CN109766802A - Flexible article recognition methods, device, computer equipment and storage medium - Google Patents
Flexible article recognition methods, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109766802A CN109766802A CN201811623177.1A CN201811623177A CN109766802A CN 109766802 A CN109766802 A CN 109766802A CN 201811623177 A CN201811623177 A CN 201811623177A CN 109766802 A CN109766802 A CN 109766802A
- Authority
- CN
- China
- Prior art keywords
- image
- cutting
- under test
- object under
- template
- 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 49
- 238000005520 cutting process Methods 0.000 claims abstract description 200
- 238000012360 testing method Methods 0.000 claims abstract description 132
- 238000004590 computer program Methods 0.000 claims description 28
- 239000000284 extract Substances 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 13
- 230000036544 posture Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 3
- 239000002131 composite material Substances 0.000 description 3
- 238000003709 image segmentation Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
This application involves a kind of flexible article recognition methods, device, computer equipment and storage mediums.The described method includes: obtaining object under test image and preset multiple template image；The object under test image and each template image are subjected to cutting respectively, obtain the corresponding multiple first cutting images of the object under test image and the corresponding multiple second cutting images of each template image；According to the second cutting image of each first cutting image and each corresponding position, identify the object under test of the object under test image, since different template images can be used to describe the different deformation state of object under test, and then suitable template image feature can be chosen, realize object identification, even with there are the flexible article of deformation, object identification also may be implemented.
Description
Technical field
This application involves computer vision fields, set more particularly to a kind of flexible article recognition methods, device, computer
Standby and storage medium.
Background technique
With the rapid development of computer technology, object recognition technique is ground as a basis in computer vision field
Study carefully, is widely used to the every aspect of industry, life and national defence scene.For example, the assembly in industrial robot application scenarios
System is just needed using object recognition technique, after first passing through industrial camera acquisition image, then by object recognition technique realizes object
Body identification, and then basis is provided to complete assembly manipulation.
The task of object recognition technique is to identify in image there is what object, and calculate this object in image expression
Scene in position and direction.Traditional object recognition technique is the recognition methods based on model, it is firstly the need of establishing object
Then body Model identifies the object most like with object model using matching algorithm from true image, realize that object is known
Not.
However, traditional recognition methods based on model is difficult to carry out object identification for there are the flexible article of deformation,
It can not even identify object.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of flexible article recognition methods, device, computer are set
Standby and storage medium.
A kind of flexible article recognition methods, which comprises
Obtain object under test image and preset multiple template image；The template image is for describing the more of object under test
A deformed state；
The object under test image and each template image are subjected to cutting respectively, obtain the object under test image pair
The corresponding multiple second cutting images of multiple first cutting images and each template image answered；
According to the second cutting image of each first cutting image and each corresponding position, the object under test image is identified
Object under test.
In one of the embodiments, the method also includes: obtain multiple original objects images；The multiple original object
Body image includes multiple posture informations of object；Each original objects image is subjected to cutting, obtains multiple third cutting figures
Picture；Extract the edge of each third cutting image；According to the edge, each third cutting image is combined, is obtained
To the template image.
It is described according to the edge in one of the embodiments, each third cutting image is combined, is obtained
Described image template, comprising: cutting is carried out to each edge, obtains multiple sub- edges；Deformation is carried out to each sub- edge
Operation, the sub- edge after obtaining deformation；According to the sub- edge after the deformation, each third cutting image is combined,
Obtain multiple strain images；Judge whether the strain image meets preset deformation range；If the strain image meets pre-
If deformation range, then the strain image is determined as the template image.
The second cutting figure according to each first cutting image and each corresponding position in one of the embodiments,
Picture identifies the object under test of the object under test image, comprising: according to preset extraction algorithm, extracts described first respectively and cuts
The side right value information of the second cutting image of partial image and corresponding position；According to each side right value information, described in identification
The object under test of object under test image.
It is described according to preset extraction algorithm in one of the embodiments, extract respectively the first cutting image with
The side right value information of the second cutting image of corresponding position, comprising: obtain the first cutting image and described the respectively
The key node of two cutting images；Subdivision processing is made to the key node, obtains subdivision result；It is mentioned according to the subdivision result
Take side right value information.
It is described according to each side right value information in one of the embodiments, identify the object under test image to
Survey object, comprising: according to the first cutting image and the second cutting image, acquisition the first cutting image with it is corresponding
The first similarity between each second cutting image of position；According to first similarity, the object under test is obtained
The second similarity between image and each template image；According to second similarity, the object under test image is identified
Object under test.
It is described according to first similarity in one of the embodiments, obtain the object under test image and each institute
State the second similarity between template image, comprising: according to first similarity and corresponding weight, obtain the determinand
The second similarity between body image and each template image.
A kind of flexible article identification device, described device include:
First obtains module, for obtaining object under test image and preset multiple template image；The template image is used
In multiple deformed states of description object under test；
Second obtains module, for the object under test image and each template image to be carried out cutting respectively, obtains
The corresponding multiple first cutting images of object under test image and the corresponding multiple second cutting images of each template image；
Identification module identifies institute for the second cutting image according to each first cutting image and each corresponding position
State the object under test of object under test image.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device realizes object identification method described in any of the above-described embodiment when executing the computer program.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
Object identification method described in any of the above-described embodiment is realized when row.
Above-mentioned flexible article recognition methods, device, computer equipment and storage medium, by obtain object under test image and
Then object under test image and each template image are carried out cutting by preset multiple template image respectively, obtain object under test figure
As corresponding multiple first cutting images and the corresponding multiple second cutting images of each template image, and then according to each first cutting
Second cutting image of image and each corresponding position identifies the object under test of object under test image, due to different template images
It can be used to describe the different deformation state of object under test, and then suitable template image feature can be chosen, realize object
Identification, even with there are the flexible article of deformation, also may be implemented object identification.
Detailed description of the invention
Fig. 1 is the internal structure chart of computer equipment in one embodiment；
Fig. 2 is the flow diagram of flexible article recognition methods in one embodiment；
Fig. 3 is template image cutting result schematic diagram in one embodiment；
Fig. 4 is the flow diagram that template image is established in one embodiment；
Fig. 5 is the flow diagram of the implementation of S404 in Fig. 4 embodiment；
Fig. 6 is the flow diagram of the implementation of S203 in Fig. 2 embodiment；
Fig. 7 is the flow diagram of the implementation of S601 in Fig. 6 embodiment；
Fig. 8 is the flow diagram of the implementation of S602 in Fig. 6 embodiment；
Fig. 9 is the first cutting image and the second cutting images match process schematic in one embodiment；
Figure 10 is the structural block diagram of flexible article identification device in one embodiment；
Figure 11 is the structural block diagram of flexible article identification device in another embodiment；
Figure 12 is the structural block diagram of flexible article identification device in another embodiment；
Figure 13 is the structural block diagram of flexible article identification device in another embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Flexible article recognition methods provided by the embodiments of the present application can be applied to computer equipment, which can
To be terminal, internal structure chart can be as shown in Figure 1.The computer equipment include by system bus connect processor,
Memory, network interface, display screen and input unit.Wherein, the processor of the computer equipment is calculated and is controlled for providing
Ability.The memory of the computer equipment includes non-volatile memory medium, built-in storage.Non-volatile memory medium storage
There are operating system and computer program.The built-in storage is operating system and computer program in non-volatile memory medium
Operation provides environment.The network interface of the computer equipment is used to communicate with external terminal by network connection.The computer
To realize a kind of flexible article recognition methods when program is executed by processor.The display screen of the computer equipment can be liquid crystal
Display screen or electric ink display screen, the input unit of the computer equipment can be the touch layer covered on display screen, can also
To be the key being arranged on computer equipment shell, trace ball or Trackpad, external keyboard, Trackpad or mouse can also be
Deng.
It will be understood by those skilled in the art that structure shown in Fig. 1, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
Technical solution of the present invention is described in detail with specifically embodiment below.These specific implementations below
Example can be combined with each other, and the same or similar concept or process may be repeated no more in certain embodiments.
In one embodiment, as shown in Fig. 2, providing a kind of flexible article recognition methods, the executing subject of this method
For computer equipment shown in FIG. 1, this application involves be flexible article identification specific implementation process, comprising the following steps:
S201 obtains object under test image and preset multiple template image；The template image is for describing determinand
Multiple deformed states of body.
Wherein, preset template image can be a variety of strain images of multiple objects, for example, it may be object A and object
A variety of strain image A1, A2, A3 of body A etc. are also possible to a variety of strain image B1, B2, B3 of object B and object B etc..
Specifically, object under test can be obtained by the image collecting device such as camera connecting with computer equipment
Image, and then object under test image is sent to computer equipment by camera, computer equipment obtains object under test image；It is default
Multiple template image can be and be stored directly in computer equipment or the template library being stored in other equipment, Jin Erji
It calculates machine equipment and obtains object under test image and preset multiple template image.
The object under test image and each template image are carried out cutting respectively, obtain the object under test by S202
The corresponding multiple first cutting images of image and the corresponding multiple second cutting images of each template image.
It specifically, can be according to the size dimension of object under test image, using image segmentation technique, by object under test image
9,16 or 25 etc. subgraphs are cut into, and then obtain the corresponding multiple first cutting images of the object under test image, for example,
Object under test image PA can be cut into the subgraph PA_1, PA_2 of the sizes such as 9, PA_3, PA_4, PA_5, PA_6, PA_7,
PA_8、PA_9。
Equally, according to the size dimension of each template image, using image segmentation technique, by each template image cutting 9,16 or
25 subgraphs, wherein object under test image is identical with the subgraph quantity of each template image cutting, i.e. object under test image cutting
At 9 subgraphs, then each template image is also cut into 9 images, and then obtains corresponding multiple second cuttings of each template image
Image, for example, as shown in figure 3, template image PA1 can be cut into the subgraph of the sizes such as 9 for template image PA1
PA1_1,PA1_2,PA1_3,PA1_4,PA1_5,PA1_6,PA1_7,PA1_8,PA1_9；For template image PA2, can incite somebody to action
Template image PA2 be cut into the subgraph PA2_1, PA2_2 of the sizes such as 9, PA2_3, PA2_4, PA2_5, PA2_6, PA2_7,
PA2_8、PA2_9。
S203 identifies the determinand according to the second cutting image of each first cutting image and each corresponding position
The object under test of body image.
Illustratively, can by the first cutting image PA_1, with each corresponding position the second cutting image PA1_1, PA2_1 into
Row comparison, obtains the matching result of PA_1 and PA1_1, PA2_1；By the subgraph PA_2 in the first cutting image, with each corresponding position
Second cutting image PA1_2, PA2_2 is set to compare, so obtain PA_2 and PA1_2, PA2_2 matching result, successively into
Row matching, acquisition and the immediate second cutting image of each first cutting image, so it is corresponding according to multiple second cutting images
Template image, identify object under test image object under test.
In above-described embodiment, by obtaining object under test image and preset multiple template image, then by object under test
Image and each template image carry out cutting respectively, obtain the corresponding multiple first cutting images of object under test image and each Prototype drawing
As corresponding multiple second cutting images, and then according to the second cutting image of each first cutting image and each corresponding position, know
The object under test of other object under test image, since different template images can be used to describe the different deformation shape of object under test
State, and then suitable template image feature can be chosen, object identification is realized, even with there are the flexible objects of deformation
Object identification also may be implemented in body.
During flexible article identification, needs to initially set up template image, template image is stored in computer and is set
It is standby.On the basis of the above embodiments, Fig. 4 provides a kind of flow diagram for establishing template image, as shown in figure 4, described
Method further include:
S401 obtains multiple original objects images；The multiple original objects image includes multiple posture informations of object.
Wherein, original objects image may include the original flexible article image of several width all angles postures, can be right
Original objects image carries out image enhancement operation, specifically includes the behaviour such as overturning, rotation, cutting, the scaling deformation to flexible article
Make, alternatively, Gaussian noise, shade noise can also be increased；Or carry out clashing the operation such as method at random, allow to extract and learn
The more features of original objects image.
Illustratively, original objects image can be different types of flexible article image, be also possible to same type of
Flexible article image, the present embodiment are not specifically limited.
Each original objects image is carried out cutting, obtains multiple third cutting images by S402.
It specifically, can be according to the size dimension of original objects image, using image segmentation technique, by original objects image
9,16 or 25 subgraphs are cut into, and then obtain multiple third cutting images.Here, the cutting quantity of original objects image is answered
The subgraph quantity with object under test image in above-described embodiment, each template image cutting is consistent.
Illustratively, for original objects image PB, original objects image PB can be cut into the subgraph of the sizes such as 9
PB_1、PB_2、PB_3、PB_4、PB_5、PB_6、PB_7、PB_8、PB_9。
S403 extracts the edge of each third cutting image.
It wherein, can will be every in third cutting image in extracting third cutting image before the edge of every subgraph object
Zhang Zitu is divided into the storage of three Color Channels, then each section is converted to grayscale image.It optionally, can be by utilizing Edge extraction
Algorithm extracts the edge of each third cutting image.Illustratively, the side of the subgraph PB_1 of original objects image PB can be extracted
Edge, extract original objects image PB subgraph PB_3 edge, can extract original objects image PC subgraph PC_1 edge
Deng.
S404 is combined each third cutting image, obtains the template image according to the edge.
Specifically, the edge of each third cutting image can be subjected to deformation, is carried out further according to the edge after deformation random
Combination, obtains template image.Illustratively, the edge of the subgraph PB_1 of original objects image PB can be revolved clockwise
Turning, the edge of the subgraph PB_2 of original objects image PB is rotated counterclockwise, other subgraphs PB_3 of original objects image B,
PB_4, PB_5, PB_6, PB_7, PB_8, PB_9 are remained unchanged, and the edge after operating after deformation combines original objects image
Subgraph PB_1, PB_2 and PB_3, PB_4, PB_5, PB_6, PB_7, PB_8, PB_9 after the deformation of B, available original objects
A strain image B1 of image B.
In above-described embodiment, by obtaining multiple original objects images, then each original objects image is cut
Point, it obtains multiple third cutting images, extracts the edge of each third cutting image, and then according to the edge, to each the
Three cutting images are combined, and obtain the template image, realize the foundation of model, provide good be applicable in for object identification
Basis.
Fig. 5 is provided according to the edge, is combined to each third cutting image, is obtained the template image
Specific implementation flow chart.As shown in figure 4, S404 " according to the edge, each third cutting image is combined,
Obtain the template image ", comprising:
S501 carries out cutting to each edge, obtains multiple sub- edges.
Specifically, each edge can be cut into the sub- edge of 3~5 pixel sizes, and then each edge according to actual needs
Multiple sub- edges can be got.
S502 carries out deformation operation to each sub- edge, the sub- edge after obtaining deformation.
Wherein it is possible to which every sub- edge is rotated clockwise the degrees such as 0.1 °, 0.2 °, 0.3 ° along straight line midpoint with such
It pushes away；Alternatively, every sub- edge is rotated degrees such as 0.1 °, 0.2 °, 0.3 ° and so on along straight line midpoint counterclockwise, deformation is obtained
Sub- edge afterwards.
S503 is combined each third cutting image, obtains multiple deformation according to the sub- edge after the deformation
Image.
Wherein it is possible to the sub- edge after each deformation of random combine, but the son after the horizontal and each deformation of vertical shift
During edge, need to guarantee to be overlapped the terminal of a line and the starting point of lower a line, it is ensured that strain image is in original graph
As on the basis of, subject image after deformation is obtained.
S504, judges whether the strain image meets preset deformation range.
Specifically, can be by judging whether strain image is more than the deformation coefficient of original objects image, and judge deformation
Whether the edge of image is continuous, judges whether strain image meets preset deformation range.
Illustratively, if the length, width and height of original objects image A, A type objects are respectively within x, y, z, then for original object
Whether the strain image A1 of body image A, can be by judging the length, width and height of A1 more than α x, β y, γ z, wherein α, β, γ are A class objects
The deformation coefficient of body, in addition it is also necessary to judge whether the edge of strain image A1 is continuous.If strain image A1 be less than α x, β y,
The continuous edge of γ z and strain image A1 then show that strain image A1 meets preset deformation range；If strain image A1 is more than
α x, β y, any or strain image A1 edge in γ z are discontinuous, then show that strain image A1 does not meet preset deformation model
It encloses.
The strain image is determined as the template if the strain image meets preset deformation range by S505
Image.
Illustratively, for the strain image A1 of original objects image A, if strain image A1 meets preset deformation model
It encloses, then strain image A1 is determined as template image；If strain image A1 does not meet preset deformation range, strain image A1
It cannot function as template image.
Optionally, for meeting the strain image of preset deformation range, not only strain image can be stored as template
The corresponding deformed state of strain image, strain image number can also be stored in model by image.
In above-described embodiment, by carrying out cutting to each edge, multiple sub- edges are obtained；Deformation behaviour is carried out to each sub- edge
Make, the sub- edge after obtaining deformation；According to the sub- edge after deformation, each third cutting image is combined, multiple shapes are obtained
Become image；Judge whether strain image meets preset deformation range, and then preset deformation range strain image will be met, really
It is set to the template image.It is operated by cutting, deformation, the combination etc. to edge, it is available to arrive different strain images, it opens up
The template image range of model has been opened up, and by judging whether strain image meets preset deformation range, will have been met preset
Deformation range strain image is determined as the template image, it is ensured that template image is the flexible article of standard deformation, is known for object
Good applicable basis is not provided.
On the basis of the above embodiments, as an optional implementation manner, as shown in fig. 6, S203 is " according to each described
Second cutting image of the first cutting image and each corresponding position, identifies the object under test of the object under test image ", comprising:
S601 extracts described the second of the first cutting image and corresponding position according to preset extraction algorithm respectively
The side right value information of cutting image.
Wherein, side right value information is dilute for indicating a little relationship between side, the common form of expression of side right value information
Matrix is dredged, the side right value information of the first cutting image can be extracted by preset extraction algorithm；And it is calculated by preset extraction
Method extracts the side right value information of the second cutting image of corresponding position.
Optionally, as an optional implementation manner, as shown in fig. 7, " according to preset extraction algorithm, extracting respectively
The side right value information of the second cutting image of the first cutting image and corresponding position ", comprising:
S701 obtains the key node of the first cutting image Yu the second cutting image respectively.
Wherein it is possible to by scale invariant feature transfer algorithm (Scale-invariant feature transform,
SIFT) or maximum stable extremal region detection algorithm (Maximally Stable Extremal Regions, MSER) extracts the
The key node of all partial images and the second cutting image.
S702 makees subdivision processing to the key node, obtains subdivision result.
Specifically, subdivision processing can be made to the first cutting image key node using Delaunay Triangulation algorithm,
Obtain the first cutting image subdivision result；The second cutting image key node can be made using Delaunay Triangulation algorithm
Subdivision processing, obtains the second cutting image subdivision result.
S703 extracts side right value information according to the subdivision result.
Wherein, side right value information is dilute for indicating a little relationship between side, the common form of expression of side right value information
Matrix is dredged, optionally, available preset extraction algorithm extracts side right value information.
By obtaining the key node of the first cutting image and the second cutting image respectively, subdivision then is made to key node
Processing obtains subdivision as a result, extracting side right value information according to subdivision result in turn, to find the first cutting image and the second cutting
Point correspondence between image provides basis, and then guarantees to realize object identification.
S602 identifies the object under test of the object under test image according to each side right value information.
Specifically, the whole phase between the first cutting image and the second cutting image can be obtained according to side right value information
Like degree, and then according to the object under test of overall similarity identification object under test image.
Optionally, as an optional implementation manner, as shown in figure 8, " according to each side right value information, identifying institute
State the object under test of object under test image ", comprising:
S801, according to the first cutting image and the second cutting image, obtain the first cutting image with it is right
Answer the first similarity between each second cutting image of position.
Illustratively, as shown in figure 9, for object under test image PA, measuring targets image PA carries out cutting, obtains the
All partial images be etc. sizes 9 equal portions PA_1, PA_2, PA_3 to PA_9, template image PA1, PA2, PA3, equally, to template
Image PA1, PA2, PA3 carry out cutting, obtain the second cutting image PA1_1 to PA1_9, PA2_1 to PA2_9, PA3_1 to PA3_
9, then the similarity between PA_1 and PA1_1, PA2_1, PA3_1 is obtained, is computed and obtains, PA_1 and PA1_1, PA2_1, PA3_1
The first similarity be respectively 0.9,0.8,0.7, then the first similarity of PA_2 Yu PA1_2, PA2_2, PA3_2 are obtained, with this
Analogize, obtains the first similarity between each first cutting image and corresponding second cutting image.
S802 obtains between the object under test image and each template image according to first similarity
Two similarities.
Specifically, the first similarity of non-maximum value can be assigned a value of 0, for example, PA_1 and PA1_1, PA2_1, PA3_1
The first similarity be respectively 0.9,0.8,0.7, then the similarity of PA_1 and PA1_1, PA2_1, PA3_1 can be converted into
0.9、0、0。
It is alternatively possible to obtain the object under test image and each institute according to first similarity and corresponding weight
State the second similarity between template image.
Illustratively, if be computed obtain object under test image PA and template image PA1 subgraph similarity be respectively 0.9,
0.8,0.9,0.6,0.9,0.8,0.7,0.8,0.9, it can preset in object under test image that feature is most heavy in the subgraph in the most upper left corner
It wants, therefore the feature weight of the subgraph in the most upper left corner is assigned a value of 2, the feature weight of other 8 position subgraphs is assigned a value of 1, then root
According to the first similarity and preset weight, the second similarity of PA and PA1 are 0.9*2+0*1+0.9*1+0*1+0.9*1+0.8*1
+ 0*1+0.8*1+0.9*1=6.By the method, available PA is similar to the second of PA2 to the second similarity of PA1, PA
Second similarity of degree and PA and PA3.
S803 identifies the object under test of the object under test image according to second similarity.
Illustratively, compare the second similarity, the second similarity and PA of PA and PA2 and the second phase of PA3 of PA and PA1
Like degree, if the second similarity of PA and PA2 is maximum, by the highest changing object PA2 of the second similarity, object under test is identified
The object under test of image.
Alternatively it is also possible to the object under test of object under test image is identified by the corresponding number of changing object PA2,
And it can posture by the identification object of the corresponding deformed state of storage changing object and spatial position.
In above-described embodiment, according to preset extraction algorithm, the first cutting image and corresponding position are extracted respectively
The side right value information of the second cutting image, and then according to each side right value information, identify the object under test image
Object under test, and then for object identification also may be implemented there are the flexible article of deformation.
It should be understood that although each step in the flow chart of Fig. 2-8 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-8
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
In one embodiment, as shown in Figure 10, a kind of flexible article identification device is provided, including first obtains module
11, second obtains module 12, identification module 13, in which:
First obtains module 11, for obtaining object under test image and preset multiple template image；The template image
For describing multiple deformed states of object under test；
Second acquisition module 12 is obtained for the object under test image and each template image to be carried out cutting respectively
Take the corresponding multiple first cutting images of the object under test image and the corresponding multiple second cutting figures of each template image
Picture；
Identification module 13, for the second cutting image according to each first cutting image and each corresponding position, identification
The object under test of the object under test image.
In one of the embodiments, as shown in figure 11, on the basis of shown in Fig. 10, described device further includes that third obtains
Modulus block the 14, the 4th obtains module 15, extraction module 16, composite module 17, in which:
Third obtains module 14, for obtaining multiple original objects images；The multiple original objects image includes object
Multiple posture informations；
4th obtains module 15, for each original objects image to be carried out cutting, obtains multiple third cutting images；
Extraction module 16, for extracting the edge of each third cutting image；
Composite module 17, for being combined to each third cutting image, obtaining the template according to the edge
Image.
In one of the embodiments, as shown in figure 12, shown in Figure 11 on the basis of, composite module 17 includes:
Cutting unit 170 obtains multiple sub- edges for carrying out cutting to each edge；
Deformation unit 171, for carrying out deformation operation to each sub- edge, the sub- edge after obtaining deformation；
Assembled unit 172, for being combined to each third cutting image according to the sub- edge after the deformation,
Obtain multiple strain images；
Judging unit 173, for judging whether the strain image meets preset deformation range；
Determination unit 174 determines the strain image if meeting preset deformation range for the strain image
For the template image.
In one of the embodiments, as shown in figure 13, on the basis of shown in Fig. 10, identification module 13 includes:
Extraction unit 130, for extracting the first cutting image and corresponding position respectively according to preset extraction algorithm
The second cutting image side right value information；
Recognition unit 131, for identifying the object under test of the object under test image according to each side right value information.
In one of the embodiments, the extraction unit 130 be specifically used for obtain respectively the first cutting image with
The key node of the second cutting image；Subdivision processing is made to the key node, obtains subdivision result；According to the subdivision
As a result side right value information is extracted.
Recognition unit 131 is specifically used for according to the first cutting image and described second in one of the embodiments,
Cutting image obtains the first similarity between the first cutting image and each second cutting image of corresponding position；
According to first similarity, the second similarity between the object under test image and each template image is obtained；According to
Second similarity identifies the object under test of the object under test image.
Recognition unit 131 is specifically used for according to first similarity and corresponding weight in one of the embodiments,
Obtain the second similarity between the object under test image and each template image.
Specific about flexible article identification device limits the limit that may refer to above for flexible article recognition methods
Fixed, details are not described herein.Modules in above-mentioned flexible article identification can be fully or partially through software, hardware and combinations thereof
To realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with soft
Part form is stored in the memory in computer equipment, executes the corresponding behaviour of the above modules in order to which processor calls
Make.
In one embodiment, a kind of computer equipment, including memory and processor are provided, is stored in memory
Computer program, the processor perform the steps of when executing computer program
Obtain object under test image and preset multiple template image；The template image is for describing the more of object under test
A deformed state；
The object under test image and each template image are subjected to cutting respectively, obtain the object under test image pair
The corresponding multiple second cutting images of multiple first cutting images and each template image answered；
According to the second cutting image of each first cutting image and each corresponding position, the object under test image is identified
Object under test.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains multiple original objects
Image；The multiple original objects image includes multiple posture informations of object；Each original objects image is subjected to cutting,
Obtain multiple third cutting images；Extract the edge of each third cutting image；According to the edge, each third is cut
Partial image is combined, and obtains the template image.
In one embodiment, it is also performed the steps of when processor executes computer program and each edge is carried out
Cutting obtains multiple sub- edges；Deformation operation is carried out to each sub- edge, the sub- edge after obtaining deformation；According to the shape
Sub- edge after change is combined each third cutting image, obtains multiple strain images；Judging the strain image is
It is no to meet preset deformation range；If the strain image meets preset deformation range, the strain image is determined as
The template image.
In one embodiment, it also performs the steps of when processor executes computer program and is calculated according to preset extraction
Method extracts the side right value information of the second cutting image of the first cutting image and corresponding position respectively；According to each institute
Side right value information is stated, identifies the object under test of the object under test image.
In one embodiment, it is also performed the steps of when processor executes computer program and obtains described first respectively
The key node of cutting image and the second cutting image；Subdivision processing is made to the key node, obtains subdivision result；Root
Side right value information is extracted according to the subdivision result.
In one embodiment, it also performs the steps of when processor executes computer program according to first cutting
Image and the second cutting image obtain between the first cutting image and each second cutting image of corresponding position
The first similarity；According to first similarity, between the object under test image and each template image is obtained
Two similarities；According to second similarity, the object under test of the object under test image is identified.
In one embodiment, it is also performed the steps of when processor executes computer program similar according to described first
Degree and corresponding weight, obtain the second similarity between the object under test image and each template image.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Obtain object under test image and preset multiple template image；The template image is for describing the more of object under test
A deformed state；
The object under test image and each template image are subjected to cutting respectively, obtain the object under test image pair
The corresponding multiple second cutting images of multiple first cutting images and each template image answered；
According to the second cutting image of each first cutting image and each corresponding position, the object under test image is identified
Object under test.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains multiple original objects
Body image；The multiple original objects image includes multiple posture informations of object；Each original objects image is cut
Point, obtain multiple third cutting images；Extract the edge of each third cutting image；According to the edge, to each described
Three cutting images are combined, and obtain the template image.
In one embodiment, also performed the steps of when computer program is executed by processor to each edge into
Row cutting obtains multiple sub- edges；Deformation operation is carried out to each sub- edge, the sub- edge after obtaining deformation；According to described
Sub- edge after deformation is combined each third cutting image, obtains multiple strain images；Judge the strain image
Whether preset deformation range is met；If the strain image meets preset deformation range, the strain image is determined
For the template image.
In one embodiment, it also performs the steps of when computer program is executed by processor according to preset extraction
Algorithm extracts the side right value information of the second cutting image of the first cutting image and corresponding position respectively；According to each
The side right value information identifies the object under test of the object under test image.
In one embodiment, it is also performed the steps of when computer program is executed by processor and obtains described respectively
The key node of all partial images and the second cutting image；Subdivision processing is made to the key node, obtains subdivision result；
Side right value information is extracted according to the subdivision result.
In one embodiment, it also performs the steps of when computer program is executed by processor and is cut according to described first
Partial image and the second cutting image, obtain the first cutting image and corresponding position each second cutting image it
Between the first similarity；According to first similarity, obtain between the object under test image and each template image
Second similarity；According to second similarity, the object under test of the object under test image is identified.
In one embodiment, it is also performed the steps of when computer program is executed by processor according to first phase
Like degree and corresponding weight, the second similarity between the object under test image and each template image is obtained.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of flexible article recognition methods, which is characterized in that the described method includes:
Obtain object under test image and preset multiple template image；The template image is used to describe multiple shapes of object under test
Change state；
The object under test image and each template image are subjected to cutting respectively, it is corresponding to obtain the object under test image
Multiple first cutting images and the corresponding multiple second cutting images of each template image；
According to the second cutting image of each first cutting image and each corresponding position, identify the object under test image to
Survey object.
2. the method according to claim 1, wherein the method also includes:
Obtain multiple original objects images；The multiple original objects image includes multiple posture informations of object；
Each original objects image is subjected to cutting, obtains multiple third cutting images；
Extract the edge of each third cutting image；
According to the edge, each third cutting image is combined, the template image is obtained.
3. according to the method described in claim 2, it is characterized in that, described according to the edge, to each third cutting figure
As being combined, described image template is obtained, comprising:
Cutting is carried out to each edge, obtains multiple sub- edges；
Deformation operation is carried out to each sub- edge, the sub- edge after obtaining deformation；
According to the sub- edge after the deformation, each third cutting image is combined, multiple strain images are obtained；
Judge whether the strain image meets preset deformation range；
If the strain image meets preset deformation range, the strain image is determined as the template image.
4. method according to claim 1-3, which is characterized in that it is described according to each first cutting image and
Second cutting image of each corresponding position, identifies the object under test of the object under test image, comprising:
According to preset extraction algorithm, the second cutting image of the first cutting image and corresponding position is extracted respectively
Side right value information；
According to each side right value information, the object under test of the object under test image is identified.
5. according to the method described in claim 4, it is characterized in that, described according to preset extraction algorithm, extract respectively described in
The side right value information of the second cutting image of first cutting image and corresponding position, comprising:
The key node of the first cutting image Yu the second cutting image is obtained respectively；
Subdivision processing is made to the key node, obtains subdivision result；
Side right value information is extracted according to the subdivision result.
6. according to the method described in claim 5, it is characterized in that, described according to each side right value information, identification it is described to
Survey the object under test of subject image, comprising:
According to the first cutting image and the second cutting image, each of the first cutting image and corresponding position is obtained
The first similarity between the second cutting image；
According to first similarity, the second similarity between the object under test image and each template image is obtained；
According to second similarity, the object under test of the object under test image is identified.
7. according to the method described in claim 6, acquisition is described to be measured it is characterized in that, described according to first similarity
The second similarity between subject image and each template image, comprising:
According to first similarity and corresponding weight, obtain between the object under test image and each template image
Second similarity.
8. a kind of flexible article identification device, which is characterized in that described device includes:
First obtains module, for obtaining object under test image and preset multiple template image；The template image is for retouching
State multiple deformed states of object under test；
Second obtains module, for the object under test image and each template image to be carried out cutting respectively, described in acquisition
The corresponding multiple first cutting images of object under test image and the corresponding multiple second cutting images of each template image；
Identification module, for the second cutting image according to each first cutting image and each corresponding position, identification it is described to
Survey the object under test of subject image.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 7 the method when executing the computer program.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811623177.1A CN109766802A (en) | 2018-12-28 | 2018-12-28 | Flexible article recognition methods, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811623177.1A CN109766802A (en) | 2018-12-28 | 2018-12-28 | Flexible article recognition methods, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109766802A true CN109766802A (en) | 2019-05-17 |
Family
ID=66451645
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811623177.1A Pending CN109766802A (en) | 2018-12-28 | 2018-12-28 | Flexible article recognition methods, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109766802A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090185715A1 (en) * | 2008-01-18 | 2009-07-23 | Mvtec Software Gmbh | System and method for deformable object recognition |
| CN103400136A (en) * | 2013-08-13 | 2013-11-20 | 苏州大学 | Target identification method based on elastic matching |
| CN103605979A (en) * | 2013-12-03 | 2014-02-26 | 苏州大学张家港工业技术研究院 | Object identification method and system based on shape fragments |
| US20150023601A1 (en) * | 2013-07-19 | 2015-01-22 | Omnivision Technologies, Inc. | Robust analysis for deformable object classification and recognition by image sensors |
-
2018
- 2018-12-28 CN CN201811623177.1A patent/CN109766802A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090185715A1 (en) * | 2008-01-18 | 2009-07-23 | Mvtec Software Gmbh | System and method for deformable object recognition |
| US20150023601A1 (en) * | 2013-07-19 | 2015-01-22 | Omnivision Technologies, Inc. | Robust analysis for deformable object classification and recognition by image sensors |
| CN103400136A (en) * | 2013-08-13 | 2013-11-20 | 苏州大学 | Target identification method based on elastic matching |
| CN103605979A (en) * | 2013-12-03 | 2014-02-26 | 苏州大学张家港工业技术研究院 | Object identification method and system based on shape fragments |
Non-Patent Citations (3)
| Title |
|---|
| 胡杰生等: "FLEXIBLE 3D OBJECT RECOGNITION FRAMEWORK USING 2D VIEWS VIA A SIMILARITY-BASED ASPECT-GRAPH APPROACH", INTERNATIONAL JOURNAL OF PATTERN RECOGNITON AND ARTIFICIAL INTELLIGENCE, vol. 22, no. 6, pages 1141 - 1169, XP001516023, DOI: 10.1142/S0218001408006685 * |
| 董本志等: "机器人视觉中障碍物刚柔性的判别方法", 《计算机工程与应用》 * |
| 董本志等: "机器人视觉中障碍物刚柔性的判别方法", 《计算机工程与应用》, vol. 49, no. 12, 18 December 2012 (2012-12-18), pages 155 - 159 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110889325B (en) | Multitasking facial motion recognition model training and multitasking facial motion recognition method | |
| EP3457357B1 (en) | Methods and systems for surface fitting based change detection in 3d point-cloud | |
| CN108062784B (en) | Three-dimensional model texture mapping conversion method and device | |
| CN111797650B (en) | Obstacle identification method, obstacle identification device, computer equipment and storage medium | |
| US9715761B2 (en) | Real-time 3D computer vision processing engine for object recognition, reconstruction, and analysis | |
| WO2019201035A1 (en) | Method and device for identifying object node in image, terminal and computer readable storage medium | |
| CN110163193B (en) | Image processing method, image processing device, computer-readable storage medium and computer equipment | |
| CN111079632A (en) | Training method and device of text detection model, computer equipment and storage medium | |
| CN109683699B (en) | Method and device for realizing augmented reality based on deep learning and mobile terminal | |
| CN111968134B (en) | Target segmentation method, device, computer readable storage medium and computer equipment | |
| CN111583184A (en) | Image analysis method, network, computer device, and storage medium | |
| CN113469092B (en) | Character recognition model generation method, device, computer equipment and storage medium | |
| CN111062324A (en) | Face detection method and device, computer equipment and storage medium | |
| CN111310758A (en) | Text detection method and device, computer equipment and storage medium | |
| CN110175974B (en) | Image significance detection method, device, computer equipment and storage medium | |
| US20120033873A1 (en) | Method and device for determining a shape match in three dimensions | |
| CN114155365A (en) | Model training method, image processing method and related device | |
| US12002260B2 (en) | Automatic topology mapping processing method and system based on omnidirectional image information | |
| CN112287730A (en) | Gesture recognition method, device, system, storage medium and equipment | |
| CN109685029B (en) | Flexible object identification method, device, equipment and medium with complex space structure | |
| CN111860582B (en) | Image classification model construction method and device, computer equipment and storage medium | |
| CN112991429B (en) | Box volume measuring method, device, computer equipment and storage medium | |
| WO2021092797A1 (en) | Image registration method, terminal, and computer storage medium | |
| CN111914890A (en) | Image block matching method between images, image registration method and product | |
| CN111291716B (en) | Sperm cell identification method, sperm cell identification device, computer equipment and storage medium |
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 | ||
| AD01 | Patent right deemed abandoned | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20240419 |