisition, self-adapted call recognizer improve 32 laser code identifications
Rate.
Detailed description of the invention
Fig. 1 is identifying system structural schematic diagram of the present invention.
Fig. 2 is secondary imaging effect picture.
Fig. 3 is to be ultimately imaged effect picture.
Specific embodiment
The present invention is described in further detail below with reference to embodiment, embodiments of the present invention are not limited thereto.
To keep the purposes, technical schemes and advantages of embodiment of the present invention clearer, implement below in conjunction with the present invention
The technical solution in embodiment of the present invention is clearly and completely described in attached drawing in mode, it is clear that described reality
The mode of applying is some embodiments of the invention, rather than whole embodiments.Based on the embodiment in the present invention, ability
Domain those of ordinary skill every other embodiment obtained without creative efforts, belongs to the present invention
The range of protection.Therefore, the detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit below and is wanted
The scope of the present invention of protection is sought, but is merely representative of selected embodiment of the invention.Based on the embodiment in the present invention,
Every other embodiment obtained by those of ordinary skill in the art without making creative efforts belongs to this
Invent the range of protection.
In the description of the present invention, it is to be understood that, term " center ", " longitudinal direction ", " transverse direction ", " length ", " width ",
" thickness ", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outside", " up time
The orientation or positional relationship of the instructions such as needle ", " counterclockwise " is to be based on the orientation or positional relationship shown in the drawings, and is merely for convenience of
The description present invention and simplified description, rather than the equipment of indication or suggestion meaning or element must have a particular orientation, with spy
Fixed orientation construction and operation, therefore be not considered as limiting the invention.
In addition, term " first ", " second " are used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance
Or implicitly indicate the quantity of indicated technical characteristic.Define " first " as a result, the feature of " second " can be expressed or
Implicitly include one or more of the features.In the description of the present invention, the meaning of " plurality " is two or more,
Unless otherwise specifically defined.
In the present invention unless specifically defined or limited otherwise, term " installation ", " connected ", " connection ", " fixation " etc.
Term shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or integral；It can be mechanical connect
It connects, is also possible to be electrically connected；It can be directly connected, can also can be in two elements indirectly connected through an intermediary
The interaction relationship of the connection in portion or two elements.It for the ordinary skill in the art, can be according to specific feelings
Condition understands the concrete meaning of above-mentioned term in the present invention.
In the present invention unless specifically defined or limited otherwise, fisrt feature second feature "upper" or "lower"
It may include that the first and second features directly contact, also may include that the first and second features are not direct contacts but pass through it
Between other characterisation contact.Moreover, fisrt feature includes the first spy above the second feature " above ", " above " and " above "
Sign is right above second feature and oblique upper, or is merely representative of first feature horizontal height higher than second feature.Fisrt feature exists
Second feature " under ", " lower section " and " following " include that fisrt feature is directly below and diagonally below the second feature, or is merely representative of
First feature horizontal height is less than second feature.
Embodiment 1:
The invention proposes Embedded 32 cigarettes laser code identifying system is based on, cigarette laser code identification is being carried out
When, based under unlike material carton cigarette, light source type and brightness are automatically selected and adjust, it under any circumstance, can blur-free imaging；
And a variety of laser code recognizers can be merged automatically as a result, being optimal recognition effect, such as according to laser code image quality
Shown in Fig. 1, following setting structures are especially used: including light-source system, image capturing system, algoritic module, the light-source system
It is connected with image capturing system, image capturing system join algorithm module；Recognizer is provided in the algoritic module
Selecting module and at least one recognizer module, and image capturing system connects recognizer selecting module, recognizer choosing
Module is selected to be connected at least one recognizer module.
As scheme is preferable to provide, it is being based on being provided with light source system in Embedded 32 cigarettes laser code identifying system
System, image capturing system and algoritic module；
The light-source system obtains target information and background information in image by light source Lighting Design appropriate
Optimal separation, can substantially reduce the difficulty of image processing algorithm segmentation, identification, while improve the positioning of system, measurement accuracy,
The reliability and comprehensive performance for making system are improved.
Described image acquisition system obtains image by high resolution industrial camera, and image is sent into algoritic module,
It is identified.
The algoritic module calls algoritic module to acquired image, carries out algorithm identification.
The algorithms selection module, according to the parameter index of light-source system, the end value of combination algorithm module is last to know
Other result judges selection.
Embodiment 2:
The present embodiment is further to optimize on the basis of the above embodiments, and further is that the present invention is better achieved
It is described based on Embedded 32 cigarettes laser code identifying system, as shown in Figure 1, especially using following set-up modes: described
There are three types of the settings of recognizer module, and is respectively the recognizer modules A being connected with recognizer selecting module, identification calculation
Method module B and recognizer module C, and be additionally provided with respectively with recognizer modules A, recognizer module B and recognizer
The result display module that module C is connected.
As the scheme that is preferable to provide, there are three types of the settings of recognizer module:
One is recognizer modules A, and inside is mounted with recognizer A, and preferred recognizer A is using one kind
OCR recognizer based on SVM, it is common that wherein SVM (Support Vector Machine), which refers to support vector machines,
A kind of method of discrimination.It is the learning model for having supervision in machine learning field, commonly used to carry out pattern-recognition, divide
Class and regression analysis.
Secondly being recognizer module B, inside is mounted with recognizer B, and preferred recognizer B is using one kind
OCR recognizer based on Adboost, Adaboost are a kind of iterative algorithms, and core concept is for the same training
The different classifier (Weak Classifier) of collection training, then gets up these weak classifier sets, constitutes one stronger final point
Class device (strong classifier).Its algorithm itself is realized by changing data distribution, it is according to each among each training set
Whether the classification of sample correct and the accuracy rate of general classification of last time, to determine the weight of each sample.Power will be modified
The new data set of value is given sub-classification device and is trained, and finally finally merges the classifier that each training obtains, makees
For last Decision Classfication device.Some unnecessary training data features can be excluded using adaboost classifier, and are placed on
Above crucial training data.
Thirdly being recognizer module C, inside is mounted with recognizer C, and preferred recognizer C uses convolutional Neural
Network (Convolutional Neural Network, CNN), CNN is a kind of feedforward neural network, its artificial neuron can
To respond the surrounding cells in a part of coverage area, there is outstanding performance for large-scale image procossing, it includes convolutional layer
(convolutional layer) and pond layer (pooling layer).
Convolutional neural networks are developed recentlies, and cause a kind of efficient identification method paid attention to extensively.20th century 60
Age, Hubel and Wiesel are in studying cat cortex for finding its uniqueness when local sensitivity and the neuron of direction selection
Network structure can be effectively reduced the complexity of Feedback Neural Network, then propose convolutional neural networks
(Convolutional Neural Networks- abbreviation CNN).Now, CNN has become the research heat of numerous scientific domains
One of point, especially can be directly defeated since the network avoids the pretreatment complicated early period to image in pattern classification field
Enter original image, thus has obtained more being widely applied.The new cognitron that K.Fukushima was proposed in 1980 is convolution mind
First realization network through network.Then, more researchers improve the network.Wherein, have and represent
Property research achievement be that Alexander and Taylor propose " improving cognitron ", this method combines various improved methods
Advantage simultaneously avoids time-consuming error back propagation.
Generally, the basic structure of CNN includes two layers, and one is characterized extract layer, the input of each neuron with it is previous
The local acceptance region of layer is connected, and extracts the feature of the part.After the local feature is extracted, it is between other feature
Positional relationship is also decided therewith；The second is Feature Mapping layer, each computation layer of network is made of multiple Feature Mappings, often
A Feature Mapping is a plane, and the weight of all neurons is equal in plane.Feature Mapping structure is small using influence function core
Activation primitive of the sigmoid function as convolutional network so that Feature Mapping has shift invariant.Further, since one
Neuron on mapping face shares weight, thus reduces the number of network freedom parameter.Each of convolutional neural networks
Convolutional layer all followed by one is used to ask the computation layer of local average and second extraction, this distinctive feature extraction structure twice
Reduce feature resolution.
CNN is mainly used to the X-Y scheme of identification displacement, scaling and other forms distortion invariance.Due to the feature of CNN
Detection layers are learnt by training data, so explicit feature extraction is avoided when using CNN, and implicitly from instruction
Practice and is learnt in data；Furthermore since the neuron weight on same Feature Mapping face is identical, so network can be learned parallel
It practises, this is also convolutional network is connected with each other a big advantage of network relative to neuron.Convolutional neural networks are with its local weight
Shared special construction has unique superiority in terms of speech recognition and image procossing, is laid out closer to actual life
Object neural network, the shared complexity for reducing network of weight, the especially image of multidimensional input vector can directly input net
This feature of network avoids the complexity of data reconstruction in feature extraction and assorting process.
Embodiment 3:
The present embodiment is to advanced optimize based on any of the above embodiments, as shown in Figure 1, being further more preferable
Ground realization is of the present invention to be based on Embedded 32 cigarettes laser code identifying system, especially uses following set-up modes: institute
Stating image capturing system includes industrial camera and image processor interconnected, and image processor connection recognizer selects
Module；The light-source system includes control circuit for light source and light source interconnected, and control circuit for light source connects image procossing
Device.
As the scheme that is preferable to provide, industrial camera interconnected (CCD) and image are provided in image capturing system
Processor；There are many photodiode of marshalling on the industrial camera (CCD), light can be incuded, and optical signal is turned
Become electric signal, is converted into data image signal through external sampling amplification and analog to digital conversion circuit；Described image processor is preferred
Using x86 platform processor, main function and function are mathematical algorithm module and algorithms selection module.
Control circuit for light source and light source interconnected, and control circuit for light source connection figure are provided in the light-source system
As processor；Light-source system includes: that light source and control circuit for light source by light source Lighting Design appropriate make the mesh in image
Mark information and background information obtain optimal separation, can substantially reduce the difficulty of image processing algorithm segmentation, identification, improve simultaneously
The positioning of system, measurement accuracy are improved the reliability of system and comprehensive performance.
Embodiment 4:
The present embodiment is to advanced optimize based on any of the above embodiments, is based on Embedded 32 cigarette laser
The recognition methods of code identifying system, especially uses following set-up modes: comprising the following specific steps
1) initial imaging picture is obtained, the texture information of present image is calculated by image capturing system；
It is preferred to use default light source (white light) in the state of defaulting exposure value, an initial imaging picture is obtained, and pass
It is defeated by image processor, image processor calculates the texture information of present image；
2) according to preliminary imaging texture information effect, light source light spectrum (preferred, the light source of adaptively selected second of imaging
Spectrum contain white light, feux rouges, blue light) and determine selected light spectrum exposure value, carry out secondary imaging；
3) texture information of second of imaging is calculated again；
4) according to the texture information of second of imaging, adaptively selected algorithm carries out high-accuracy and identifies a cigarette laser code,
And it shows.
Embodiment 5:
The present embodiment is further to optimize on the basis of the above embodiments, and further is that the present invention is better achieved
The recognition methods based on Embedded 32 cigarettes laser code identifying system especially uses following set-up modes: described
Texture information includes the mean value of image and the mean square deviation of image；The mean value of image uses formula:It calculates；The mean square deviation of image uses formula:It calculates；Wherein, n indicates this group of data amount check,
x1、x2、x3……xnIndicate current picture value specific value.
Embodiment 6:
The present embodiment is further optimized on the basis of embodiment 4 or 5, and further is that the present invention is better achieved
The recognition methods based on Embedded 32 cigarettes laser code identifying system especially uses following set-up modes: described
The light source light spectrum of adaptively selected second of imaging uses following manner:
(1) when mean square deviation is being greater than 4096, white light source is selected, while exposure value takes range: 100-130, it is preferred that white
The exposure value of radiant is 128；
(2) when equal sides is looked into greater than 1024 and when less than 4096, select blue light source, while exposure value takes range:
150-180, it is preferred that the exposure value of blue light source is 160；
(3) when mean square deviation is less than 1024, red-light source is selected, while exposure value takes range: 70-90, it is preferred that light source
Exposure value is 98.
Embodiment 7:
The present embodiment is advanced optimized on the basis of embodiment 4 or 5 or 6, and further is that this hair is better achieved
The bright recognition methods based on Embedded 32 cigarettes laser code identifying system, especially uses following set-up modes: institute
Stating algorithm in step 4) includes algorithm A, algorithm B and algorithm C；
4.1) when mean square deviation be greater than 4096, selection algorithm A；
4.2) when mean square deviation is greater than 1024 and when less than 4096, selection algorithm B；
4.3) when mean square deviation is less than 1024, selection algorithm C.
Embodiment 8:
The present embodiment is advanced optimized on the basis of embodiment 4 or 5 or 6 or 7, and further is that this is better achieved
The invention recognition methods based on Embedded 32 cigarettes laser code identifying system, especially uses following set-up modes:
The algorithm A is the OCR recognizer based on SVM, and wherein SVM (Support Vector Machine) refers to supporting
Vector machine is a kind of common method of discrimination.The learning model for having supervision in machine learning field, commonly used into
Row pattern-recognition, classification and regression analysis.
Embodiment 9:
The present embodiment is advanced optimized on the basis of embodiment 4 or 5 or 6 or 7 or 8, and further is preferably real
The existing recognition methods of the present invention based on Embedded 32 cigarettes laser code identifying system, especially uses following setting sides
Formula: the algorithm B is the OCR recognizer based on Adboost, and Adaboost is a kind of iterative algorithm, core concept
It is the classifier (Weak Classifier) different for the training of the same training set, then these weak classifier sets is got up, are constituted
One stronger final classification device (strong classifier).Its algorithm itself is realized by changing data distribution, it is according to every
Whether the classification of each sample correct among secondary training set and the accuracy rate of general classification of last time, to determine each sample
Weight.It gives the new data set for modifying weight to sub-classification device to be trained, the classification for finally obtaining each training
Device finally merges, as last Decision Classfication device.Some unnecessary instructions can be excluded using adaboost classifier
Practice data characteristics, and is placed on above crucial training data.
Embodiment 10:
The present embodiment is advanced optimized on the basis of embodiment 4 or 5 or 6 or 7 or 8 or 9, further for preferably
It realizes the recognition methods of the present invention based on Embedded 32 cigarettes laser code identifying system, especially uses following settings
Mode: the algorithm C is convolutional neural networks recognizer, convolutional neural networks (Convolutional Neural
Network, CNN), CNN is a kind of feedforward neural network, its artificial neuron can respond the week in a part of coverage area
Unit is enclosed, has outstanding performance for large-scale image procossing, it includes convolutional layer (convolutional layer) and pond layer
(pooling layer)。
Convolutional neural networks are developed recentlies, and cause a kind of efficient identification method paid attention to extensively.20th century 60
Age, Hubel and Wiesel are in studying cat cortex for finding its uniqueness when local sensitivity and the neuron of direction selection
Network structure can be effectively reduced the complexity of Feedback Neural Network, then propose convolutional neural networks
(Convolutional Neural Networks- abbreviation CNN).Now, CNN has become the research heat of numerous scientific domains
One of point, especially can be directly defeated since the network avoids the pretreatment complicated early period to image in pattern classification field
Enter original image, thus has obtained more being widely applied.The new cognitron that K.Fukushima was proposed in 1980 is convolution mind
First realization network through network.Then, more researchers improve the network.Wherein, have and represent
Property research achievement be that Alexander and Taylor propose " improving cognitron ", this method combines various improved methods
Advantage simultaneously avoids time-consuming error back propagation.
Generally, the basic structure of CNN includes two layers, and one is characterized extract layer, the input of each neuron with it is previous
The local acceptance region of layer is connected, and extracts the feature of the part.After the local feature is extracted, it is between other feature
Positional relationship is also decided therewith；The second is Feature Mapping layer, each computation layer of network is made of multiple Feature Mappings, often
A Feature Mapping is a plane, and the weight of all neurons is equal in plane.Feature Mapping structure is small using influence function core
Activation primitive of the sigmoid function as convolutional network so that Feature Mapping has shift invariant.Further, since one
Neuron on mapping face shares weight, thus reduces the number of network freedom parameter.Each of convolutional neural networks
Convolutional layer all followed by one is used to ask the computation layer of local average and second extraction, this distinctive feature extraction structure twice
Reduce feature resolution.
CNN is mainly used to the X-Y scheme of identification displacement, scaling and other forms distortion invariance.Due to the feature of CNN
Detection layers are learnt by training data, so explicit feature extraction is avoided when using CNN, and implicitly from instruction
Practice and is learnt in data；Furthermore since the neuron weight on same Feature Mapping face is identical, so network can be learned parallel
It practises, this is also convolutional network is connected with each other a big advantage of network relative to neuron.Convolutional neural networks are with its local weight
Shared special construction has unique superiority in terms of speech recognition and image procossing, is laid out closer to actual life
Object neural network, the shared complexity for reducing network of weight, the especially image of multidimensional input vector can directly input net
This feature of network avoids the complexity of data reconstruction in feature extraction and assorting process.
Embodiment 11:
The present embodiment is to advanced optimize based on any of the above embodiments, using technology disclosed in this invention into
Row identifying processing is transferred to image processor, image processor calculates present image firstly, obtaining an initial imaging picture
Texture information, mean value and mean square deviation including image, according to preliminary imaging texture information effect, adaptively selected second at
The light source light spectrum (white light, feux rouges, blue light) of picture has determined brightness value, the exposure value of selected spectrum, carries out secondary imaging, imaging
Effect picture is as shown in Figure 2；
The texture information of second of imaging, including image mean value and variance are calculated again, according to second of imaging texture letter
Breath, adaptively selected algorithm (algorithm A, algorithm B, algorithm C), high-accuracy identify a cigarette laser code, and show, finally at
As effect is as shown in Figure 3.
Embodiment 12:
The present embodiment is advanced optimized on the basis of any one of Examples 1 to 10 embodiment, and initial acquisition one opens figure
Piece obtains an initial imaging picture, is transferred to image processor, and image processor calculates the texture information of present image, packet
The mean value and mean square deviation for including image, according to preliminary imaging texture information effect, the light source light spectrum of adaptively selected second of imaging
(white light, feux rouges, blue light) carries out secondary imaging, then directlys adopt algorithm and carries out 32 laser code identifications.
Embodiment 13:
The present embodiment is advanced optimized on the basis of any one of Examples 1 to 10 embodiment, and initial acquisition one opens figure
Then piece calculates the texture information of acquisition imaging, including image mean value and variance, adaptive according to second of imaging texture information
It answers selection algorithm (algorithm A, algorithm B, algorithm C), high-accuracy identifies a cigarette laser code, and shows.
Embodiment 14:
The present embodiment is advanced optimized on the basis of any one of embodiment 1~13 embodiment, and initial acquisition one opens figure
Piece obtains an initial imaging picture, is transferred to image processor, and image processor calculates the texture information of present image, packet
The mean value and mean square deviation for including image, according to preliminary imaging texture information effect, the light source light spectrum of adaptively selected second of imaging
(white light, feux rouges, blue light) and brightness value, the exposure value for determining selected spectrum, carries out secondary imaging, then calculates and adopt for the second time
Collect the texture information of imaging, including image mean value and variance, according to second of imaging texture information, adaptively selected algorithm (is calculated
Method A, algorithm B, algorithm C), high-accuracy identifies a cigarette laser code, and shows.
The above is only presently preferred embodiments of the present invention, not does limitation in any form to the present invention, it is all according to
According to technical spirit any simple modification to the above embodiments of the invention, equivalent variations, protection of the invention is each fallen within
Within the scope of.
Claims (10)
1. being based on Embedded 32 cigarettes laser code identifying system, it is characterised in that: including light-source system, Image Acquisition system
System, algoritic module, the light-source system are connected with image capturing system, image capturing system join algorithm module；Described
Recognizer selecting module and at least one recognizer module, and image capturing system connection identification are provided in algoritic module
Algorithms selection module, recognizer selecting module are connected at least one recognizer module.
2. according to claim 1 be based on Embedded 32 cigarettes laser code identifying system, it is characterised in that: the knowledge
There are three types of other algoritic module settings, and is respectively recognizer modules A, the recognizer being connected with recognizer selecting module
Module B and recognizer module C, and be additionally provided with respectively with recognizer modules A, recognizer module B and recognizer mould
The result display module that block C is connected.
3. according to claim 1 or 2 be based on Embedded 32 cigarettes laser code identifying system, it is characterised in that: institute
Stating image capturing system includes industrial camera and image processor interconnected, and image processor connection recognizer selects
Module；The light-source system includes control circuit for light source and light source interconnected, and control circuit for light source connects image procossing
Device.
4. the recognition methods as claimed in any one of claims 1 to 3 based on Embedded 32 cigarettes laser code identifying system,
It is characterized by comprising steps in detail below:
1) initial imaging picture is obtained, the texture information of present image is calculated by image capturing system；
2) the preliminary imaging texture information effect of basis, the light source light spectrum and determining selected light of adaptively selected second of imaging
The exposure value of spectrum carries out secondary imaging；
3) texture information of second of imaging is calculated again；
4) according to the texture information of second of imaging, adaptively selected algorithm carries out high-accuracy and identifies a cigarette laser code, and shows
It shows and.
5. the recognition methods according to claim 4 based on Embedded 32 cigarettes laser code identifying system, feature
Be: the texture information includes the mean value of image and the mean square deviation of image；The mean value of image uses formula:It calculates；The mean square deviation of image uses formula:It calculates.
6. the recognition methods according to claim 4 based on Embedded 32 cigarettes laser code identifying system, feature
Be: the light source light spectrum of adaptively selected second of imaging is using following manner:
(1) when mean square deviation is being greater than 4096, white light source is selected, while exposure value takes range: 100-130；
(2) when equal sides is looked into greater than 1024 and when less than 4096, select blue light source, while exposure value takes range: 150-
180；
(3) when mean square deviation is less than 1024, red-light source is selected, while exposure value takes range: 70-90.
7. the recognition methods according to claim 4 based on Embedded 32 cigarettes laser code identifying system, feature
Be: algorithm includes algorithm A, algorithm B and algorithm C in the step 4)；
4.1) when mean square deviation be greater than 4096, selection algorithm A；
4.2) when mean square deviation is greater than 1024 and when less than 4096, selection algorithm B；
4.3) when mean square deviation is less than 1024, selection algorithm C.
8. the recognition methods according to claim 7 based on Embedded 32 cigarettes laser code identifying system, feature
Be: the algorithm A is the OCR recognizer based on SVM.
9. the recognition methods according to claim 7 based on Embedded 32 cigarettes laser code identifying system, feature
Be: the algorithm B is the OCR recognizer based on Adboost.
10. the recognition methods according to claim 7 based on Embedded 32 cigarettes laser code identifying system, feature
Be: the algorithm C is convolutional neural networks recognizer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811009972.1A CN109214222A (en) | 2018-08-31 | 2018-08-31 | Based on Embedded 32 cigarette laser code identifying systems and its recognition methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811009972.1A CN109214222A (en) | 2018-08-31 | 2018-08-31 | Based on Embedded 32 cigarette laser code identifying systems and its recognition methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109214222A true CN109214222A (en) | 2019-01-15 |
Family
ID=64985777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811009972.1A Withdrawn CN109214222A (en) | 2018-08-31 | 2018-08-31 | Based on Embedded 32 cigarette laser code identifying systems and its recognition methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109214222A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111614880A (en) * | 2020-06-02 | 2020-09-01 | 江苏易高烟草机械有限公司 | Cigarette product specification identification method based on visual technology |
| CN113465505A (en) * | 2021-06-28 | 2021-10-01 | 七海测量技术（深圳）有限公司 | Visual detection positioning system and method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1420041A (en) * | 2001-11-21 | 2003-05-28 | 北京汉王科技有限公司 | Embedded integrative vehicle licensing plate distinguishing apparatus |
| CN101009747A (en) * | 2007-01-10 | 2007-08-01 | 刘强 | The method for accurate digit extraction based on multiple OCR scheme combination verification |
| CN103607524A (en) * | 2013-10-18 | 2014-02-26 | 湖南省烟草公司长沙市公司 | Cigarette case 32-bit code image acquisition and processing device and cigarette case 32-bit code identification method |
| CN104146724A (en) * | 2014-08-29 | 2014-11-19 | 重庆邮电大学 | Digital X-ray machine automatic exposure control method and device |
| CN106407863A (en) * | 2016-09-22 | 2017-02-15 | 湘潭大学 | Tobacco carton fast identification device and method |
| CN106971215A (en) * | 2016-01-14 | 2017-07-21 | 北京柯斯元科技有限公司 | A kind of random grain false-proof method, anti-counterfeit recognition system and Antiforge system |
-
2018
- 2018-08-31 CN CN201811009972.1A patent/CN109214222A/en not_active Withdrawn
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1420041A (en) * | 2001-11-21 | 2003-05-28 | 北京汉王科技有限公司 | Embedded integrative vehicle licensing plate distinguishing apparatus |
| CN101009747A (en) * | 2007-01-10 | 2007-08-01 | 刘强 | The method for accurate digit extraction based on multiple OCR scheme combination verification |
| CN103607524A (en) * | 2013-10-18 | 2014-02-26 | 湖南省烟草公司长沙市公司 | Cigarette case 32-bit code image acquisition and processing device and cigarette case 32-bit code identification method |
| CN104146724A (en) * | 2014-08-29 | 2014-11-19 | 重庆邮电大学 | Digital X-ray machine automatic exposure control method and device |
| CN106971215A (en) * | 2016-01-14 | 2017-07-21 | 北京柯斯元科技有限公司 | A kind of random grain false-proof method, anti-counterfeit recognition system and Antiforge system |
| CN106407863A (en) * | 2016-09-22 | 2017-02-15 | 湘潭大学 | Tobacco carton fast identification device and method |
Non-Patent Citations (1)
| Title |
|---|
| 程刚喜: "嵌入式32位激光防伪码识别系统的研究与实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111614880A (en) * | 2020-06-02 | 2020-09-01 | 江苏易高烟草机械有限公司 | Cigarette product specification identification method based on visual technology |
| CN113465505A (en) * | 2021-06-28 | 2021-10-01 | 七海测量技术（深圳）有限公司 | Visual detection positioning system and method |
| CN113465505B (en) * | 2021-06-28 | 2024-03-22 | 七海测量技术（深圳）有限公司 | Visual detection positioning system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ma et al. | GANMcC: A generative adversarial network with multiclassification constraints for infrared and visible image fusion | |
| Fan et al. | LeukocyteMask: An automated localization and segmentation method for leukocyte in blood smear images using deep neural networks | |
| Zhao et al. | An effective automatic system deployed in agricultural Internet of Things using Multi-Context Fusion Network towards crop disease recognition in the wild | |
| Maltezos et al. | Building extraction from LiDAR data applying deep convolutional neural networks | |
| CN111639544B (en) | Expression recognition method based on multi-branch cross-connection convolutional neural network | |
| CN108875821A (en) | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing | |
| CN110263705A (en) | Towards two phase of remote sensing technology field high-resolution remote sensing image change detecting method | |
| CN105740912B (en) | The recognition methods and system of low-rank image characteristics extraction based on nuclear norm regularization | |
| CN109063737A (en) | Image processing method, device, storage medium and mobile terminal | |
| Qadri et al. | Machine vision approach for classification of citrus leaves using fused features | |
| Zeng et al. | Identification of maize leaf diseases by using the SKPSNet-50 convolutional neural network model | |
| CN109684906A (en) | The method of detection red turpentine beetle based on deep learning | |
| Abualigah et al. | Multilevel thresholding image segmentation using meta-heuristic optimization algorithms: Comparative analysis, open challenges and new trends | |
| JP2018136767A (en) | Optical sensor, learning apparatus, and image processing system | |
| CN109214222A (en) | Based on Embedded 32 cigarette laser code identifying systems and its recognition methods | |
| Shi et al. | Diurnal and nocturnal cloud segmentation of all-sky imager (ASI) images using enhancement fully convolutional networks | |
| CN109190458A (en) | A kind of person of low position's head inspecting method based on deep learning | |
| Liu et al. | Farmland aerial images fast-stitching method and application based on improved sift algorithm | |
| CN109800812A (en) | CT image classification feature selection approach and system based on counterfeit filter | |
| Zhang et al. | Robust image segmentation method for cotton leaf under natural conditions based on immune algorithm and PCNN algorithm | |
| CN110472495B (en) | Deep learning face recognition method based on graphic reasoning global features | |
| Lu et al. | Tobacco leaf grading based on deep convolutional neural networks and machine vision | |
| Zambare et al. | A theoretical study on cnn based cotton crop disease detection | |
| Guo et al. | Dual-concentrated network with morphological features for tree species classification using hyperspectral image | |
| Sosa et al. | An algorithm for detection of nutritional deficiencies from digital images of coffee leaves based on descriptors and neural networks |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190115 |