CN108090434A - A kind of ore method for quickly identifying - Google Patents
A kind of ore method for quickly identifying Download PDFInfo
- Publication number
- CN108090434A CN108090434A CN201711326240.0A CN201711326240A CN108090434A CN 108090434 A CN108090434 A CN 108090434A CN 201711326240 A CN201711326240 A CN 201711326240A CN 108090434 A CN108090434 A CN 108090434A
- Authority
- CN
- China
- Prior art keywords
- ore
- image
- quickly identifying
- notable
- training
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
A kind of ore method for quickly identifying, including：S1, obtain ore training image and based on deep learning the ore training image is trained to obtain multiple modeling coefficients；S2, processing model is built based on the multiple modeling coefficients；S3, based on the processing model measurement Ore Image to be measured to obtain mask image；S4, Ore Image is obtained based on the Ore Image to be measured and the mask image to identify ore.Implement the ore method for quickly identifying and computer readable storage medium of the present invention, it, can be with very high precision rapidly and accurately automatic identification ore by the processing Model Identification ore obtained based on deep learning training.
Description
Technical field
The present invention relates to ores to identify field, more specifically to a kind of ore method for quickly identifying.
Background technology
Tungsten belongs to high-melting-point rare metal or refractory metal in metallurgical and metal material field.Tungsten and its alloy are
One of particularly important functional material in modern industry, national defence and high-technology, be widely used in space flight, atomic energy,
The numerous areas such as ship, auto industry, electrical industry, electronics industry, chemical industry.In the prior art, generally use picking
Tungsten ore ore is identified selecting in mode, thus there are low production efficiency, of high cost, labor strength is big and sorting index is low
The defects of.
The content of the invention
The technical problem to be solved in the present invention is, for the drawbacks described above of the prior art, provides a kind of ore and quickly knows
Other method uses processor that can carry out ore automated optical identification, therefore production efficiency is high, at low cost and sorting index is high.
The technical solution adopted by the present invention to solve the technical problems is：A kind of ore method for quickly identifying is constructed, including：
S1, obtain ore training image and be to obtain multiple modelings based on the deep learning training ore training image
Number；
S2, processing model is built based on the multiple modeling coefficients；
S3, based on the processing model measurement Ore Image to be measured to obtain mask image；
S4, Ore Image is obtained based on the Ore Image to be measured and the mask image to identify ore.
In this hair ore method for quickly identifying, the step S1 further comprises：
S11, obtain ore training image and extract the characteristic properties of the ore training image and ore space coordinates；
S12, the property notable figure of the ore training image is obtained based on the characteristic texture and based on the ore
Space coordinates obtains the position notable figure of the ore training image；
S13, the multiple modeling coefficients are calculated based on the property notable figure and the position notable figure.
In this hair ore method for quickly identifying, the step S12 further comprises：
S121, multiple characteristic properties structure ore instructions are based on using gaussian pyramid and central peripheral difference algorithm
Practice multiple characteristic properties figures of image；
S122, using intersecting, scale combines and normalization operator is based on multiple characteristic properties figures acquisitions ore and instructs
Practice multiple property notable figures of image；And
S123, shown using position of the dimensional gaussian distribution based on the ore space coordinates acquisition ore training image
Write figure.
In this hair ore method for quickly identifying, the step S13 further comprises：
S131, multiple property notable figures and the position notable figure meter are based on using scale invariant feature transfer algorithm
Calculate the multiple modeling coefficients.
In this hair ore method for quickly identifying, the characteristic properties include：Color, brightness, transparency and anti-
Penetrate rate.
In this hair ore method for quickly identifying, the step S2 further comprises：
S21, the multiple modeling coefficients structure processing model is based on using convolutional neural networks, wherein the convolution
Neutral net uses ReLU activation primitives.
In this hair ore method for quickly identifying, the step S3 further comprises:
S31, Ore Image to be measured is obtained using the scanning of CCD area array cameras；
S32, Ore Image to be measured described in the processing model measurement is used to generate test notable figure；
Notable figure is tested described in S33, optimization processing to generate the mask image；
Wherein described optimization processing includes Threshold segmentation, Morphological scale-space and median filter process.
In this hair ore method for quickly identifying, the step S4 further comprises
S41, corner points detection is carried out to the Ore Image to be measured to obtain corner points feature point set；
S42, the corner points feature point set and the mask image are multiplied to obtain Ore Image；
S43, processing is split to the Ore Image to obtain Ore Image region and background area；
S44, by the Ore Image region recognition be ore.
In this hair ore method for quickly identifying, the step S4 further comprises:
S45, the movement position that the position in the Ore Image region is converted into the ore；
S46, driving injection apparatus are based on the movement position and spray the ore.
It is to construct a kind of computer readable storage medium that the present invention, which solves another technical solution that its technical problem uses,
It is stored thereon with computer program, which is characterized in that realize that the ore quickly identifies when described program is executed by processor
Method.
Implement the ore method for quickly identifying and computer readable storage medium of the present invention, trained by being based on deep learning
The processing Model Identification ore of acquisition, can be with very high precision rapidly and accurately automatic identification ore.Further, by adopting
More ore characteristic properties can be obtained in synchronization with CCD area array cameras, improve spatial resolution, treating capacity is big.Again
It further, can be more efficiently right by handling model based on the convolutional neural networks structure using ReLU activation primitives
Even if the Ore Image to be measured be subjected to displacement, deformed is identified, and then further improves the accuracy of identification.
Description of the drawings
Below in conjunction with accompanying drawings and embodiments, the invention will be further described, in attached drawing：
Fig. 1 is the flow chart of the first embodiment of the ore method for quickly identifying of the present invention；
Fig. 2 is the flow chart of the second embodiment of the ore method for quickly identifying of the present invention；
Fig. 3 is the model structure of the convolutional neural networks used in ore method for quickly identifying shown in Fig. 2；
Fig. 4 A-4E are the effect signals using the ore of the first mining site of ore method for quickly identifying shown in Fig. 2 identification
Figure；
Fig. 5 A-5D are the effect signals using the ore of the second mining site of ore method for quickly identifying shown in Fig. 2 identification
Figure.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, it is right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The present invention relates to a kind of ore method for quickly identifying.Various aspects of the invention include obtaining ore training image simultaneously
Based on deep learning the ore training image is trained to obtain multiple modeling coefficients；At the multiple modeling coefficients structure
Manage model；And Ore Image is obtained to identify ore based on the Ore Image to be measured and the mask image.Implement this hair
Bright ore method for quickly identifying and computer readable storage medium are known by the processing model obtained based on deep learning training
Other ore, can be with very high precision rapidly and accurately automatic identification ore.
Fig. 1 is the flow chart of the first embodiment of the ore method for quickly identifying of the present invention.As shown in Figure 1, the present invention
Ore method for quickly identifying comprises the following steps.In step sl, obtain ore training image and institute is trained based on deep learning
Ore training image is stated to obtain multiple modeling coefficients.In a preferred embodiment of the invention, industrial camera may be employed,
Such as area array cameras and line-scan digital camera are shot.Then the ore training image of shooting is passed through into kilomega network, camera
The directly incoming processor of the interfaces such as link, USB3.0, such as computer.Then, computer utilizes these ore training images, base
The ore training image is trained in deep learning, so as to construct required modeling coefficients.It such as can be to belonging to different ore deposits
The ore training image on mountain is largely sampled, and establishes to self-organizing the color on building stones to be measured, brightness, transparency and anti-
Penetrate whether the characteristic parameters such as rate and the building stones to be measured are optional relation between ore, by learning and training, build required
Modeling coefficients.
In the step S2, processing model is built based on the multiple modeling coefficients.For example, one in the present invention is excellent
It selects in embodiment, convolutional neural networks may be employed and be based on the multiple modeling coefficients structure processing model.In the present invention
In, due to for the identification of ore being handled using problem of image recognition as an energy function minimization problem, energy function
Determine network structure.The basic thought of image identification segmentation based on neutral net is obtained by training multi-layer perception (MLP)
Then linear decision function classifies pixel with decision function to achieve the purpose that segmentation.Neutral net is generally with instruction
Practice sample to be trained it.Image identification is considered as a constraint satisfaction problemx (CSP), and utilizes constraint satisfaction nerve net
Network solves.Neocognitron is more much better than than cognitron to the identification of mode signal, no matter the conversion of signal, conversion, distortion, big
Small change can be handled.One visual pattern is resolved into many subpatterns (feature) by neocognitron, subsequently into point
Layer is passed the characteristic plane that stepwise is connected and is handled, it is attempted vision system model, even if can have position in object
When shifting or slight deformation, it can also complete to identify.Usual neocognitron includes two class neurons, that is, undertakes feature extraction
The C- members of S- members and resistance to deformation.Two important parameters, i.e. receptive field and threshold parameter involved in S- members, the former determines input connection
Number, the latter then controls the extent of reaction to feature subpattern.It therefore, in a preferred embodiment of the invention, can be with
Using convolutional neural networks LeNet-5.
In step s3, based on the processing model measurement Ore Image to be measured to obtain mask image.The present invention's
In one preferred embodiment, industrial camera may be employed, such as area array cameras and line-scan digital camera are shot.Then by shooting
Image passes through the directly incoming processor of the interfaces such as kilomega network, camera link, USB3.0, such as computer.It then can be by this
A little images are as Ore Image to be measured.By these Ore Images to be measured using it is described processing model measurement after, it is possible to directly
Generate mask image.Certainly, in other preferred embodiments of the present invention, processing can also be optimized to the image of generation,
Then again using the image after optimization as mask image.
In step s 4, Ore Image is obtained based on the Ore Image to be measured and the mask image to identify ore.
In a preferred embodiment of the invention, Ore Image to be tested can be pre-processed, can also be directly treated described
Survey Ore Image is multiplied with the mask image directly obtains Ore Image.At this moment can by the direct dividing processing of Ore Image into
Ore Image region and background area.Then it is ore by Ore Image region recognition.In the further preferred reality of the present invention
It applies in example, the position in the Ore Image region can also be converted into the movement position of the ore, then driving injection dress
It puts and the ore is sprayed based on the movement position.
Those skilled in the art know, in the present invention, establish the step S1-S2 of processing model and test identification step
It is not absolutely required to consecutive order execution by S3-S4.For example, step S1-S2 can be first carried out to obtain suitable processing model,
Then the processing model is recycled and reused for performing step S3-S4 to complete ore identification.
Implement the ore method for quickly identifying of the present invention, pass through the processing Model Identification ore deposit obtained based on deep learning training
Stone, can be with very high precision rapidly and accurately automatic identification ore.
Fig. 2 is the flow chart of the second embodiment of the ore method for quickly identifying of the present invention.As shown in Fig. 2, in step S1
In, it obtains ore training image and extracts the characteristic properties of the ore training image and ore space coordinates.In the present invention
A preferred embodiment in, industrial camera may be employed, such as area array cameras and line-scan digital camera are shot.It then will shooting
Ore training image pass through the directly incoming processor of the interfaces such as kilomega network, camera link, USB3.0, such as computer.So
Afterwards, the characteristic properties of the ore training image and ore space coordinates can be extracted respectively.Such as can extract one, two
A, three or multiple characteristic properties.The characteristic properties can include color, brightness, transparency and reflectivity etc. feature
One of parameter or more.In a preferred embodiment of the invention, multiple characteristic properties can be extracted simultaneously with multithreading
And ore space coordinates.
In step s 2, the property notable figure of the ore training image is obtained and based on described based on the characteristic properties
Ore space coordinates obtains the position notable figure of the ore training image.It in a preferred embodiment of the invention, can be with
Gaussian pyramid is respectively adopted in simultaneous multi-threading and central peripheral difference algorithm is based on multiple characteristic properties and builds the ore
Multiple characteristic properties figures of training image.Then, using intersecting, scale combines and normalization operator is based on multiple characteristics
Matter figure obtains multiple property notable figures of the ore training image.At the same time or afterwards, two-dimentional height may be employed
This distribution obtains the position notable figure of the ore training image based on the ore space coordinates.
In step s3, the multiple modeling coefficients are calculated based on the property notable figure and the position notable figure.
In a preferred embodiment of the present invention, may be employed scale invariant feature transfer algorithm be based on multiple property notable figures and
The position notable figure calculates the multiple modeling coefficients.
In step s 4, the multiple modeling coefficients are based on using convolutional neural networks and build the processing model.At this
In embodiment, the convolutional neural networks use ReLU activation primitives.Fig. 3 is adopted in ore method for quickly identifying shown in Fig. 2
The model structure of convolutional neural networks, as shown in figure 3, the convolutional neural networks from input layer and output layer from front to back
Respectively conv1, pool1, conv2, pool2, interior lamination 1, ReLU activation primitives, interior lamination 2.Input data is big for 64
It is small be 256*256 ore training image be one group (sharing 64*3*256*256 element)；Conv1 layers are read the input data
Convolution algorithm is carried out, filter (namely convolution kernel) size is 5*5 in conv1 layers, and step-length 1, exporting 20 sizes is
The characteristic pattern (sharing 64*20*252*252 element) of 252*252；Conv1 to pool1 layers carry out maximum pond, output characteristic
For the width and Gao Chihua of figure to the half size of last layer, characteristic pattern quantity is constant, exports the characteristic pattern that 20 sizes are 126*126
(sharing 64*20*126*126 element)；Similarly, conv2 exports characteristic pattern (the shared 64*50* that 50 sizes are 122*122
122*122 element)；Pool2 exports the characteristic pattern (sharing 64*50*61*61 element) that 50 sizes are 61*61.Interior lamination
1 500 characteristic patterns (common 64*500 element) of output；Afterwards by ReLU, number of elements does not change；Interior lamination 2 exports spy
Sign figure be N (common 64*N, N are integer) more than or equal to 2, it is intended that expression present networks model was done is that N- classifies, finally with
SoftMaxWithLoss functions result of calculation is as output result.
In the present embodiment, the activation primitive used is ReLU, certainly in other embodiments of the invention, can also adopt
By the use of sigmoid as activation primitive.In the present invention, not only it is only that using the advantage of ReLU activation primitives and effectively avoids part
Optimal problem, it is also possible that output layer, the data sample allowed in output layer become linear separability to the end by input data mapping.
In step s 5, Ore Image to be measured is obtained using the scanning of CCD area array cameras.One in the present invention is preferably implemented
In example, industrial camera may be employed, such as area array cameras and line-scan digital camera are shot.Then the image of shooting is passed through into gigabit
The directly incoming processor of the interfaces such as net, camera link, USB3.0, such as computer.It then can be using these images as treating
Survey Ore Image.
In step s 6, use Ore Image to be measured described in the processing model measurement obtained in step S4 and surveyed with generating
Try notable figure.In a preferred embodiment of the invention, interpretation of result and vision optimization processing model may be employed.Ability
Field technique personnel any of processing method may be employed utilize it is described processing model measurement described in Ore Image to be measured with
Generation test notable figure.
In the step s 7, notable figure is tested described in optimization processing to generate the mask image.One in the present invention is excellent
It selects in embodiment, Threshold segmentation, Morphological scale-space and median filter process may be employed and optimize processing.In its of the present invention
In his preferred embodiment, the processing method that other can also be used suitable carries out related optimization.
In step s 8, corner points detection is carried out to the Ore Image to be measured obtained from step S5 to obtain corner
Point feature point set.Certainly, in other preferred embodiments of the present invention, the Ore Image to be measured can not be handled,
It can also carry out other kinds of pretreatment.
In step s 9, by the corner points feature point set obtained from step S8 with obtain from the step S7 described in
Mask image is multiplied to obtain Ore Image.
In step slo, can the Ore Image be split processing to obtain Ore Image region and background area.
In the present invention, any image segmentation algorithm as known in the art, such as water ridge partitioning algorithm, pyramid can be utilized to split
Algorithm and mean shift segmentation algorithm etc. carry out image segmentation.
In step s 11, can be ore by the Ore Image region recognition.In a preferred embodiment of the invention,
The position in the Ore Image region can also be converted into the movement position of the ore, then injection apparatus is driven to be based on institute
It states movement position and sprays the ore.In a preferred embodiment of the invention, region and the reality of ore will can also be identified as
Situation is compared, the step of so as to calculate the accuracy rate of ore method for quickly identifying.
Fig. 4 A-4E are the effect signals using the ore of the first mining site of ore method for quickly identifying shown in Fig. 2 identification
Figure.Fig. 5 A-5D are the effect diagrams using the ore of the second mining site of ore method for quickly identifying shown in Fig. 2 identification.Such as
Shown in Fig. 4 A-5D, although image is subjected to displacement, deformation also can accurately be identified.
Those skilled in the art know, as previously mentioned, for above-mentioned steps S1-S11, in addition to defined otherwise in text, hold
Row order can be upset, and can also perform simultaneously or backward performs, can not also continuously perform, but be spaced and perform.
Implement the ore method for quickly identifying of the present invention, pass through the processing Model Identification ore deposit obtained based on deep learning training
Stone, can be with very high precision rapidly and accurately automatic identification ore.It further, can be by using CCD area array cameras
Synchronization obtains more ore characteristic properties, improves spatial resolution, the smaller ore of grade can be sorted, and locates
Reason amount is big, it can be achieved that 40t/h,.Yet further, by based on the convolutional neural networks structure processing using ReLU activation primitives
Model even if can more efficiently pair be subjected to displacement, the Ore Image to be measured that deforms is identified, and then further improves knowledge
Other accuracy.
An alternative embodiment of the invention provides a kind of machine-readable memory and/or storage medium, the machine of memory storage
Code and/or computer program include at least one code segment, by machine and/or computer perform and cause the machine and/or
Computer performs each step of ore method for quickly identifying described in this application.
In the present invention, by carrying out global design to convolutional neural networks, suitable activation primitive and Chi Huafang are defined
Case using more GPU Parallel Designs thinkings, realizes the fast recognition technology of ore.In addition, it is obtained using the scanning of CCD area array cameras
Ore Image is taken, more mineral surface characteristic informations can be obtained in synchronization, spatial resolution is increased to 0.05/
mm.In addition the disaggregated model of ore and non-ore is constructed by convolutional neural networks, and devises the Millisecond based on FPGA
Image processing engine realizes in the case of free fall type feed-type, fast and accurately sub-elects containing lode ore, have
Higher precision.
Therefore, the present invention can be by hardware, software or soft and hardware with reference to realizing.The present invention can be at least one
It realizes in a computer system or is divided by being distributed in the different piece in the computer system of several interconnection in a centralised manner
Scattered mode is realized.Any computer system that can realize the method for the present invention or miscellaneous equipment are all applicatory.It is common soft or hard
The combination of part can be equipped with the general-purpose computing system of computer program, by installing and performing program-con-trolled computer system
System, makes it be run by the method for the present invention.
The present invention can also be implemented by computer program product, and program, which includes, can realize the complete of the method for the present invention
Portion's feature when it is installed in computer system, can realize method of the invention.Computer program in this document is signified
Be：Any expression formula for one group of instruction that any program language, code or symbol are write may be employed, which makes system
With information processing capability, to be directly realized by specific function or realize specific work(after one or two following step are carried out
Energy：A) other Languages, coding or symbol are converted into；B) reproduce in a different format.
Although the present invention is illustrated by specific embodiment, it will be appreciated by those skilled in the art that, it is not departing from
In the case of the scope of the invention, various conversion and equivalent substitute can also be carried out to the present invention.In addition, for particular condition or material
Material, can do various modifications, without departing from the scope of the present invention to the present invention.Therefore, the present invention is not limited to disclosed tool
Body embodiment, and the whole embodiments fallen within the scope of the appended claims should be included.
Claims (10)
1. a kind of ore method for quickly identifying, which is characterized in that including：
S1, obtain ore training image and based on deep learning the ore training image is trained to obtain multiple modeling coefficients；
S2, processing model is built based on the multiple modeling coefficients；
S3, based on the processing model measurement Ore Image to be measured to obtain mask image；
S4, Ore Image is obtained based on the Ore Image to be measured and the mask image to identify ore.
2. ore method for quickly identifying according to claim 1, which is characterized in that the step S1 further comprises：
S11, obtain ore training image and extract the characteristic properties of the ore training image and ore space coordinates；
S12, the property notable figure of the ore training image is obtained based on the characteristic texture and based on the ore space
Coordinate obtains the position notable figure of the ore training image；
S13, the multiple modeling coefficients are calculated based on the property notable figure and the position notable figure.
3. ore method for quickly identifying according to claim 2, which is characterized in that the step S12 further comprises：
S121, multiple characteristic properties structure ore training figures are based on using gaussian pyramid and central peripheral difference algorithm
Multiple characteristic properties figures of picture；
S122, using intersecting, scale combines and normalization operator is based on multiple characteristic properties figures acquisition ore training and schemes
Multiple property notable figures of picture；
S123, the position for obtaining the ore training image based on the ore space coordinates using dimensional gaussian distribution are notable
Figure.
4. ore method for quickly identifying according to claim 3, which is characterized in that the step S13 further comprises：
S131, multiple property notable figures and position notable figure calculating institute are based on using scale invariant feature transfer algorithm
State multiple modeling coefficients.
5. ore method for quickly identifying according to claim 4, which is characterized in that the characteristic properties include：It is color, bright
Degree, transparency and reflectivity.
6. the ore method for quickly identifying according to any one claim in claim 1-5, which is characterized in that described
Step S2 further comprises：
S21, the multiple modeling coefficients structure processing model is based on using convolutional neural networks, wherein the convolutional Neural
Network uses ReLU activation primitives.
7. ore method for quickly identifying according to claim 6, which is characterized in that the step S3 further comprises：
S31, Ore Image to be measured is obtained using the scanning of CCD area array cameras；
S32, Ore Image to be measured described in the processing model measurement is used to generate test notable figure；
Notable figure is tested described in S33, optimization processing to generate the mask image；
Wherein described optimization processing includes Threshold segmentation, Morphological scale-space and median filter process.
8. ore method for quickly identifying according to claim 6, which is characterized in that the step S4 further comprises:
S41, corner points detection is carried out to the Ore Image to be measured to obtain corner points feature point set；
S42, the corner points feature point set and the mask image are multiplied to obtain Ore Image；
S43, processing is split to the Ore Image to obtain Ore Image region and background area；
S44, by the Ore Image region recognition be ore.
9. ore method for quickly identifying according to claim 7, which is characterized in that the step S4 further comprises:
S45, the movement position that the position in the Ore Image region is converted into the ore；
S46, driving injection apparatus are based on the movement position and spray the ore.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is processed
The ore method for quickly identifying according to any one claim in claim 1-9 is realized when device performs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711326240.0A CN108090434B (en) | 2017-12-13 | 2017-12-13 | Rapid ore identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711326240.0A CN108090434B (en) | 2017-12-13 | 2017-12-13 | Rapid ore identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108090434A true CN108090434A (en) | 2018-05-29 |
| CN108090434B CN108090434B (en) | 2020-06-23 |
Family
ID=62175225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711326240.0A Active CN108090434B (en) | 2017-12-13 | 2017-12-13 | Rapid ore identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108090434B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109284780A (en) * | 2018-09-10 | 2019-01-29 | 中山大学 | Ore mineral image automatic identification and classification method |
| CN109800728A (en) * | 2019-01-28 | 2019-05-24 | 济南浪潮高新科技投资发展有限公司 | A kind of minerals based on deep learning quickly know method for distinguishing |
| CN110000116A (en) * | 2019-04-19 | 2019-07-12 | 福建铂格智能科技股份公司 | A kind of freely falling body fruits and vegetables method for separating and system based on deep learning |
| CN110458119A (en) * | 2019-08-15 | 2019-11-15 | 中国水利水电科学研究院 | A kind of aggregate gradation method for quickly identifying of non-contact measurement |
| CN110706267A (en) * | 2019-09-04 | 2020-01-17 | 五邑大学 | Mining process-based ore three-dimensional coordinate acquisition method and device |
| CN110930393A (en) * | 2019-11-28 | 2020-03-27 | 兰州石化职业技术学院 | Chip material pipe counting method, device and system based on machine vision |
| CN112317362A (en) * | 2020-09-24 | 2021-02-05 | 赣州好朋友科技有限公司 | Method and device for sorting quartz associated gold ore and readable storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6988610B2 (en) * | 2002-01-14 | 2006-01-24 | Carnegie Mellon University | Conveyor belt inspection system and method |
| CN102521572A (en) * | 2011-12-09 | 2012-06-27 | 中国矿业大学 | Image recognition method of coal and gangue |
| CN103530621A (en) * | 2013-11-04 | 2014-01-22 | 中国矿业大学(北京) | Coal and rock image identification method based on back propagation (BP) neural network |
| CN103886290A (en) * | 2014-03-13 | 2014-06-25 | 西安交通大学 | Coal flow video detection method under complex background |
| CN103984940A (en) * | 2014-06-03 | 2014-08-13 | 核工业北京地质研究院 | Method for identifying hematitization based on hyperspectral data |
| CN104657988A (en) * | 2015-02-04 | 2015-05-27 | 中国矿业大学 | Image segmentation method for micro-fine cohesive core particles based on angular point and curvature detection |
| CN106204570A (en) * | 2016-07-05 | 2016-12-07 | 安徽工业大学 | A kind of angular-point detection method based on non-causal fractional order gradient operator |
| CN103927528B (en) * | 2014-05-05 | 2017-02-01 | 中国矿业大学(北京) | Coal and rock recognition method based on close neighborhood pixel gray level joint distribution characteristics |
-
2017
- 2017-12-13 CN CN201711326240.0A patent/CN108090434B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6988610B2 (en) * | 2002-01-14 | 2006-01-24 | Carnegie Mellon University | Conveyor belt inspection system and method |
| CN102521572A (en) * | 2011-12-09 | 2012-06-27 | 中国矿业大学 | Image recognition method of coal and gangue |
| CN103530621A (en) * | 2013-11-04 | 2014-01-22 | 中国矿业大学(北京) | Coal and rock image identification method based on back propagation (BP) neural network |
| CN103886290A (en) * | 2014-03-13 | 2014-06-25 | 西安交通大学 | Coal flow video detection method under complex background |
| CN103927528B (en) * | 2014-05-05 | 2017-02-01 | 中国矿业大学(北京) | Coal and rock recognition method based on close neighborhood pixel gray level joint distribution characteristics |
| CN103984940A (en) * | 2014-06-03 | 2014-08-13 | 核工业北京地质研究院 | Method for identifying hematitization based on hyperspectral data |
| CN104657988A (en) * | 2015-02-04 | 2015-05-27 | 中国矿业大学 | Image segmentation method for micro-fine cohesive core particles based on angular point and curvature detection |
| CN106204570A (en) * | 2016-07-05 | 2016-12-07 | 安徽工业大学 | A kind of angular-point detection method based on non-causal fractional order gradient operator |
Non-Patent Citations (6)
| Title |
|---|
| 周龙萍: "基于改进的Harris算法检测角点", 《计算机技术与发展》 * |
| 张永梅: "基于生物视觉机制的图像感兴趣区域提取研究", 《中国优秀硕士学位论文全文数据库》 * |
| 田慧卿: "图像识别技术在煤岩识别应用中的研究", 《中国优秀硕士学位论文全文数据库》 * |
| 程潭武，等: "混合像元分解法在植被覆盖区矿化蚀变信息提取中的应用——以江西大浩山金矿区为例", 《地质学刊》 * |
| 高隽: "图像语义分析与理解综述", 《模式识别与人工智能》 * |
| 龚涛: "《摄影测量学》", 31 December 2014, 西南交通大学出版社 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109284780A (en) * | 2018-09-10 | 2019-01-29 | 中山大学 | Ore mineral image automatic identification and classification method |
| CN109800728A (en) * | 2019-01-28 | 2019-05-24 | 济南浪潮高新科技投资发展有限公司 | A kind of minerals based on deep learning quickly know method for distinguishing |
| CN110000116A (en) * | 2019-04-19 | 2019-07-12 | 福建铂格智能科技股份公司 | A kind of freely falling body fruits and vegetables method for separating and system based on deep learning |
| CN110458119A (en) * | 2019-08-15 | 2019-11-15 | 中国水利水电科学研究院 | A kind of aggregate gradation method for quickly identifying of non-contact measurement |
| CN110706267A (en) * | 2019-09-04 | 2020-01-17 | 五邑大学 | Mining process-based ore three-dimensional coordinate acquisition method and device |
| CN110706267B (en) * | 2019-09-04 | 2022-11-15 | 五邑大学 | Mining process-based ore three-dimensional coordinate acquisition method and device |
| CN110930393A (en) * | 2019-11-28 | 2020-03-27 | 兰州石化职业技术学院 | Chip material pipe counting method, device and system based on machine vision |
| CN112317362A (en) * | 2020-09-24 | 2021-02-05 | 赣州好朋友科技有限公司 | Method and device for sorting quartz associated gold ore and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108090434B (en) | 2020-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108090434A (en) | A kind of ore method for quickly identifying | |
| CN113160192B (en) | Visual sense-based snow pressing vehicle appearance defect detection method and device under complex background | |
| CN108416266B (en) | Method for rapidly identifying video behaviors by extracting moving object through optical flow | |
| CN106446894B (en) | A method of based on outline identification ball-type target object location | |
| CN107123106A (en) | Show the image processing apparatus of the object detected from input picture | |
| US8320682B2 (en) | Evaluation of edge direction information | |
| EP0363828A2 (en) | Method and apparatus for adaptive learning type general purpose image measurement and recognition | |
| CN110148162A (en) | A kind of heterologous image matching method based on composition operators | |
| CN107230203A (en) | Casting defect recognition methods based on human eye vision attention mechanism | |
| CN108470338B (en) | A kind of water level monitoring method | |
| CN111414931B (en) | Multi-branch multi-scale small target detection method based on image depth | |
| CN109523583B (en) | Infrared and visible light image registration method for power equipment based on feedback mechanism | |
| CN104751475B (en) | A kind of characteristic point Optimum Matching method towards still image Object identifying | |
| CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
| CN110414571A (en) | A kind of website based on Fusion Features reports an error screenshot classification method | |
| CN108108678B (en) | Tungsten ore identification and separation method | |
| CN111553422A (en) | Automatic identification and recovery method and system for surgical instruments | |
| CN109740609A (en) | A kind of gauge detection method and device | |
| CN111665199A (en) | Wire and cable color detection and identification method based on machine vision | |
| Meng et al. | A robust and efficient method for license plate recognition | |
| CN111339974B (en) | Method for identifying modern ceramics and ancient ceramics | |
| CN117576079A (en) | Industrial product surface abnormality detection method, device and system | |
| Kusanti et al. | Combination of otsu and canny method to identify the characteristics of solo batik as Surakarta traditional batik | |
| CN113252103A (en) | Method for calculating volume and mass of material pile based on MATLAB image recognition technology | |
| CN110334727B (en) | Intelligent matching detection method for tunnel cracks |
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 | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A Fast Identification Method for Ores Effective date of registration: 20230803 Granted publication date: 20200623 Pledgee: Everbright Xinglong Trust Co.,Ltd. Pledgor: GANZHOU GOOD FRIEND TECHNOLOGY Co.,Ltd. Registration number: Y2023980050931 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |