CN109858504A - A kind of image-recognizing method, device, system and calculate equipment - Google Patents
A kind of image-recognizing method, device, system and calculate equipment Download PDFInfo
- Publication number
- CN109858504A CN109858504A CN201711240738.5A CN201711240738A CN109858504A CN 109858504 A CN109858504 A CN 109858504A CN 201711240738 A CN201711240738 A CN 201711240738A CN 109858504 A CN109858504 A CN 109858504A
- Authority
- CN
- China
- Prior art keywords
- template
- image
- matching
- tested
- subgraphs
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000012545 processing Methods 0.000 claims abstract description 70
- 238000005065 mining Methods 0.000 claims description 12
- 238000004891 communication Methods 0.000 claims description 7
- 235000013399 edible fruits Nutrition 0.000 claims description 7
- 238000003909 pattern recognition Methods 0.000 claims description 7
- 230000004044 response Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 abstract description 20
- 238000004422 calculation algorithm Methods 0.000 description 14
- 230000002441 reversible effect Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000004380 ashing Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 241000238876 Acari Species 0.000 description 1
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
This application provides a kind of image-recognizing methods, this method comprises: server obtains different template subgraphs by zooming in and out processing to template image, it recycles at least two template subgraphs to be matched respectively with tested image, finally determines final matching results using matched matching result at least twice;It can be seen that, when size is inconsistent in the target and template image being detected in image, server obtains multiple template subgraph by zooming in and out processing to template image, it is matched again using this multiple template subgraph as template with tested image, this allows for template has Dimensional variability in the matching process, so that finally participating in adjoining dimensions of the matched template as far as possible with target in tested image, a matching process based on a template image is transformed into the multiple matching process based on multiple template subgraph by server, final matching results are determined by multiple matched matching result, to improve the precision of identification.
Description
Technical field
This application involves image identification technical field, in particular to a kind of image-recognizing method, device, system and calculating
Equipment.
Background technique
With the development of image recognition technology, template identification is applied and is being counted as representative method in image recognition
Be used widely in the multiple fields such as calculation machine field, e-commerce field, medicine, but in every field business development get over
Come it is faster, to template identification accuracy requirement it is also higher and higher.
And traditional template matching algorithm is to directly determine matching knot by a template matching based on a template image
Fruit;Conventional template matching algorithm is only applicable to target size and the consistent this scene of template image in tested image, traditional mould
Plate matching algorithm is it is not intended that target size and the inconsistent scene of template image size in tested image, in such a scenario
Accuracy of identification tend not to meet practical business demand.
Summary of the invention
In order to improve the precision of image recognition, this application provides a kind of image-recognizing method, this method passes through to mould
Plate image zooms in and out processing, and the matching based on template image is transformed into based at least two moulds after template image scaling
The matching of plank image, so that final matching results are determined using at least two matching results, to improve accuracy of identification.
In addition, present invention also provides a kind of image recognitions in order to guarantee that the above method is realization and application in practice
Device, system and calculating equipment image recognition.
Technical solution provided by the present application is specific as follows:
A kind of image identification system is provided in the application first aspect, which includes:
Server zooms in and out processing to the template image and obtains at least for determining tested image and template image
Two template subgraphs carry out template matching to the tested image using at least two template subgraphs and obtain and each template
The corresponding matching result of subgraph determines final matching results according at least two matching results;
Client, for showing the final matching results on the tested image.
A kind of image-recognizing method is provided in the application second aspect, comprising:
Determine tested image and template image;
Processing is zoomed in and out to the template image and obtains at least two template subgraphs;
Template matching is carried out to the tested image respectively using at least two template subgraphs to obtain and each template
The corresponding matching result of image;
Final matching results are determined according at least two matching results.
A kind of pattern recognition device is provided in the application third aspect, comprising:
First determining module, for determining tested image and template image;
Scaling processing module obtains at least two template subgraphs for zooming in and out processing to the template image;
Matching module is obtained for carrying out template matching to the tested image respectively using at least two template subgraphs
Matching result corresponding with each template subgraph;
Second determining module, for determining final matching results according at least two matching results.
A kind of calculating equipment is provided in the application fourth aspect, which includes:
Processor, memory, network interface and bus system.
The bus system, for each hardware component for calculating equipment to be coupled;
The network interface, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory, for storing program instruction;
The processor, for reading the instruction stored in the memory, the following operation of execution:
Determine tested image and template image;
Processing is zoomed in and out to the template image and obtains at least two template subgraphs;
Template matching is carried out to the tested image respectively using at least two template subgraphs to obtain and each template
The corresponding matching result of image;
Final matching results are determined according at least two matching results.
A kind of image-recognizing method is provided in the application fourth aspect, comprising:
Show user configuration interface;
Configuration information is sent to server;The configuration information is used to indicate server and zooms in and out processing to template image
At least two template subgraphs are obtained, and template matching is carried out to tested image according at least two template subgraphs;Receive clothes
The final matching results for device feedback of being engaged in;
Position of the template image in the tested image is marked according to the final matching results.In the application
Six aspects provide a kind of pattern recognition device, comprising:
First display module, for showing user configuration interface;
Sending module, for sending configuration information to server;The configuration information is used to indicate server to Prototype drawing
At least two template subgraphs are obtained as zooming in and out processing, and mould is carried out to tested image according at least two template subgraphs
Plate matching;Receiving module, for receiving the final matching results of server feedback;
Second display module, for marking the figure being identified in the tested image according to the final matching results
As region.A kind of calculating equipment is provided at the 7th aspect of the application, comprising:
Processor, memory, network interface and bus system.
The bus system, for each hardware component for calculating equipment to be coupled;
The network interface, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory, for storing program instruction;
The processor, for reading the instruction stored in the memory, the following operation of execution:
Show user configuration interface;
Configuration information is sent to server;The configuration information is used to indicate server and zooms in and out processing to template image
At least two template subgraphs are obtained, and template matching is carried out to tested image according at least two template subgraphs;Receive clothes
The final matching results for device feedback of being engaged in;
The image-region being identified is marked in the tested image according to the final matching results.With the prior art
It compares, the application has the advantages that
In this application, server obtains different template subgraphs, then benefit by zooming in and out processing to template image
It is matched with tested image at least two template subgraphs, is finally determined using matched matching result at least twice respectively
Final matching results out;As it can be seen that server passes through to mould when size is inconsistent in the target and template image being detected in image
Plate image zooms in and out processing and obtains multiple template subgraph, then as template and is detected figure using this multiple template subgraph
As being matched, this allows for template has Dimensional variability in the matching process, so that finally participating in matched template
As far as possible with the adjoining dimensions of target in tested image, a matching process based on a template image is transformed by server
Multiple matching process based on multiple template subgraph determines final matching results by multiple matched matching result, to mention
The precision of height identification.
Certainly, any product for implementing the application does not necessarily require achieving all the advantages described above at the same time.
Detailed description of the invention
In order to more clearly explain the technical solutions in the embodiments of the present application, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for
For those of ordinary skill in the art, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 is the Sample Scenario figure of the application in practical applications;
Fig. 2 is a kind of structure chart of image identification system provided by the embodiments of the present application;
Fig. 3 is a kind of flow chart of image-recognizing method provided by the embodiments of the present application;
Fig. 4 is the interaction schematic diagram provided by the embodiments of the present application based on user configuration interface;
Fig. 5 is the flow chart of another image-recognizing method provided by the embodiments of the present application;
Fig. 5 A is with a kind of target display renderings on Fig. 4 scene basis;
Fig. 6 is a kind of structure chart of pattern recognition device provided by the embodiments of the present application;
Fig. 7 is a kind of hardware structure diagram for calculating equipment provided by the embodiments of the present application;
Fig. 8 is the structure chart of another pattern recognition device provided by the embodiments of the present application;
Fig. 9 is another hardware structure diagram for calculating equipment provided by the embodiments of the present application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
Technical solution provided by the present application in order to facilitate understanding below first carries out the research background of technical scheme
Simple declaration.
Inventor has found during studying traditional images identification technology, in electric business field, medical domain, calculating
It requires to realize related service using image recognition technology in the multiple fields such as machine field;It is said by taking e-commerce field as an example
It is bright, traditional images identification technology is often utilized in electric business platform, and legitimate verification is carried out to the commodity associated picture that businessman uploads,
Whether include invalid information in authentication image, and positions the invalid information.It lives in addition, electric business platform can usually issue some promotion
Dynamic, for the advertising campaign page, perhaps the relevant image of commodity is verified whether to verify loose-leaf or commodity image
Comprising throbber, to guarantee that advertising campaign is normally carried out.Again by taking medical domain as an example, doctor would generally utilize traditional images
Identification technology carries out image recognition to the medical image picture of patient, to orient lesions position.And traditional images identify
Technology specifically calls directly template matching algorithm when realizing, carries out a template to tested image based on template image
Match, directly determines matching result, this method target size and when inconsistent template image size in tested image, identification
Precision is not high, tends not to meet business demand, especially in electric business field, medical domain, be detected image in target often
It is unknown, and its target size is often inconsistent with template size.
Based on this, inventor proposes the technical solution of the application after study, during template matching, by mould
Plate image zooms in and out processing and obtains at least two template subgraphs, recycles at least two template subgraphs to participate in matching, most
Best matching result is determined using multiple matching result eventually, to improve identification accuracy.
The application in order to facilitate understanding by those skilled in the art, below with reference to practical application scene to the application in practice
Applicable cases lay down a definition explanation.
Referring to Fig. 1, it illustrates the Sample Scenario figure of the application in practical applications, and under the scene, the application is provided
Image-recognizing method be applied to server 101 in, server 101 as hardware, refers to computing capability, passes through net
Network can supply the calculating equipment that multiple client user uses.Wherein, server 101 is that client 102 provides image recognition
The data of processing business are supported;Wherein, client 102 refers to corresponding with server, provides image recognition processing for user
The program of service;The client 102 can be independent application program, the functional module being also possible in some application programs.
In practical applications, for hardware point of view, client 102 can be deployed in server 101, can also be deployed in independently of
In the equipment of server 101, for example, client 102 deployment in the user terminal, client 102 specifically can with browser, answer
It is disposed in the user terminal with forms such as programs (APP), and user terminal can be the meter such as smart phone, notebook, computer
Calculate equipment.
In practical applications, user is by the specified tested image of client 102 and template image, server 101 according to
The specified tested image in family and template image carry out template matching, during template matching, and indirect utilize template image
It is matched, but size scaling is carried out to template image, a template image scaling is generated at least two template subgraphs,
Then, this at least two templates subgraph is recycled to be matched respectively with tested image, then it is last to utilize at least two matchings
As a result final matching results are determined;Server 101 sends the final matching results to client, and client 102 is final according to this
Matching result displaying target on tested image, to facilitate user to understand identification situation.It compares, traditional images identification technology is direct
It is matched using a template image, the image-recognizing method of the application will be changed based on the matching process of a template image
At the multiple matching process based on a template image size scaling treated at least two template subgraphs, to pass through
Multiple matching operation improves the precision of final matching results, so that the accuracy of target identification is higher.
Based on above-mentioned application scenarios, the embodiment of the present application provides a kind of image identification system, is to this below with reference to Fig. 2
System is introduced.
Referring to fig. 2, it illustrates a kind of structure chart of image identification system provided by the embodiments of the present application, the system packets
It includes:
Server 201, for determining tested image and template image, to the template image zoom in and out processing obtain to
Few two template subgraphs carry out template matching to the tested image using at least two template subgraphs and obtain and each mould
The corresponding matching result of plank image determines final matching results according at least two matching results;Preferably, server 201
Specific implementation can be found in following FIG. 3 shown in embodiment of the method realization.
Client, for marking the image district being identified in the tested image according to the final matching results
Domain.Preferably, the specific implementation of client 202 can be found in the realization of embodiment of the method shown in following FIG. 4.
Using the system provided by the embodiments of the present application, by the network connection mode of server and client, by servicing
Device zooms in and out processing to template image and generates at least two template subgraphs, the matching process of a template is transformed into multiple
The matching process of template subgraph determines final matching results using multiple matching results, to improve the accurate of target identification
Property, then server sends final matching results to client, displays for a user the final matching results by client, facilitates use
The higher final matching results of accuracy are watched by client in family in time.
Referring to Fig. 3, it illustrates a kind of flow chart of image-recognizing method provided by the embodiments of the present application, this method can be with
Applied in server, comprising the following steps:
Step 301 determines tested image and template image;
In the present embodiment, server first determines tested image and template image, wherein tested image refers to be detected
Source images, template image refer to given target image.Wherein, source images and target image can be bianry image, grayscale image
Picture or color image, certain the present embodiment are not especially limited this.
In the present embodiment, tested image and template image how to be determined about server, additionally provided several optional
Implementation is separately below described the optional implementation of these types, wherein the first optional implementation includes:
Server in response to client access request, to client feedback user configuration interface;Then server responds
In configuration operation of the user on the user configuration interface, tested image and template image are determined.
In specific implementation, the first control and the second control can be carried on the user configuration interface, wherein the first control
Part is for determining tested image, and the second control is for determining template image;User is referred in client by operating the two controls
Surely tested image and template image, client generates configuration information according to operation of the user on the user configuration interface, to clothes
Business device sends the configuration information, and server determines tested image and template image according to the configuration information.
More specifically, the first control can enable shooting function with the shooting function of associated terminal, user's touch-control first control
Can, tested image is generated by shooting image;First control can also with associated terminal local image data, then user's touch-control this
One control selects image as tested image from terminal storage space;First control can be also used for access and connect with terminal
Other equipment memory space, then user accesses the image that stores in other equipment by the first control, to specify
Image is as image to be checked.
And the second control can be associated with the template database pre-established in server, the second control passes through template number
Template image is provided for user according to library, user this time matches required template image by the way that the second control is specified, certainly, the second control
The concrete function of part and realization can also be by the way of above-mentioned first controls, and details are not described herein again.
In practical applications, user can once determine a tested image, can also once determine multiple tested images,
And its applicable template image is determined for identified tested image;A template image can be determined for tested image
It can determine multiple tested images.
The first above-mentioned optional implementation is illustrated below by example.
Referring to fig. 4, it illustrates the interaction schematic diagram based on user configuration interface, user accesses user by client and matches
Set interface, the first control on the touch-control user configuration interface, then display reminding information on user configuration interface, to prompt user
One of " shooting " or " being selected from photograph album " mode of selection, it is assumed that user's touch-control " selects " control, Jin Ercong from photograph album
Select a photo as tested image in photograph album, it is tested image that the upper left corner as shown in Figure 4, which ticks selected photo,.With
After tested image has been chosen at family, " determination " control is triggered, selected photo is uploaded to server as tested image.So
Afterwards, the template database data in the second control on user's touch-control user configuration interface, second control and server are closed
Connection, second control are associated with search box, and to facilitate user's fast search template, user can pass through template name, template master
The keywords such as topic scan for, and server searches for relevant template image in response to searching request from template database, and anti-
Correlate template image is presented, then user determines the template image needed for this time matching.
As it can be seen that server provides user configuration interface for user, by this using the first optional implementation
User oriented interactive mode specifies tested image and template image, which can be well according to actual needs by user
It is adapted in some user oriented business scenarios.
Second of optional implementation include: server excavated from service database according to default mining rule it is tested
Image;The default mining rule is for describing using the image with specified attribute as tested image;Server is according to tested
Specified attribute possessed by image excavates template image from template database.Wherein, specified attribute includes: service attribute, page
Face attribute, item property etc., wherein item property includes one or more kinds of combinations of attributes such as commodity price, merchandise classification
Etc..
Below only by taking the application scenarios of electric business platform as an example, preset mining rule is illustrated.
For example, some advertising campaigns can be usually held in electric business platform, such as double 11 activities, Valentine's Day activity etc.
Deng for these advertising campaigns, electric business platform can issue some specific activities pages, such as double 11 activities, in homepage
Show " double 11 gratitude feedbacks " small picture;It is small that " full three subtract one " is shown in the service icon of service dress class business in homepage
Icon;" fufty percent discount " small icon is shown in the commodity picture that price is more than 2000 yuan of electric type commodity;Etc., for
This advertising campaign, electric business platform generally require to position by image-recognizing method in these pages or these commodity images
The small icon for being included, the page to guarantee advertising campaign are normal.
It is then directed to this business scenario, following mining rule can be preset:
Mining rule 1, for describing using the image of homepage as tested image;
Mining rule 2, for describing using the service icon of clothing business in homepage as tested image;
Mining rule 3, the commodity image for the electric type commodity using commodity price higher than 2000 yuan is as tested image.
It is understood that mining rule believes the marks such as specified services mark, page iden-tity, commodity sign for describing
Identified image is ceased as tested image.
After server excavates tested image, according to attribute possessed by tested image, excavated from template database
Corresponding template image;For example, server excavates homepage image according to mining rule 1, then server is dug for homepage image
Excavating corresponding template image is " double 11 gratitude feedbacks " small picture;Server excavates the industry of clothing according to mining rule 2
Business icon, excavating corresponding template image for the service icon is " full three subtract one " small icon;Server is advised according to excavation
Then 3 businessman's picture is excavated, excavating corresponding template image for businessman's picture is " fufty percent discount " small icon.
Server is after having determined tested image and template image, it is necessary to be known in tested image using template image
Other target, in the present embodiment, server not conventionally directly adopt template image and carry out template to tested image
Matching, but need to be implemented step 302 and processing is zoomed in and out to template image.
Step 302 zooms in and out processing to the template image and obtains at least two template subgraphs;
Step 303, using at least two template subgraphs respectively to the tested image carry out template matching obtain with often
The corresponding matching result of a template subgraph;
In the present embodiment, as soon as server can after scaling processing of every execution obtains a template subgraph,
It executes a template matching and obtains a matching result;In this way, by scaling processing at least twice and it is corresponding at least twice
Matching treatment just obtains at least two matching results.
In the present embodiment, server can also first seek unity of action scaling processing at least twice, and each scaling processing obtains
One template subgraph, i.e. server first generate template subgraph corresponding with template image;Then at least two moulds are recycled
Plank image makees template matching to tested image respectively.
The present embodiment is directed to the specific implementation of step 302, several optional implementations is additionally provided, below to this several
Kind optional implementation is introduced respectively, wherein the first optional implementation includes:
Server determines that the first zooming parameter collection, the first zooming parameter collection include: at least two scalings;Service
Device zooms in and out processing to the template image according to the first zooming parameter collection and obtains at least two template subgraphs.
In specific implementation, server can provide a user the first zooming parameter collection configuration clothes by user configuration interface
Business, user configures the size of the first zooming parameter collection on the user configuration interface, in this way, user can be according to actual template
The requirements for high precision matched configures the first zooming parameter collection.Certainly, which can also be pre-configured in server
In, the first zooming parameter collection is stored in the server, specifically, different first can be arranged for different types of service
Zooming parameter collection, server are read and type of service matched the in zooming in and out processing Shi Xiancong database to template image
One zooming parameter collection.
For example, the size that target in image is detected in electric business platform is often below the size of template image, therefore,
In order to improve the precision of target identification, template image can be reduced so that the size of template image as far as possible with quilt
The adjoining dimensions of target in image are examined, this is based on, scaling can be set smaller than to 1 numerical value, for example, preset first
Zooming parameter collection includes: 80%, 50% and 30%, then server according to the preset first zooming parameter collection to template image into
Scaling processing obtains three template subgraphs to row three times.
Above-mentioned first zooming parameter collection also may include at least two groups scaling, wherein every group of scaling includes length
It spends scaling and width scaling, the length scale ratio and the width scaling is of different sizes.Based on this, take
Business device concentrates contracting of each group scaling to template image progress length and width inequality proportion using such first zooming parameter
It puts.
The present embodiment is not construed as limiting the specific value of scaling parameter, can be the numerical value greater than 1, or
Numerical value equal to 1 can also be the numerical value less than 1.
Second of optional implementation, comprising:
Server determines the second zooming parameter collection;The second zooming parameter collection include: zoom direction a, scaling step-length b and
Unidirectional scaling number c;Server zooms in and out processing to the template image according to the second zooming parameter collection and obtains at least
Two template subgraphs.Wherein, zoom direction a is positive, reverse or two-way scaling for identifying, and so-called forward direction refers to pair
Template image amplifies processing, inversely refers to and carries out diminution processing to template image, and it is two-way refer to both amplify diminution,
In, when a mark is positive, value 1;Value is -1 when a mark is reverse, if a is identified as just reverse, value is ± 1.Unidirectional contracting
Number c is put, for identifying the highest number of one direction offset.And step-length b is scaled, for identifying single scaling pixel transform amount.
In practical applications, scaling step-length b can be the length pixel transform amount of template image, be also possible to the width of template image
Pixel transform amount.Server can calculate scaling corresponding to each scaling processing according to scaling step-length b, be then based on
The scaling of calculating carries out length and width to template image, and size scaling handles to obtain corresponding template subgraph in proportion.In addition,
Above-mentioned scaling step-length may include the length pixel transform amount of template image and the width pixel transform amount of template image, wherein
The width of ratio and the width pixel transform amount and template image between the length pixel transform amount and the length of template image
Ratio is of different sizes between degree.Based on this, server can be realized according to the second zooming parameter and carry out length and width not to template image
Scaling in proportion.
In specific implementation, server can provide a user the second zooming parameter collection configuration clothes by user configuration interface
Business, user configures the size of the second zooming parameter collection on the user configuration interface, in this way, user can be according to actual template
The requirements for high precision matched configures the second zooming parameter collection.Certainly, which can also be pre-configured in server
In, the second zooming parameter collection is stored in the server, specifically, different second can be arranged for different types of service
Zooming parameter collection, server are read and type of service matched the in zooming in and out processing Shi Xiancong database to template image
Two zooming parameter collection.
The specific implementation of this second optional implementation is illustrated below, specific:
The size [long Tl, wide Tw] for three parameters and template image T that server is concentrated according to the second zooming parameter calculates
The zoom ranges of template image T include:
Length scale range is Tl to Tl+a*b*c;
Width zoom ranges are Tw to (Tw+a*b (Tw/Tl) * c);
Wherein, if value is 1 when a mark is positive;Value is -1 when a mark is reverse;If value is when a is identified as just reverse
±1.Known to then:
When a identifies forward direction, then the length scale range of template image T is Tl as Tl+b*c;Length scale range is
Tw is to (Tw+b (Tw/Tl) * c);
When a mark is reverse, then the length scale range of template image T is Tl-b*c to Tl;Width zoom ranges are
(Tw-b (Tw/Tl) * c) is to Tw;
When a mark is just reverse, then the length scale range of template image T are as follows: Tw-b*c to Tw+b*c;And width scales
Range is (Tw-b (Tw/Tl) * c) to (Tw+b (Tw/Tl) * c);
It is illustrated by taking positive reverse process as an example, it is N that server, which records current zoom number, then N value is from 0 to 2c+1;
If N≤2c+1, server calculates the corresponding scaling of each scaling processing according to the corresponding length and width zoom ranges of template image
Then ratio zooms in and out processing to template image according to the scaling, using the template image after each scaling processing as
Template subgraph.Specifically, the corresponding scaling of n-th scaling processing is equal to the corresponding length Tl-b* of n-th scaling processing
The ratio of the length Tl of c+N*b and template image;Alternatively, the corresponding scaling of n-th scaling processing is equal at n-th scaling
Manage the ratio between corresponding width Tw-b (Tw/Tl) * c+N*b (Tw/Tl) and the width Tw of template image.Server is to template
Image carries out total 2c+1 size change over and handles to obtain total 2c+1 template subgraph, what needs to be explained here is that, each template
The characteristics of image that subgraph is included is identical as template image.In turn, the 2c+1 mould obtained after server by utilizing scaling processing
Plank image carries out template matching to tested image respectively, and template matching obtains a matching result each time.
For example, the size of tested image S is (200 × 100), the size of template image T is (10 × 10), the second contracting
Putting parameter set is specially a=± 1, b=2, c=3;Then server is according to the second zooming parameter collection and template image T, to template
Image carries out altogether 11 scaling processings, the 7 various sizes of template subgraphs obtained after scaling processing, respectively template
The size of image T1 is (4 × 4), the size of template subgraph T2 is (6 × 6), the size of template subgraph T3 is (8 × 8), mould
The size of plank image T4 is (10 × 10), the size of template subgraph T5 is (12 × 12), the size of template subgraph T6 is
The size of (14 × 14), template subgraph T7 are (16 × 16);Server by utilizing this template subgraph respectively with tested image S
Template matching is carried out, 7 corresponding matching results are obtained, specifically: T1 and S is subjected to template matching and obtains matching result M1,
T2 and S is subjected to template matching and obtains matching result M2 ..., T7 and S is subjected to template matching and obtains matching result M7.
The template matching process between each template subgraph and tested image is introduced below.
Server carries out the matching treatment between template subgraph and tested image using template matching algorithm, specifically,
Refer to the position that template subgraph is found in tested image, this is detected in image and the most matched image-region of template subgraph
It is exactly target area, i.e. matching result.During template matching, as long as all subregions and template in tested image
Image is compared, and determines that most matched subregion is target area.Specifically, server can be according to template subgraph
Size establish window, by drawing window mode sliding window on tested image, compare window in template subgraph and tested image
Similarity between openning figure finds the maximum window subgraph of similarity, as finds target.Specifically, server can be adopted
It is calculated with the template matching algorithm based on gray scale: MAD algorithm (MAD), absolute error and algorithm (SAD), error sum of squares
Method (SSD), mean error quadratic sum algorithm (MSD), normalization product correlation al gorithm (NCC), sequential similarity detection algorithm
(SSDA) scheduling algorithm.
Certainly, server can also use other kinds of template matching algorithm, and the present embodiment is not especially limited this,
However, server, which carries out a template matching processing with template subgraph to tested image, to be obtained using any algorithm
A corresponding matching result, the matching result include at least the size of target position, matching degree and target area, the target
Position can be the top left co-ordinate of target area or the target position is also possible to other apex coordinates of target area;
The matching degree refers to matching degree or similarity degree between the target area and template subgraph;The size of the target area is
Refer to the size for the template subgraph being matched.
If server uses the template matching algorithm based on gray scale, server, can be with before carrying out template matching
The processing of image gray-scale edges is made to tested image and each template subgraph.Wherein, image grayscale edge processing refers to figure
As making ashing processing and edge detection process, wherein be in order to which image is made gray scale conversion to the purpose that image makees ashing processing
Processing is converted into grayscale image;The purpose for carrying out edge detection process to image is apparent in order to identify brightness change in image
Point, i.e. edge.
Step 304 determines final matching results according at least two matching results.
The present embodiment provides several optional implementations for the specific implementation of step 304, wherein the first is optional
Implementation, comprising:
Server selects the highest matching result of matching degree as final from least two matching result
With result.
The implementation is illustrated based on above-mentioned example, server will obtain after template image scaling processing
To 7 template subgraphs, template matching is carried out to tested image respectively based on this 7 template subgraphs and obtains corresponding 7
With as a result, therefrom selecting that highest matching result of matching degree as final matching results, it is assumed that the 2nd template subgraph with
The matching degree highest of the matching result of tested image, then server selects the matching knot of the 2nd template subgraph and tested image
Fruit is as final matching results.
Using the first implementation, server selects the highest matching result of matching degree from the angle of matching degree
As final matching results, to improve the accuracy of target identification.
Second of optional implementation, comprising:
When the matching degree maximum value at least two matching result is less than the first preset threshold, from described at least two
Selection is greater than the matching result of the second preset threshold in a matching result;First preset threshold is greater than the described second default threshold
Value;
Determine that the coordinates of targets in selected matching result is formed by centre of figure position coordinates as final matching knot
Target area size in coordinates of targets in fruit, and the determining maximum matching result of matching degree is as in final matching results
Target area size.
For example, being sat if selected matching result includes two matching results according to the target in the two matching results
Mark is formed by line segment, determines coordinates of targets of the midpoint coordinate of the line segment as final matching results;If selected
Include at least three matching results with result, then region is formed by according to the coordinates of targets at least three matching results, really
Coordinates of targets of the center position coordinates in the fixed region as final matching results.
In practical applications, since to may result in multiple matched matching degree not high for the factor of picture quality, if single
The pure angle from matching degree height determines final matching results, and accuracy is not so good, is based on this, and the embodiment of the present application provides
Above-mentioned second optional implementation, coordinates of targets and matching degree in comprehensive multiple matching result determine final matching knot
Fruit, to improve its accuracy.
The third optional implementation, comprising:
Server selects matching degree to be greater than the matching result of third predetermined threshold value from least two matching result;
Server determines a matching result as final matching results at random from selected matching result.In addition, in the above method
On the basis of, it is contemplated that in real image identification business, need to judge by image-recognizing method in image whether include
Some target, then whether server can also judge the final matching results less than the 4th preset threshold, if being less than, generate and accuse
Alert information.
For example, if the matching degree for the final matching results that server determines is less than the 4th preset threshold, then it is assumed that this is final
Matching result shows in tested image and does not include the content of template image, and server generates warning information, which uses
The template image is not included in describing the tested image.
Using this method provided in this embodiment, processing is zoomed in and out to template image by server and generates at least two
The matching process of one template is transformed into the matching process of multiple template subgraph by template subgraph, is tied using multiple matchings
Fruit determines final matching results, to improve the precision of target identification.
Referring to Fig. 5, it illustrates the embodiment of the present application to provide another image-recognizing method, and this method can be with client
The form at end is applied to calculate in equipment, method includes the following steps:
Step 501, display user configuration interface;
In the present embodiment, client can send access request to server, server in response to the access request, to
Client feedback user configuration interface, client show the user configuration interface, to facilitate user to identify industry according to real image
Business demand configures the information such as tested image, template image.Client is generated according to operation of the user on the user configuration interface
Configuration information, the configuration information are used to indicate server and determine tested image and template image.
A kind of optional implementation, user can upload tested image and template image by the user configuration interface,
Then client generates the configuration information comprising the tested image and template image;
Another optional implementation, user is by the specified tested image in the user configuration interface and template image, then
Client generates the configuration information of the mark of the mark comprising the tested image and template image.
In the present embodiment, which can also include: zooming parameter, then the configuration information is used to indicate server
Image recognition processing is carried out according to the zooming parameter, the form of specific zooming parameter may refer to the implementation of method shown in above-mentioned Fig. 3
Description in example.
Step 502 sends configuration information to server;The configuration information is used to indicate server and carries out to template image
Scaling processing obtains at least two template subgraphs, and carries out template to tested image according at least two template subgraphs
Match;Wherein, the configuration information may include: tested image, template image and zooming parameter collection;The zooming parameter collection can
To see above the first zooming parameter collection or the second zooming parameter collection of description.
Step 503, the final matching results for receiving server feedback;
In the present embodiment, after client generates configuration information, the configuration information is sent to server, to indicate to service
Device carries out image recognition processing according to the configuration information;Server carries out image according to the step in Fig. 3 the method embodiment
Identification obtains final matching results, to the client feedback final matching results.
Step 504 marks the image-region being identified in the tested image according to the final matching results.?
In the present embodiment, client receives the final matching results of server feedback, which includes at least: target is sat
It is marked with and the size of target area, also may include matching degree.It is marked and is known in tested image according to the final matching results
Not Chu image-region.For example, client can select the image-region being identified by frame in such a way that line color highlights.
In the present embodiment, if client is received for multiple tested corresponding final matching results of image,
Can the image-region being identified be marked in tested image in the display interface in the form of a list.
Using method provided in this embodiment, user configures image recognition correlation by client according to practical business demand
Information, client indicates that server carries out image recognition processing by data communication between server, accurate to improve identification
Degree, and then the image-region being identified is marked according to the final matching results of server feedback.
In order to make it easy to understand, for the scene shown in Fig. 4, client receive server feedback final matching results it
Afterwards, identified image-region is marked, display effect is as shown in Figure 5A.
Corresponding with method shown in above-mentioned Fig. 3, the embodiment of the present application also provides a kind of servers, below with reference to Fig. 6
The server is introduced.
Referring to Fig. 6, it illustrates a kind of structure chart of server provided by the embodiments of the present application, which includes:
First determining module 601, for determining tested image and template image;
Scaling processing module 602 obtains at least two template subgraphs for zooming in and out processing to the template image;
Matching module 603, for carrying out template matching to the tested image respectively using at least two template subgraphs
Obtain matching result corresponding with each template subgraph;
Second determining module 604, for determining final matching results according at least two matching results.
It should be noted that the specific implementation of each functional module of above-mentioned server may refer to side shown in figure 3 above
The realization of method embodiment, details are not described herein again.
Referring to Fig. 7, it illustrates a kind of hardware structure diagram for calculating equipment provided by the embodiments of the present application, the server packets
It includes:
Processor 701, memory 702, network interface 703 and bus system 704.
The bus system 704, for each hardware component for calculating equipment to be coupled;
The network interface 703, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory 702, for storing program instruction and data;
The processor 701, for reading the instruction stored in the memory, the following operation of execution:
Determine tested image and template image;
Processing is zoomed in and out to the template image and obtains at least two template subgraphs;
Template matching is carried out to the tested image respectively using at least two template subgraphs to obtain and each template
The corresponding matching result of image;
Final matching results are determined according at least two matching results.
It should be noted that the specific implementation of each functional module of above-mentioned server may refer to side shown in figure 3 above
The realization of method embodiment, details are not described herein again.
Corresponding with method shown in above-mentioned Fig. 5, referring to Fig. 8, it illustrates the embodiment of the present application to provide a kind of image
The structure chart of identification device, the pattern recognition device include:
First display module 801, for showing user configuration interface;
Sending module 802, for sending configuration information to server;The configuration information is used to indicate server and determines quilt
Examine image and template image;
Receiving module 803, for receiving the final matching results of server feedback;
Second display module 804 is identified for being marked in the tested image according to the final matching results
Image-region.Corresponding with method shown in above-mentioned Fig. 9, referring to Fig. 9, it illustrates the embodiment of the present application to provide a kind of meter
The hardware structure diagram of equipment is calculated, which includes:
Processor 901, memory 902, network interface 903 and bus system 904.
The bus system 904, for each hardware component for calculating equipment to be coupled;
The network interface 903, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory 902, for storing program instruction;
The processor 901, for reading the instruction stored in the memory, the following operation of execution:
Show user configuration interface;
Configuration information is sent to server;The configuration information is used to indicate server and zooms in and out processing to template image
At least two template subgraphs are obtained, and template matching is carried out to tested image according at least two template subgraphs;Receive clothes
The final matching results for device feedback of being engaged in;
The image-region being identified is marked in the tested image according to the final matching results.The processor
The specific implementation of the 901 each steps executed may refer to each step in embodiment of the method shown in figure 5 above, herein no longer
It repeats.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
For device class embodiment, since it is basically similar to the method embodiment, so being described relatively simple, related place ginseng
See the part explanation of embodiment of the method.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or apparatus that includes the element.
A kind of image-recognizing method provided herein, device, system and calculating equipment have been carried out in detail above
It introduces, specific examples are used herein to illustrate the principle and implementation manner of the present application, the explanation of above embodiments
It is merely used to help understand the present processes and its core concept;At the same time, for those skilled in the art, according to this
The thought of application, there will be changes in the specific implementation manner and application range, in conclusion the content of the present specification is not answered
It is interpreted as the limitation to the application.
Claims (17)
1. a kind of image identification system characterized by comprising
Server zooms in and out processing to the template image and obtains at least two for determining tested image and template image
Template subgraph carries out template matching to the tested image using at least two template subgraphs and obtains and each template subgraph
As corresponding matching result, final matching results are determined according at least two matching results;
Client, for marking the image-region being identified in the tested image according to the final matching results.
2. a kind of image-recognizing method characterized by comprising
Determine tested image and template image;
Processing is zoomed in and out to the template image and obtains at least two template subgraphs;
Template matching is carried out to the tested image respectively using at least two template subgraphs to obtain and each template subgraph
Corresponding matching result;
Final matching results are determined according at least two matching results.
3. method according to claim 2, which is characterized in that the determination is detected image and template image, comprising:
Tested image is excavated from service database according to default mining rule;The default mining rule is for describing to have
The image of specified attribute is as tested image;
Template image is excavated from template database according to specified attribute possessed by tested image.
4. method according to claim 3, which is characterized in that the specified attribute includes: service attribute, page properties, quotient
Product attribute.
5. method according to claim 2, which is characterized in that the determination is detected image and template image, comprising:
In response to the access request of client, to client feedback user configuration interface;
In response to configuration operation of the user on the user configuration interface, tested image and template image are determined.
6. method according to claim 2, which is characterized in that described to zoom in and out processing to the template image and obtain at least
Two template subgraphs, comprising:
Determine the first zooming parameter collection;The first zooming parameter collection includes: at least two scalings;
Processing is zoomed in and out to the template image according to the first zooming parameter collection and obtains at least two template subgraphs.
7. method according to claim 2, which is characterized in that described to zoom in and out processing to the template image and obtain at least
Two template subgraphs, comprising:
Determine the second zooming parameter collection;The second zooming parameter collection includes: zoom direction, scaling step-length and unidirectional scaling
Number;
Processing is zoomed in and out to the template image according to the second zooming parameter collection and obtains at least two template subgraphs.
8. method according to claim 2, which is characterized in that described to determine final matching knot according at least two matching results
Fruit, comprising:
From at least two matching result, select the highest matching result of matching degree as final matching results.
9. method according to claim 2, which is characterized in that described to determine final matching knot according at least two matching results
Fruit, comprising:
When the matching degree maximum value at least two matching result is less than the first preset threshold, from described at least two
With the matching result for selecting matching degree to be greater than the second preset threshold in result;It is default that first preset threshold is greater than described second
Threshold value;
Determine that the coordinates of targets in selected matching result is formed by centre of figure position coordinates as in final matching results
Coordinates of targets, and determine the maximum matching result of matching degree in target area size as the mesh in final matching results
Mark area size.
10. method according to claim 2, which is characterized in that described to determine final matching according at least two matching results
As a result, comprising:
From at least two matching result, matching degree is selected to be greater than the matching result of third predetermined threshold value;
From selected matching result, determine a matching result as final matching results at random.
11. method according to claim 2, which is characterized in that utilize at least two template subgraphs respectively to institute described
Tested image is stated to carry out before template matching obtains matching result corresponding with each template subgraph, the method also includes:
The processing of image gray-scale edges is made respectively to the tested image and each template subgraph.
12. method according to claim 2, which is characterized in that the method also includes:
Whether the matching degree of the final matching results is judged less than the 4th preset threshold, if being less than, generates warning information.
13. a kind of pattern recognition device characterized by comprising
First determining module, for determining tested image and template image;
Scaling processing module obtains at least two template subgraphs for zooming in and out processing to the template image;
Matching module, for being obtained respectively to the tested image progress template matching using at least two template subgraphs and often
The corresponding matching result of a template subgraph;
Second determining module, for determining final matching results according at least two matching results.
14. a kind of calculating equipment characterized by comprising
Processor, memory, network interface and bus system.
The bus system, for each hardware component for calculating equipment to be coupled;
The network interface, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory, for storing program instruction;
The processor, for reading the instruction stored in the memory, the following operation of execution:
Determine tested image and template image;
Processing is zoomed in and out to the template image and obtains at least two template subgraphs;
Template matching is carried out to the tested image respectively using at least two template subgraphs to obtain and each template subgraph
Corresponding matching result;
Final matching results are determined according at least two matching results.
15. a kind of image-recognizing method characterized by comprising
Show user configuration interface;
Configuration information is sent to server, the configuration information, which is used to indicate server and zooms in and out processing to template image, to be obtained
At least two template subgraphs, and template matching is carried out to tested image according at least two template subgraphs;Receive server
The final matching results of feedback;
The image-region being identified is marked in the tested image according to the final matching results.
16. a kind of pattern recognition device characterized by comprising
First display module, for showing user configuration interface;
Sending module, for server send configuration information, the configuration information be used to indicate server to template image into
Row scaling processing obtains at least two template subgraphs, and carries out template to tested image according at least two template subgraphs
Match;
Receiving module, for receiving the final matching results of server feedback;
Second display module, for marking the image district being identified in the tested image according to the final matching results
Domain.
17. a kind of calculating equipment characterized by comprising
Processor, memory, network interface and bus system.
The bus system, for each hardware component for calculating equipment to be coupled;
The network interface, for realizing the communication connection between the calculating equipment and at least one other equipment;
The memory, for storing program instruction;
The processor, for reading the instruction stored in the memory, the following operation of execution:
Show user configuration interface;
Configuration information is sent to server;The configuration information, which is used to indicate server and zooms in and out processing to template image, to be obtained
At least two template subgraphs, and template matching is carried out to tested image according at least two template subgraphs;Receive server
The final matching results of feedback;
The image-region being identified is marked in the tested image according to the final matching results.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711240738.5A CN109858504A (en) | 2017-11-30 | 2017-11-30 | A kind of image-recognizing method, device, system and calculate equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711240738.5A CN109858504A (en) | 2017-11-30 | 2017-11-30 | A kind of image-recognizing method, device, system and calculate equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109858504A true CN109858504A (en) | 2019-06-07 |
Family
ID=66888465
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711240738.5A Pending CN109858504A (en) | 2017-11-30 | 2017-11-30 | A kind of image-recognizing method, device, system and calculate equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109858504A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111027118A (en) * | 2019-11-19 | 2020-04-17 | 广东博智林机器人有限公司 | Actual measurement actual quantity task point searching and task dispatching method and system |
CN111079730A (en) * | 2019-11-20 | 2020-04-28 | 北京云聚智慧科技有限公司 | Method for determining area of sample image in interface image and electronic equipment |
CN111258938A (en) * | 2019-11-07 | 2020-06-09 | 深圳市远行科技股份有限公司 | Method and equipment for processing application docking service provider to be adapted |
CN111309618A (en) * | 2020-02-24 | 2020-06-19 | 广州市百果园信息技术有限公司 | Page element positioning method, page testing method and related device |
CN111797933A (en) * | 2020-07-10 | 2020-10-20 | 百度在线网络技术(北京)有限公司 | Template matching method and device, electronic equipment and storage medium |
CN113128543A (en) * | 2020-01-15 | 2021-07-16 | 阿里巴巴集团控股有限公司 | Image matching method, application testing method, device and system |
CN113466233A (en) * | 2020-03-31 | 2021-10-01 | 北京配天技术有限公司 | Visual inspection method, visual inspection apparatus, and computer storage medium |
CN114489915A (en) * | 2022-02-17 | 2022-05-13 | 平安国际智慧城市科技股份有限公司 | Open screen picture display method, device, equipment and medium based on mobile equipment |
CN117252956A (en) * | 2023-09-04 | 2023-12-19 | 深圳市经纬线科技有限公司 | Image editing method, device, equipment and medium |
CN117714894A (en) * | 2023-08-03 | 2024-03-15 | 荣耀终端有限公司 | Target identification method, electronic equipment and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1301369A (en) * | 1998-05-18 | 2001-06-27 | 数据立方公司 | Image tecognition and correlation system |
WO2009113598A1 (en) * | 2008-03-13 | 2009-09-17 | Kabushiki Kaisha Toshiba | Template matching apparatus and method thereof |
CN101739692A (en) * | 2009-12-29 | 2010-06-16 | 天津市亚安科技电子有限公司 | Fast correlation tracking method for real-time video target |
CN101996398A (en) * | 2009-08-12 | 2011-03-30 | 睿励科学仪器(上海)有限公司 | Image matching method and equipment for wafer alignment |
CN105865423A (en) * | 2016-03-30 | 2016-08-17 | 沈阳泰科易科技有限公司 | A binocular range finding method, a binocular range finding device, a panoramic image mosaicking method and a system thereof |
WO2016144800A1 (en) * | 2015-03-06 | 2016-09-15 | Matthew Lee | Method and device for capturing images using image templates |
CN106228194A (en) * | 2016-08-05 | 2016-12-14 | 腾讯科技(深圳)有限公司 | Image lookup method and device |
CN106504238A (en) * | 2016-10-31 | 2017-03-15 | 成都交大光芒科技股份有限公司 | Railway contact line defect inspection method based on image procossing and convolutional neural networks |
-
2017
- 2017-11-30 CN CN201711240738.5A patent/CN109858504A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1301369A (en) * | 1998-05-18 | 2001-06-27 | 数据立方公司 | Image tecognition and correlation system |
WO2009113598A1 (en) * | 2008-03-13 | 2009-09-17 | Kabushiki Kaisha Toshiba | Template matching apparatus and method thereof |
CN101996398A (en) * | 2009-08-12 | 2011-03-30 | 睿励科学仪器(上海)有限公司 | Image matching method and equipment for wafer alignment |
CN101739692A (en) * | 2009-12-29 | 2010-06-16 | 天津市亚安科技电子有限公司 | Fast correlation tracking method for real-time video target |
WO2016144800A1 (en) * | 2015-03-06 | 2016-09-15 | Matthew Lee | Method and device for capturing images using image templates |
CN105865423A (en) * | 2016-03-30 | 2016-08-17 | 沈阳泰科易科技有限公司 | A binocular range finding method, a binocular range finding device, a panoramic image mosaicking method and a system thereof |
CN106228194A (en) * | 2016-08-05 | 2016-12-14 | 腾讯科技(深圳)有限公司 | Image lookup method and device |
CN106504238A (en) * | 2016-10-31 | 2017-03-15 | 成都交大光芒科技股份有限公司 | Railway contact line defect inspection method based on image procossing and convolutional neural networks |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111258938A (en) * | 2019-11-07 | 2020-06-09 | 深圳市远行科技股份有限公司 | Method and equipment for processing application docking service provider to be adapted |
CN111027118A (en) * | 2019-11-19 | 2020-04-17 | 广东博智林机器人有限公司 | Actual measurement actual quantity task point searching and task dispatching method and system |
CN111027118B (en) * | 2019-11-19 | 2024-01-19 | 广东博智林机器人有限公司 | Actual measurement real-quantity task point searching and task dispatching method and system |
CN111079730B (en) * | 2019-11-20 | 2023-12-22 | 北京云聚智慧科技有限公司 | Method for determining area of sample graph in interface graph and electronic equipment |
CN111079730A (en) * | 2019-11-20 | 2020-04-28 | 北京云聚智慧科技有限公司 | Method for determining area of sample image in interface image and electronic equipment |
CN113128543B (en) * | 2020-01-15 | 2024-02-13 | 阿里巴巴集团控股有限公司 | Image matching method, application testing method, device and system |
CN113128543A (en) * | 2020-01-15 | 2021-07-16 | 阿里巴巴集团控股有限公司 | Image matching method, application testing method, device and system |
CN111309618A (en) * | 2020-02-24 | 2020-06-19 | 广州市百果园信息技术有限公司 | Page element positioning method, page testing method and related device |
CN113466233A (en) * | 2020-03-31 | 2021-10-01 | 北京配天技术有限公司 | Visual inspection method, visual inspection apparatus, and computer storage medium |
CN111797933B (en) * | 2020-07-10 | 2023-07-28 | 百度在线网络技术(北京)有限公司 | Template matching method, device, electronic equipment and storage medium |
CN111797933A (en) * | 2020-07-10 | 2020-10-20 | 百度在线网络技术(北京)有限公司 | Template matching method and device, electronic equipment and storage medium |
CN114489915A (en) * | 2022-02-17 | 2022-05-13 | 平安国际智慧城市科技股份有限公司 | Open screen picture display method, device, equipment and medium based on mobile equipment |
CN117714894A (en) * | 2023-08-03 | 2024-03-15 | 荣耀终端有限公司 | Target identification method, electronic equipment and storage medium |
CN117252956A (en) * | 2023-09-04 | 2023-12-19 | 深圳市经纬线科技有限公司 | Image editing method, device, equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109858504A (en) | A kind of image-recognizing method, device, system and calculate equipment | |
Yang et al. | Copy-move forgery detection based on hybrid features | |
Bi et al. | Fast copy-move forgery detection using local bidirectional coherency error refinement | |
Bae et al. | High-precision vision-based mobile augmented reality system for context-aware architectural, engineering, construction and facility management (AEC/FM) applications | |
US20130148860A1 (en) | Motion aligned distance calculations for image comparisons | |
WO2021093329A1 (en) | Interactive behavior identification method and apparatus, computer device and storage medium | |
US9014467B2 (en) | Image processing method and image processing device | |
Lu et al. | A Delaunay-based temporal coding model for micro-expression recognition | |
Hosny et al. | Copy-move forgery detection of duplicated objects using accurate PCET moments and morphological operators | |
JP2012014544A (en) | Coordinate recommendation apparatus, coordinate recommendation method and program therefor | |
CN109886223B (en) | Face recognition method, bottom library input method and device and electronic equipment | |
CN109840524A (en) | Kind identification method, device, equipment and the storage medium of text | |
Vaghela et al. | A review of image mosaicing techniques | |
Buoncompagni et al. | Saliency-based keypoint selection for fast object detection and matching | |
US10600202B2 (en) | Information processing device and method, and program | |
CN111881804A (en) | Attitude estimation model training method, system, medium and terminal based on joint training | |
Zhou et al. | Deep learning-based instance segmentation for indoor fire load recognition | |
CN109409388A (en) | A kind of bimodulus deep learning based on graphic primitive describes sub- building method | |
Zheng et al. | Multi-scale density detection for yarn-dyed fabrics with deformed repeat patterns | |
CN108345687A (en) | A kind of 3D web page display method and apparatus | |
US10991085B2 (en) | Classifying panoramic images | |
Matusiak et al. | Unbiased evaluation of keypoint detectors with respect to rotation invariance | |
Juvela | Template matching method for the analysis of interstellar cloud structure | |
Yu et al. | Scale-invariant feature transform based on the frequency spectrum and the grid for remote sensing image registration | |
Fan et al. | COPRNet: correspondence confidence and overlap score guided network for indoor partial point cloud registration |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190607 |