CN107818539A - Conversion masterplate, fingerprint recognition or the computer vision or artificial intelligence or product quality detecting system of image profile data structure - Google Patents
Conversion masterplate, fingerprint recognition or the computer vision or artificial intelligence or product quality detecting system of image profile data structure Download PDFInfo
- Publication number
- CN107818539A CN107818539A CN201710995309.2A CN201710995309A CN107818539A CN 107818539 A CN107818539 A CN 107818539A CN 201710995309 A CN201710995309 A CN 201710995309A CN 107818539 A CN107818539 A CN 107818539A
- Authority
- CN
- China
- Prior art keywords
- point
- value
- data structure
- image
- profile data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/60—Memory management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The conversion masterplate of image profile data structure, constructed for orderly profile point set in dot chart is converted into image profile data structure;At least one packet of the data set of the image profile data structure contains tilt value, the tilt value that the member of the data of the image profile data structure is included be probably on the occasion of, zero, one kind in three kinds of situations of negative value, the statement of the tilt value that the members of the data of the image profile data structure is included be this point last point and this point last point the first line and the second line of the last point of this point and this point this two lines between deflection;Construction inquiry table array int M [a1] [a2] [b1] [b2] are used for the tilt value for storing this point for needing to be calculated in advance, a1, a2, b1, b2 numerical value are 3, assuming that there is the orderly point set { P0 of profile, P1, P2 }, then P2 tilt value is equal to the value stored by M [P0.X P1.X+1] [P0.Y P1.Y+1] [P2.X P1.X+1] [P2.Y P1.Y+1] D.The present invention can greatly improve operation efficiency, reduce cost, be more suitable for concurrent operation, improve recognition speed, there is provided new approaches.
Description
Technical field
The present invention relates to computer vision, specific image profile data structure, fingerprint recognition system, computer vision system
System, artificial intelligence system, product quality detecting system.
Background technology
Computer vision is the important module of artificial intelligence.
Vision is each application field, as various in the fields such as manufacturing industry, inspection, document analysis, medical diagnosis, and military affairs
Inalienable part in intelligence/autonomous system.
The processing of image outline is a more important step in computation vision, is had in a variety of application scenarios important
Meaning such as character identification system, certificate recognition system, fingerprint recognition system, iris authentication system, material surface texture recognition,
Products in kind quality testing, the identification of identification code code(Such as bar code, Quick Response Code etc.).
Computer vision system always with possess the mankind identify error correcting capability direction develop, but human brain have it is thousands of
The operational capability of hundred million neuron is that modern science and technology is still difficult to pursue(The most strong supercomputer in the whole world can not all simulate),
So function of many mankind in terms of figure identification, we are difficult to reach, and product size energy consumption is required at some
Occasion(Such as the occasion such as laptop computer, smart mobile phone, monitoring system, autonomous driving vehicle), existing computer vision system
It is difficult to remain some, room for improvement be present.
Human brain can easily identify the figure that diverse location in image, different rotary angle are put, and figure be rotated any
Angle, the mankind easily can quickly identify(Reason is the relative pass that the mankind can quickly obtain lines of outline various pieces
System), the operation efficiency of existing computer picture recognition system, which also exists, in this regard improves space, existing identifying system
Certain conflict between the processing speed of this aspect and discrimination being present, the discrimination that the high computing of discrimination is slow, computing is fast is low,
Room for improvement be present.
Concurrent operation(Such as CUDA technologies)It is to improve artificial intelligence system, the important side of NI Vision Builder for Automated Inspection operation efficiency
Method, but used by existing Image outline identification technology there is room for improvement in method in terms of parallel processing.
Present applicant proposes a kind of brand-new image profile data structure, develops a kind of brand-new computer picture lines
Or profile processing method, overcome above-mentioned technical problem.
The content of the invention
To overcome problem above, present applicant proposes image profile data structure, fingerprint recognition system, computer vision system
System, artificial intelligence system, product quality detecting system;With following scheme.
Scheme 1, image profile data structure, it is a kind of ordered data collection, it is characterised in that:The image profile data structure
The number of members of data set be at least 3, at least one packet of the data set of the image profile data structure contains tilt value, should
The tilt value that the member of the data of image profile data structure is included be probably on the occasion of, zero, one kind in three kinds of situations of negative value,
The statement for the tilt value that the member of the data of the image profile data structure is included be this point last point and this point it is upper
Deflection between this two lines of second line of the first line and this point of individual point and the last point of this point;The image outline
The method in the computing direction of the tilt value included of all members of the data set of data structure is consistent;The image outline
The sign symbol of the yawing moment of the tilt value included of all members of the data set of data structure represents yawing moment, number
Value represents deflection, and numerical value is bigger, and deflection is bigger.
Scheme 2, the image profile data structure as described in scheme 1, it is characterised in that:The data set of data structure owns
The plus sign of the yawing moment of the tilt value included of member represents deflection of turning left.(Assuming that a visual human is from last point
Set out via it is last point advance form ray path, if the left-hand side that this point is located at visual human the symbol of tilt value be just,
The symbol of tilt value is negative if the right-hand side that this point is located at visual human)
Scheme 3, the image profile data structure as described in scheme 1, it is characterised in that:All members of the data set of data structure
The plus sign of yawing moment of the tilt value included represent and turn right deflection.(Assuming that a visual human is from last point
Via it is last point advance form ray path, if the right-hand side that this point is located at visual human the symbol of tilt value be just, if
The symbol that this point is located at the left-hand side of visual human then tilt value is negative)
Scheme 4, the image profile data structure as described in scheme 1, it is characterised in that:If the position of the last point of this point with
This point overlaps, and the tilt value of this point is zero, that is to say, that the tilt value of this point is if this point is identical with last point
Zero.Such as in ordered data collection { P0, P1, P2 } data P2 coordinate it is equal with P0 coordinate, P2 tilt value is zero.
Scheme 5, the image profile data structure as described in scheme 1, it is characterised in that:The number range of tilt value is -3 to arrive
Positive integer or negative integer or zero between 3.
The conversion masterplate of scheme 6, image profile data structure, it is characterised in that:For by orderly profile point set in dot chart
The image profile data structure that is converted into described in scheme 1 and construct;Inquiry table array int M [a1] [a2] [b1] [b2] are constructed to use
In the tilt value for this point that advance storage needs to be calculated, a1, a2, b1, b2 numerical value are 3, that is to say, that M is one four
Dimension group, 3 members that each dimension includes in M four dimensions, a1 Query Value be the last point of this point X-coordinate and
The Query Value that the difference of the two X-coordinate value of the X-coordinate of last point of this point adds 1, a2 is the y-coordinate of the upper last point of this point
It is the X-coordinate and this point of this point plus 1, b1 Query Value with the differences of the y-coordinate of last point of this point the two y-coordinate values
The difference of the two X-coordinate value of the X-coordinate of last point is plus last point of the y-coordinate with this point that 1, b2 Query Value is this point
The difference of the two y-coordinate values of y-coordinate adds 1;Assuming that there is the orderly point set of profile { P0, P1, P2 }, then P2 tilt value is equal to M
Value stored by [P0.X-P1.X+1] [P0.Y-P1.Y+1] [P2.X-P1.X+1] [P2.Y-P1.Y+1] D;Array M is stored
Value be ready for before image is handled.1 is added to be in order to which the array index inserted is negative when preventing inquiry, array member
Subscript be since zero or 1, it is impossible to insert negative.
The conversion masterplate of the image profile data structure of scheme 7, such as scheme 6, it is characterised in that:The value of array M storages is led to
Cross operation program and realize and calculate, and write as fixed data storage in the storage device, real-time calling during for image procossing.
The conversion masterplate of the image profile data structure of scheme 8, such as scheme 1 or 6 image profile data structure, its feature
It is:The value of array M storages is obtained by artificial computing, and is write as fixed data storage in the storage device, for image procossing
When real-time calling.
The conversion masterplate of the image profile data structure of scheme 9, such as scheme 1 or 6 image profile data structure, its feature
It is:
The value of array M storages calculates acquisition by the following method:
The point centered on the last point of this point, using 8 points being centered around around central point as peripheral point, this point and this point it is upper
Last point must belong to peripheral point, and imparting auxiliary fortune is carried out to peripheral point according to clocking sequence with the last point position starting point of this point
Calculation value U, the auxiliary operation value of peripheral point is equal to the sequence number that peripheral point presses hour hands direction sequencing in addition to the last point of this point, this
The auxiliary operation value of the last point of point is 4;
The tilt value of this point subtracts 4 equal to auxiliary operation value.
Scheme 10, the clocking sequence as described in scheme 9 are clockwise.
Scheme 11, the clocking sequence as described in scheme 9 are counterclockwise.Such as Figure 11.
The conversion masterplate of the image profile data structure of scheme 12, such as scheme 1 or 6 image profile data structure, it is special
Sign is:
The value of array M storages calculates acquisition by the following method:
The point centered on the last point of this point, using 8 points being centered around around central point as peripheral point, this point and this point it is upper
Last point must belong to peripheral point, and the tilt value of peripheral point is followed successively by according to clocking sequence with the last point position starting point of this point
{ 3,2,1,0, -1, -2, -3 }, the tilt value of the last point of this point is 0.
Scheme 13, the clocking sequence as described in scheme 12 are clockwise.Such as Figure 10, three dot sequencies are P0- in Figure 12>P1->
P2
Scheme 14, the clocking sequence as described in scheme 12 are counterclockwise.Such as Figure 12, three dot sequencies are P0- in Figure 12>P1->P2.
Scheme 15, fingerprint recognition system, it is characterised in that:At the technical scheme described in either a program in scheme 1-11
Manage image.
Scheme 16, computer vision system, it is characterised in that:Using the technical scheme described in either a program in scheme 1-11
Handle image.
Scheme 17, artificial intelligence system, it is characterised in that:At the technical scheme described in either a program in scheme 1-11
Manage image.
Scheme 18, product quality detecting system, it is characterised in that:Using the technical side described in either a program in scheme 1-11
Case handles image.
The present invention can greatly improve operation efficiency, reduce cost, be more suitable for concurrent operation, improve recognition speed,
Provide a kind of image outline processing new approaches.The storage of image profile data structure can be the number commonly used in computer programming
Group may also mean that pin chained list or orderly hardware storage apparatus, and the acquisition of tilt value can be transported by computer software
Calculate and obtain and can also be obtained by the hardware specially designed for it.The present invention can handle annular closed point set(For example close
Orderly point set)Non-closed edge point set can also be handled(Such as straight line, camber line etc.).
Brief description of the drawings
Fig. 1 is the pending image for implementing example 1, and digital in bracket is meant that(X-coordinate, Y-coordinate, color),
Wherein(-1,-1,-1)It is the auxiliary magnet conveniently constructed for computing, facilitates computing.
Fig. 2-7 is the calculating process displaying to the tilt value of the point in Fig. 1, and wherein U is the value for auxiliary operation, and S is
Tilt value, there is this point for the point position current operation for writing S in figure.
Fig. 8 is schematic diagrames of the Fig. 1 via the tilt value ordered set obtained after process computing shown in Fig. 2-7, and tilt value has
Sequence collection is closed hoop.
Fig. 9 is the image after Fig. 8 middle figure is rotated and translated, and the tilt value that can be obtained after being handled is orderly
Collection displaying, Fig. 9 tilt value ordered set are equal with Fig. 8 tilt value ordered set.
Figure 10-12 is in order to which a variety of examples are done in the computing of the tilt value to the present invention.
Embodiment
Implement example 1, using the processing of the 4x4 pixel images shown in Fig. 1 as example, obtained using the method shown in Fig. 2-7
Image profile data structure, the number of members of the data set of the image profile data structure are at least 3, the image profile data structure
Data set at least one packet contain tilt value, the tilt value that the members of the data of the image profile data structure is included
Be probably on the occasion of, zero, one kind in three kinds of situations of negative value, the deflection that the members of the data of the image profile data structure is included
The statement of value is that the last point of this point and the first line of last point of this point and the second of the last point of this point and this point connect
Deflection between this two lines of line;The tilt value included of all members of the data set of the image profile data structure
The method in computing direction be consistent;The tilt value included of all members of the data set of the image profile data structure
The sign symbol of yawing moment represent yawing moment, numerical value represents deflection, and numerical value is bigger, and deflection is bigger.
First the point set of the image outline in image 1 is marked in order with prior art during this implementation example process
(Worm is such as but not limited to multiple means be present with method, raster scanning method, broken method etc. in acquisition of the prior art to image outline
Deng)The border marked in order is changed, its conversion method is:To be currently needed for this point of the point position of computing, with this point
Point centered on last point, 8 points to be centered around around central point must belong to as peripheral point, the last point of this point and this point
In peripheral point, imparting auxiliary operation value U is carried out according to clocking sequence to peripheral point with the last point position starting point of this point, except this point
Last point beyond the auxiliary operation value of peripheral point be equal to peripheral point and press the sequence number of hour hands direction sequencing, the last point of this point
Auxiliary operation value be 4;The tilt value of this point subtracts 4 equal to auxiliary operation value;Clocking sequence is clockwise.
This point refers to during computing to need the point for obtaining tilt value at that time, can be in order by the point set of image outline during computing
Computing is carried out one by one, can also be calculated using concurrent operation(Because the computing of this point is only related to the first two point, and need
The point of write-in data only has this point, it is possible to concurrent operation is carried out to each point of the point set of image outline, so can be with
Obtain faster arithmetic speed).This point is in Fig. 2(1,2,1), this point is in Fig. 3(2,2,1)This point is in Fig. 4(3,2,1),
This point is in Fig. 5(3,1,1), this point is in Fig. 6(2,0,1), this point is in Fig. 7(1,1,1), arrow represents point set in Fig. 1-7
Order direction.Fig. 1 image obtains Fig. 8 mark result, image profile data structure by the operation method shown in Fig. 2-7
A logically end to end annular array data { 1,2,1,2,0,2 } can be converted into.
Implement example 2, the profile of Fig. 9 image is changed with implementing method used in example 1, obtain in logic
Upper the first connected annular array data { 2,1,2,1,2,0 };Show by implementation example 1 obtains annular array data and implementation
It is logically end to end annular array data that example 2, which obtains annular array data, thus { 1,2,1,2,0,2 } with
2, and 1,2,1,2,0 } it is equal, diversified forms be present such as in array arrangement mode of the annular data in calculator memory
{ 0.2,1,2,1,2 }, { 2,0,2,1,2,1 }, { 1,2,0,2,1,2 } etc., no longer illustrate one by one.
Implement example 3, as shown in Figure 10, modification implement example 1 or this point tilt value operation method, with this point
Last point centered on point, using 8 points being centered around around central point as peripheral point, the last point of this point and this point must
Belong to peripheral point, the tilt value of peripheral point is followed successively by according to clocking sequence with the last point position starting point of this point 3,2,1,0 ,-
1, -2, -3 }, the tilt value of the last point of this point is 0.Clocking sequence is clockwise.
Implement example 4, as shown in figure 11, modification implement example 1 or this point tilt value operation method, with this point
Last point centered on point, using 8 points being centered around around central point as peripheral point, the last point of this point and this point must
Belong to peripheral point, the tilt value of peripheral point is followed successively by according to clocking sequence with the last point position starting point of this point 3,2,1,0 ,-
1, -2, -3 }, the tilt value of the last point of this point is 0.Clocking sequence is counterclockwise.
Implement example 5, as shown in figure 12, illustrate the processing method of this point situation equal with the last point of this point,
P2 is the last point that this point P1 is this point in figure, and P2 is the last point of this point, and P2 tilt value is 0.
Implement example 6, change the operation method for implementing example 1-5, make this point situation equal with the last point of this point
The tilt value of this point is equal to S=4.
Other explanations:Fig. 8, Fig. 9 illustrate that the profile of the similar image of different rotary angle can be unified for by the present invention
Ordered data collection equal or with close rule, enables to computer vision system to can be good at the word that identification is rotated
Symbol or image.
Claims (10)
1. the conversion masterplate of image profile data structure, it is characterised in that:For orderly profile point set in dot chart is converted into figure
Constructed as outline data structure;
Image profile data structure, it is a kind of ordered data collection, it is characterised in that:The data set of the image profile data structure
Number of members is at least 3, and at least one packet of the data set of the image profile data structure contains tilt value, the image outline number
The tilt value included according to the member of the data of structure be probably on the occasion of, zero, one kind in three kinds of situations of negative value, the image outline
The statement for the tilt value that the member of the data of data structure is included is the last point and the first of the last point of this point of this point
Deflection between this two lines of second line of the last point of line and this point and this point;The image profile data structure
The method in the computing direction of the tilt value included of all members of data set is consistent;The image profile data structure
The sign symbol of the yawing moment of the tilt value included of all members of data set represents yawing moment, and numerical value represents deflection
Degree, numerical value is bigger, and deflection is bigger;
Construction inquiry table array intM [a1] [a2] [b1] [b2] is used for the tilt value for storing this point for needing to be calculated in advance,
A1, a2, b1, b2 numerical value are 3, that is to say, that M is a four-dimensional array, each dimension includes in M four dimensions 3
Member, a1 Query Value are the differences of the two X-coordinate value of the X-coordinate Yu the X-coordinate of last point of this point of the last point of this point
The Query Value that value adds 1, a2 is the y-coordinate and the two y-coordinate values of the y-coordinate of last point of this point of the last point of this point
Difference adds plus the difference that 1, b1 Query Value is the X-coordinate with the X-coordinate of last point of this point of this point the two X-coordinate value
1, b2 Query Value is that the difference of the y-coordinate Yu the y-coordinate of last point of this point of this point the two y-coordinate values adds 1;Assuming that have
The orderly point set of profile { P0, P1, P2 }, then P2 tilt value be equal to M [P0.X-P1.X+1] [P0.Y-P1.Y+1] [P2.X-
P1.X+1] value stored by [P2.Y-P1.Y+1] D;The value of array M storages is ready for before image is handled.
2. the conversion masterplate of image profile data structure as claimed in claim 1, it is characterised in that:The value of array M storages is led to
Cross operation program and realize and calculate, and write as fixed data storage in the storage device, real-time calling during for image procossing.
3. the conversion masterplate of image profile data structure as claimed in claim 1, it is characterised in that:The value of array M storages is led to
Remarkable labour movement, which is calculated, to be obtained, and is write as fixed data storage in the storage device, real-time calling during for image procossing.
4. the conversion masterplate of image profile data structure as claimed in claim 1, it is characterised in that:
The value of array M storages calculates acquisition by the following method:
The point centered on the last point of this point, using 8 points being centered around around central point as peripheral point, this point and this point it is upper
Last point must belong to peripheral point, and imparting auxiliary fortune is carried out to peripheral point according to clocking sequence with the last point position starting point of this point
Calculation value U, the auxiliary operation value of peripheral point is equal to the sequence number that peripheral point presses hour hands direction sequencing in addition to the last point of this point, this
The auxiliary operation value of the last point of point is 4;
The tilt value of this point subtracts 4 equal to auxiliary operation value.
5. the conversion masterplate of image profile data structure as claimed in claim 1, it is characterised in that:Described clocking sequence is
Clockwise.
6. the conversion masterplate of image profile data structure as claimed in claim 1, it is characterised in that:Described clocking sequence is
Counterclockwise.
7. fingerprint recognition system, it is characterised in that:Image is handled using the technical scheme described in either a program in claim 1-6
Profile.
8. computer vision system, it is characterised in that:Figure is handled using the technical scheme described in either a program in claim 1-6
As profile.
9. artificial intelligence system, it is characterised in that:Image is handled using the technical scheme described in either a program in claim 1-6
Profile.
10. product quality detecting system, it is characterised in that:At the technical scheme described in either a program in claim 1-6
Manage image outline.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710995309.2A CN107818539B (en) | 2017-10-23 | 2017-10-23 | Construction method of conversion template of image contour data structure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710995309.2A CN107818539B (en) | 2017-10-23 | 2017-10-23 | Construction method of conversion template of image contour data structure |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107818539A true CN107818539A (en) | 2018-03-20 |
CN107818539B CN107818539B (en) | 2021-04-23 |
Family
ID=61607494
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710995309.2A Active CN107818539B (en) | 2017-10-23 | 2017-10-23 | Construction method of conversion template of image contour data structure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107818539B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1380986A2 (en) * | 2002-07-10 | 2004-01-14 | Northrop Grumman Corporation | System and method for analyzing a contour of an image by applying a sobel operator thereto |
CN101118544A (en) * | 2006-08-01 | 2008-02-06 | 华为技术有限公司 | Method for constructing picture shape contour outline descriptor |
CN103530639A (en) * | 2013-10-30 | 2014-01-22 | 刘伟 | Picture contour ordered point set extraction method |
CN104077775A (en) * | 2014-06-28 | 2014-10-01 | 中国科学院光电技术研究所 | Shape matching method and device combined with framework feature points and shape contexts |
-
2017
- 2017-10-23 CN CN201710995309.2A patent/CN107818539B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1380986A2 (en) * | 2002-07-10 | 2004-01-14 | Northrop Grumman Corporation | System and method for analyzing a contour of an image by applying a sobel operator thereto |
CN101118544A (en) * | 2006-08-01 | 2008-02-06 | 华为技术有限公司 | Method for constructing picture shape contour outline descriptor |
CN103530639A (en) * | 2013-10-30 | 2014-01-22 | 刘伟 | Picture contour ordered point set extraction method |
CN106960209A (en) * | 2013-10-30 | 2017-07-18 | 刘伟 | A kind of orderly point set extracting method of image outline, computer vision system |
CN104077775A (en) * | 2014-06-28 | 2014-10-01 | 中国科学院光电技术研究所 | Shape matching method and device combined with framework feature points and shape contexts |
Also Published As
Publication number | Publication date |
---|---|
CN107818539B (en) | 2021-04-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Merget et al. | Robust facial landmark detection via a fully-convolutional local-global context network | |
EP2858030B1 (en) | Performing a histogram using an array of addressable registers | |
JP6493163B2 (en) | Density search method and image processing apparatus | |
TWI469086B (en) | Method for image/video segmentation using texture feature | |
CN111310775A (en) | Data training method and device, terminal equipment and computer readable storage medium | |
CN110751037A (en) | Method for recognizing color of vehicle body and terminal equipment | |
US11017262B2 (en) | Method and device to determine image features using multi-resolution gradient analysis | |
CN110852311A (en) | Three-dimensional human hand key point positioning method and device | |
Codreanu et al. | GPU-ASIFT: A fast fully affine-invariant feature extraction algorithm | |
CN112025693B (en) | Pixel-level target capture detection method and system of asymmetric three-finger grabber | |
CN108256520B (en) | Method for identifying coin year, terminal equipment and computer readable storage medium | |
CN110210505A (en) | Generation method, device and the electronic equipment of sample data | |
CN113762159A (en) | Target grabbing detection method and system based on directional arrow model | |
JP5201184B2 (en) | Image processing apparatus and program | |
CN107818539A (en) | Conversion masterplate, fingerprint recognition or the computer vision or artificial intelligence or product quality detecting system of image profile data structure | |
CN107704861A (en) | Image profile data structure, character identification system, computer vision system, artificial intelligence system | |
Gopalan et al. | Pattern averaging technique for facial expression recognition using support vector machines | |
CN108229498B (en) | Zipper piece identification method, device and equipment | |
CN116486151A (en) | Image classification model training method, image classification method, device and storage medium | |
CN111260723B (en) | Barycenter positioning method of bar and terminal equipment | |
Beaini et al. | Novel convolution kernels for computer vision and shape analysis based on electromagnetism | |
Zhou et al. | E3CM: Epipolar-constrained cascade correspondence matching | |
CN111401365A (en) | OCR image automatic generation method and device | |
CN113111921A (en) | Object recognition method, object recognition device, electronic equipment and storage medium | |
Fulco et al. | Empirical evaluation of convolutional neural networks prediction time in classifying German traffic signs |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |