CN110210346A - A kind of optimization method that video template matching is handled in real time - Google Patents
A kind of optimization method that video template matching is handled in real time Download PDFInfo
- Publication number
- CN110210346A CN110210346A CN201910422598.6A CN201910422598A CN110210346A CN 110210346 A CN110210346 A CN 110210346A CN 201910422598 A CN201910422598 A CN 201910422598A CN 110210346 A CN110210346 A CN 110210346A
- Authority
- CN
- China
- Prior art keywords
- data
- pixel value
- sliding
- image
- 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 34
- 238000005457 optimization Methods 0.000 title claims abstract description 15
- 238000005070 sampling Methods 0.000 claims abstract description 6
- 230000015654 memory Effects 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012512 characterization method Methods 0.000 claims description 2
- 230000001186 cumulative effect Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 2
- 238000013500 data storage Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/48—Matching video sequences
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of optimization method that video template matching is handled in real time is proposed, method includes the following steps: reading and storing the pixel Value Data of template image；Sliding window identical with template image T size is constructed, sliding sampling is carried out to background image using sliding window, sliding is corresponding every time obtains a secondary subgraph, reads the pixel Value Data of subgraph, and carry out real-time storage and update to the pixel Value Data of subgraph；Using minimum error method, the similarity of the pixel Value Data of the pixel Value Data and template image that slide corresponding subgraph every time is obtained, and then obtains the matching position of template image.The present invention is accelerated to calculate, finally realizes the real-time processing to video on the basis of guaranteeing effect using the global search strategy of template matching and sliding window traversal background image based on similarity using field programmable gate array.
Description
Technical field
The invention belongs to technical field of image processing, further relate to a kind of optimization that video template matching is handled in real time
Method can be used for matching template image in background image, and export matching position in real time.
Background technique
Template matching is the technology searched out in a sub-picture with given target template image, is widely used in image
Processing and field of signal processing, such as image retrieval, image registration, image recognition etc..However as the at full speed of science and technology
The information processing capacity of development, the work of image template matching treatment is increasing, while demand of the people to scan picture
It is higher and higher.During realizing the template matching of realtime graphic using general processor, the processing speed of image is
As the bottleneck of the technology, high-definition image is extremely difficult to handle in real time.Some outstanding matching algorithms are in desktop computer or work
It can be appropriately carried out on standing, and after they are applied to embedded system, since the limitations such as dominant frequency, storage resource must be right
Algorithm is simplified, and usually implementation effect is very poor for simplified algorithm, and some cannot even execute.So fast with hardware
Speed development, use site programmable gate array (FPGA) solve the problems, such as that scan picture has become a kind of trend.
Existing template matching implementation is undesirable in terms of real-time processing, and the principal element for influencing speed is search
It calculates and matched operation.In some schemes, in order to carry out carrying out traversal search to image, a large amount of image data is stored,
Frequently reading reduces speed for it；In other schemes, calculating section design is unreasonable, limits the clock frequency of system, shadow
Whole processing speed is rung.In addition to this largely storage and calculate can also occupy a large amount of hardware resource, increase realization at
This.
Summary of the invention
To solve the above-mentioned problems, the purpose of the present invention is to propose to a kind of video templates to match the optimization side handled in real time
Method, the calculating speed that the present invention has comprehensively considered matching effect and field programmable gate array are fast but be bad at the spy of complex logic
Point, the present invention are being protected using the global search strategy of template matching and sliding window traversal background image based on similarity
On the basis of demonstrate,proving effect, accelerates to calculate using field programmable gate array, finally realize the real-time processing to video.
In order to achieve the above object, the present invention is resolved using following technical scheme.
A kind of optimization method that video template matching is handled in real time, comprising the following steps:
Step 1, the size of template image T is set as m × n, and the size of background image G is M × N；It reads and stores template
The pixel Value Data ref [m × n] of image.
Step 2, sliding window A identical with template image T size is constructed, i.e. the size of sliding window A is m × n；Using
Sliding window A carries out sliding sampling to background image G, and sliding is corresponding every time obtains the subgraph that a secondary size is m × n, reads
The pixel Value Data img of subgraphp[m × n], and to the pixel Value Data img of subgraphp[m × n] carries out real-time storage and more
Newly.
Wherein, imgp[m × n] indicates the sub-image pixels Value Data that pth time sliding obtains.
Step 3, using minimum error method, the pixel Value Data img for sliding corresponding subgraph every time is obtained respectivelyp[m×
N] similarity with the pixel Value Data ref [m × n] of template image, and then the matching position of template image is obtained, complete template
The matching of image and background image.
Compared with prior art, the invention has the benefit that
Processing the present invention is based on programmable logic array to template image and background image, and lead to during realization
It crosses and designs the storage of reasonable data and improve the speed entirely handled with update method, and save field programmable gate array
Resource.Its basic ideas are as follows: global search is carried out to background image by sliding window, then using similarity calculation to template
Image is matched；By using data of multiple memories to the subgraph that sliding window obtains carry out implement storage and more
Newly, the speed for accelerating data storage and updating, saves resource, the matching position of final output template image.
Detailed description of the invention
The present invention is described in further details in the following with reference to the drawings and specific embodiments.
Fig. 1 is a kind of flow diagram for the optimization method that template matching is handled in real time provided by the invention.
Fig. 2 is the sliding process schematic diagram of sliding window.
Specific embodiment
The embodiment of the present invention and effect are described in further detail with reference to the accompanying drawing.
Step 1, the size of template image T is set as m × n, and the size of background image G is M × N；It reads and stores template
The pixel Value Data ref [m × n] of image；
It reads to obtain the pixel Value Data of the template image of m × n from external memory, due to carrying out a matching process
It says, template image is determining constant, and image is relatively small, and therefore, the pixel Value Data of template image is stored to deposit
In device, the pixel Value Data of template image is denoted as ref [m × n].
Step 2, sliding window A identical with template image T size is constructed, i.e. the size of sliding window A is m × n；Using
Sliding window A carries out sliding sampling to background image G, and sliding is corresponding every time obtains the subgraph that a secondary size is m × n, reads
The pixel Value Data img of subgraphp[m × n], and to the pixel Value Data img of subgraphp[m × n] carries out real-time storage and more
Newly；
Sliding sampling are as follows: sliding window A is since the upper left corner of background image G, from left to right, from top to bottom successively
Sliding, each clock cycle sliding is primary, slides a pixel every time.
The pixel Value Data img for reading subgraphp[m × n] specifically:
When sliding window A slides into the Far Left of every a line of background image, i.e., sliding window A is in background image G
Position top left co-ordinate be (0, y) when, by m clock cycle, read m × n pixel of background image left end
Value Data；
When sliding window A is not located at the Far Left of every a line of background image, the data of reading are by upper sliding position
(m-1) column pixel Value Data set, in addition last column pixel value data of current sliding position form, i.e., sliding corresponds to every time
One n row m column pixel Value Data；
The pixel Value Data img to subgraphp[m × n] carries out real-time storage, the specific steps are that:
To the pixel Value Data that each sliding obtains, (n-1) row correspondence is stored in (n-1) a memory by before, every row
Use a memory；Then each clock cycle reads (n-1) a picture of a upper sliding position from (n-1) a memory
Plain Value Data, as (n-1) a data before last column of current sliding position；It is read from external memory and works as front slide
One data of position, nth data of last column as current sliding position is to get last to current sliding position
One column data.
The pixel Value Data img to subgraphp[m × n] carries out real-time update, the specific steps are that:
Data in memory update: firstly, being 0 by address in memory when sliding window slides to the right a pixel
Data be updated, be that the first row address is written is to complete at 0 to the first row data for data at 0 by the second row address
It updates, and so on, the data address by (n-1) row is that data write-in (n-2) row address at 0 is to complete (n- at 0
2) update of row data；Secondly, the data from being 0 for the line n address read from external memory write (n-1) row address
At 0, the update of (n-1) row data is completed；Finally, when sliding window slides to the right a pixel again, then by memory
The data that middle address is 1 are updated, and are that the first row address is written is to complete at 1 to first for data at 1 by the second row address
The update of row data, and so on, the data address by (n-1) row is that the data at 1 are written (n-2) row address and are at 1,
Complete the update of (n-2) row data；Secondly, the data from being 1 for the line n address read from external memory write (n-
1) row address is to complete the update of (n-1) row data at 1.And so on, in the memory for completing each sliding window sliding
Data update.
Every time by the pixel Value Data of m × n background image in current window A, storage is in a register；Work as sliding window
When mouth sliding, the corresponding data in register are updated.
It in such a manner, has been updated data in memory, it is only necessary to press when sliding window A is moved down
Repeat according to operation before, until sliding window A slides into the lower right corner of background image, the i.e. right side of sliding window A
Lower angular coordinate is consistent with the bottom right angular coordinate of entire image, i.e. completion shiding matching process.
Step 3, using minimum error method, the pixel Value Data img for sliding corresponding subgraph every time is obtained respectivelyp[m×
N] similarity with the pixel Value Data ref [m × n] of template image, and then the matching position of template image is obtained, complete template
The matching of image and background image.
Template matching based on similarity, main thought are: according to different similarity functions come calculation template image
With the similarity of sliding window institute overlay area data in image to be matched.Comprehensively consider the effect of template matching and calculates complicated
Degree, the present invention calculate similarity using minimum error method.
The main thought of minimum error method is: according to the difference of template image and sliding window institute overlay area image come table
Show their correlation.After sliding window has traversed all pixels in background image, taking the smallest region of difference is template
Position of the image in background image.Common smallest error function has variance, mean square deviation, mean absolute error and absolutely misses
Difference.The calculating of minimum error method is mainly concerned with addition and subtraction, seldom uses multiplication and division.Therefore, minimum error method
Have the characteristics that realize that simple, operand is small.
Sliding sliding window every time, correspondence obtain the similarity of current sliding window mouth position virgin's image and template image,
The corresponding similarity in current sliding window mouth position similarity corresponding with last sliding window position is compared, similarity
Higher position is the position of the template image of current matching；To get arriving and template after sliding window traverses background image
The position of the highest sliding window of image similarity, the as matching position of template image；
Wherein, similarity uses absolute error and characterization, and similarity is higher, absolute error and smaller.
Use smallest error function for absolute error and (SAD) in the present invention.
The calculation formula of absolute error and function are as follows:G (i, j) is in sliding window
Background image pixel value, T (i, j) is the pixel value of template image, | | to take absolute value symbol, (i, j) indicates image
In pixel coordinate, D be background image and template image absolute error and.
The pixel value of the corresponding position of template image and background image is subtracted each other, then is taken absolute value to difference, finally,
M × n absolute value is added, obtain two images absolute error and, absolute error and smaller shows that similarity is bigger, absolutely
It is matched template image position to error and the smallest position.
The present invention using DSP48E1 calculate background image and template image absolute error and, DSP48E1 is a kind of scene
Programmable gate array hardware resource, it is digital signal processing unit, it can be achieved that the operations such as multiplication, addition, multiply-add, use the money
Source carries out calculating other resources that can save field programmable gate array.There is more signed magnitude arithmetic(al) when calculating similarity,
It can be used DSP48E1 to save resource.
Using DSP48E1 calculate background image and template image absolute error and, specific calculating process are as follows:
Firstly, calculating corresponding difference: the respective pixel value of template image and the subgraph of background image is subtracted each other；Wherein,
Each pixel value is 8 data, and DSP48E1 is able to carry out the addition and subtraction of 48 data, and four pixel values are combined meter
Subtraction is calculated, i.e., calculates four pixel values every time；Addition 0 adds 0, so that often for borrowing before minuend before subtrahend again
A data become 9 digits, are 36 digits after combination, recall DSP48E1 primitive and carry out subtraction, again will be poor after the completion of calculating
The correspondence digit of value is split, and reverts to four pixel values；Traverse background image, obtain the subgraph of each background image with
The correspondence difference of template image；
Secondly, taking absolute value to each corresponding difference: the highest order of the corresponding difference is sign bit, and least-significant byte is data；
Judge whether the highest order of the corresponding difference borrows, if highest order is 0, borrow, shows that corresponding difference is negative,
Then least-significant byte, which is negated, becomes positive number to get the absolute value of the corresponding difference；Otherwise, the least-significant byte data of the corresponding difference are
For the absolute value of corresponding difference；
Finally, carry out cumulative summation to the absolute value of all corresponding differences, the absolute of background image and template image is obtained
Error and；
Specifically: the absolute value of all corresponding differences is first obtained into first order summed result by summing in pairs,
First order summed result is obtained into second level summed result by summing in pairs again, and so on, obtain all correspondences
The absolute error of the sum of the absolute value of difference, as background image and template image and.It is above-mentioned to add up to all absolute values
Summation also can be used DSP48E1 and carry out summation operation.
Above-mentioned matching process can also state are as follows: sliding correspondence obtains current sad value every time, by current sad value with
The high sad value of the obtained similarity of last time sliding is compared, using smaller sad value as working as the high sad value of previous similarity,
To get being the highest sad value of similarity to the smallest sad value at the end of sliding, the corresponding position of the window is to match
The matching of template image and background image is completed in position.
Above step is carried out by way of assembly line, each clock cycle available sad value, and background image is defeated
The sad value of all positions has been had stepped through after entering, and has completed the comparison of all sad values, exports matching position.It is final simultaneously
The clock frequency of realization can achieve 100MHz, can be quickly obtained its matching position for each frame image, can
Image is handled in real time.
This can be accomplished by hardware associated with program instructions for all or part of the steps of the present invention, and program above-mentioned can deposit
It is stored in a computer-readable storage medium, which when being executed, executes step including the steps of the foregoing method embodiments；And it is preceding
The storage medium stated includes: the various media that can store program code such as ROM, RAM, magnetic or disk.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (8)
1. a kind of video template matches the optimization method handled in real time, which comprises the following steps:
Step 1, the size of template image T is set as m × n, and the size of background image G is M × N；It reads and stores template image
Pixel Value Data ref [m × n]；
The pixel Value Data of template image T is read from external memory, and is stored in register；
Step 2, sliding window A identical with template image T size is constructed, i.e. the size of sliding window A is m × n；Using sliding
Window A carries out sliding sampling to background image G, and sliding is corresponding every time obtains the subgraph that a secondary size is m × n, reads subgraph
The pixel Value Data img of picturep[m × n], and to the pixel Value Data img of subgraphp[m × n] carries out real-time storage and update；
Wherein, imgp[m × n] indicates the sub-image pixels Value Data that pth time sliding obtains；
Step 3, using minimum error method, the pixel Value Data img for sliding corresponding subgraph every time is obtained respectivelyp[m × n] with
The similarity of the pixel Value Data ref [m × n] of template image, and then the matching position of template image is obtained, complete template image
With the matching of background image.
2. a kind of video template according to claim 1 matches the optimization method handled in real time, which is characterized in that the cunning
Dynamic sampling are as follows: sliding window A is successively slided, each clock from left to right, from top to bottom since the upper left corner of background image G
Period sliding is primary, slides a pixel every time.
3. a kind of video template according to claim 1 matches the optimization method handled in real time, which is characterized in that the reading
Take the pixel Value Data img of subgraphp[m × n] specifically:
When sliding window A slides into the Far Left of every a line of background image, i.e. position of the sliding window A in background image G
When the top left co-ordinate set is (0, y), by m clock cycle, m × n pixel value number of background image left end is read
According to；
When sliding window A is not located at the Far Left of every a line of background image, the data of reading are by a upper sliding position
(m-1) column pixel Value Data, in addition last column pixel value data of current sliding position form, i.e., sliding is one corresponding every time
N row m column pixel Value Data.
4. a kind of video template according to claim 3 matches the optimization method handled in real time, which is characterized in that described right
The pixel Value Data img of subgraphp[m × n] carries out real-time storage, the specific steps are that:
To the pixel Value Data that each sliding obtains, (n-1) row correspondence is stored in (n-1) a memory by before, and every enforcement is used
One memory；Then each clock cycle reads (n-1) a pixel value of a upper sliding position from (n-1) a memory
Data, as (n-1) a data before last column of current sliding position；Current sliding position is read from external memory
A data, as current sliding position last column nth data to get to current sliding position last column
Data.
5. a kind of video template according to claim 4 matches the optimization method handled in real time, which is characterized in that described right
The pixel Value Data img of subgraphp[m × n] carries out real-time update, the specific steps are that:
Data in memory update: firstly, when sliding window slides to the right a pixel, being by address in memory0Data
It is updated, is that the first row address is written is to complete the update to the first row data at 0 for data at 0 by the second row address, according to
Secondary to analogize, the data address by (n-1) row is that data write-in (n-2) row address at 0 is to complete (n-2) row data at 0
Update；Secondly, the data from being 0 for the line n address read from external memory write (n-1) row address is at 0, it is complete
At the update of (n-1) row data；Finally, being then 1 by address in memory when sliding window slides to the right a pixel again
Data be updated, be that the first row address is written is to complete at 1 to the first row data for data at 1 by the second row address
It updates, and so on, the data address by (n-1) row is that data write-in (n-2) row address at 1 is to complete (n- at 1
2) update of row data；Secondly, the data from being 1 for the line n address read from external memory write (n-1) row address
At 1, the update of (n-1) row data is completed；And so on, complete data in the memory of each sliding window sliding more
Newly；
Every time by the pixel Value Data of m × n background image in current window A, storage is in a register；When sliding window is sliding
When dynamic, the corresponding data in register are updated.
6. a kind of video template according to claim 1 matches the optimization method handled in real time, which is characterized in that it is described most
The smallest error function that the small theory of error uses for absolute error and, its calculation formula is:
Wherein, G (i, j) is the pixel value of the background image in sliding window, and T (i, j) is the pixel value of template image, | | be
Take absolute value symbol, and (i, j) indicates the pixel coordinate in image, D be background image and template image absolute error and.
7. a kind of video template according to claim 1 matches the optimization method handled in real time, which is characterized in that described point
The pixel Value Data img for sliding corresponding subgraph every time is not obtainedp[m × n] and template image pixel Value Data ref [m ×
N] similarity, and then obtain the matching position of template image, the specific steps are that:
Sliding sliding window every time, correspondence obtain the similarity of current sliding window mouth position virgin's image and template image, will work as
The corresponding similarity of front slide the window's position similarity corresponding with last sliding window position is compared, and similarity is higher
Position be current matching template image position；To get arriving and template image after sliding window traverses background image
The position of the highest sliding window of similarity, the as matching position of template image；
Wherein, similarity uses absolute error and characterization, and similarity is higher, absolute error and smaller.
8. a kind of video template according to claim 1 matches the optimization method handled in real time, which is characterized in that use
DSP48E1 calculate background image and template image absolute error and, specific calculating process are as follows:
Firstly, calculating corresponding difference: the respective pixel value of template image and the subgraph of background image is subtracted each other；Wherein, each
Pixel value is 8 data, and DSP48E1 is able to carry out the addition and subtraction of 48 data, and four pixel values are combined calculating and are subtracted
Method calculates four pixel values every time；Addition 0 adds 0, so that every number for borrowing before minuend before subtrahend again
It is 36 digits after combination according to 9 digits are become, recalls DSP48E1 primitive and carry out subtraction, again by difference after the completion of calculating
Corresponding digit is split, and reverts to four pixel values；Background image is traversed, the subgraph and template of each background image are obtained
The correspondence difference of image；
Secondly, taking absolute value to each corresponding difference: the highest order of the corresponding difference is sign bit, and least-significant byte is data；To institute
Whether the highest order for stating corresponding difference, which borrows, is judged, if highest order is 0, is borrowed, and is shown that corresponding difference is negative, then will
Least-significant byte, which negates, becomes positive number to get the absolute value of the corresponding difference；Otherwise, the least-significant byte data of the corresponding difference are pair
Answer the absolute value of difference；
Finally, carrying out cumulative summation to the absolute value of all corresponding differences, the absolute error of background image and template image is obtained
With；
Specifically: first by the absolute value of all corresponding differences by summing in pairs, first order summed result is obtained, then will
First order summed result obtains second level summed result by summing in pairs, and so on, obtain all corresponding differences
Absolute value sum, as the absolute error of background image and template image and.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910422598.6A CN110210346A (en) | 2019-05-21 | 2019-05-21 | A kind of optimization method that video template matching is handled in real time |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910422598.6A CN110210346A (en) | 2019-05-21 | 2019-05-21 | A kind of optimization method that video template matching is handled in real time |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110210346A true CN110210346A (en) | 2019-09-06 |
Family
ID=67787919
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910422598.6A Pending CN110210346A (en) | 2019-05-21 | 2019-05-21 | A kind of optimization method that video template matching is handled in real time |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110210346A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111598177A (en) * | 2020-05-19 | 2020-08-28 | 中国科学院空天信息创新研究院 | Self-adaptive maximum sliding window matching method facing low-overlapping image matching |
| CN111780781A (en) * | 2020-06-23 | 2020-10-16 | 南京航空航天大学 | Template matching vision and inertia combined odometer based on sliding window optimization |
| CN112967310A (en) * | 2021-02-04 | 2021-06-15 | 成都国翼电子技术有限公司 | FPGA-based template matching acceleration method |
| CN114584673A (en) * | 2020-12-01 | 2022-06-03 | 京东方科技集团股份有限公司 | Image processing method and device |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060002472A1 (en) * | 2004-06-30 | 2006-01-05 | Mehta Kalpesh D | Various methods and apparatuses for motion estimation |
| US20060002471A1 (en) * | 2004-06-30 | 2006-01-05 | Lippincott Louis A | Motion estimation unit |
| US20090028440A1 (en) * | 2007-07-27 | 2009-01-29 | Sportvision, Inc. | Detecting an object in an image using multiple templates |
| US20100020880A1 (en) * | 2008-07-22 | 2010-01-28 | Mathstar, Inc. | Field programmable object array having image processing circuitry |
| CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
| CN105761233A (en) * | 2014-12-15 | 2016-07-13 | 南京理工大学 | FPGA-based real-time panoramic image mosaic method |
| US20170339404A1 (en) * | 2016-05-17 | 2017-11-23 | Arris Enterprises Llc | Template matching for jvet intra prediction |
| CN109671042A (en) * | 2018-12-19 | 2019-04-23 | 西安电子科技大学 | Gray-scale image processing system and method based on FPGA morphological operator |
| CN109743562A (en) * | 2019-01-10 | 2019-05-10 | 中国科学技术大学 | Matching cost counting circuit structure and its working method based on Census algorithm |
-
2019
- 2019-05-21 CN CN201910422598.6A patent/CN110210346A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060002472A1 (en) * | 2004-06-30 | 2006-01-05 | Mehta Kalpesh D | Various methods and apparatuses for motion estimation |
| US20060002471A1 (en) * | 2004-06-30 | 2006-01-05 | Lippincott Louis A | Motion estimation unit |
| US20090028440A1 (en) * | 2007-07-27 | 2009-01-29 | Sportvision, Inc. | Detecting an object in an image using multiple templates |
| US20100020880A1 (en) * | 2008-07-22 | 2010-01-28 | Mathstar, Inc. | Field programmable object array having image processing circuitry |
| CN103514293A (en) * | 2013-10-09 | 2014-01-15 | 北京中科模识科技有限公司 | Method for video matching in video template library |
| CN105761233A (en) * | 2014-12-15 | 2016-07-13 | 南京理工大学 | FPGA-based real-time panoramic image mosaic method |
| US20170339404A1 (en) * | 2016-05-17 | 2017-11-23 | Arris Enterprises Llc | Template matching for jvet intra prediction |
| CN109671042A (en) * | 2018-12-19 | 2019-04-23 | 西安电子科技大学 | Gray-scale image processing system and method based on FPGA morphological operator |
| CN109743562A (en) * | 2019-01-10 | 2019-05-10 | 中国科学技术大学 | Matching cost counting circuit structure and its working method based on Census algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 周文晖;杜歆;叶秀清;顾伟康;: "基于FPGA的双目立体视觉系统" * |
| 沙莎;刘锦峰;: "基于差分有序数组的图像匹配快速算法" * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111598177A (en) * | 2020-05-19 | 2020-08-28 | 中国科学院空天信息创新研究院 | Self-adaptive maximum sliding window matching method facing low-overlapping image matching |
| CN111780781A (en) * | 2020-06-23 | 2020-10-16 | 南京航空航天大学 | Template matching vision and inertia combined odometer based on sliding window optimization |
| CN114584673A (en) * | 2020-12-01 | 2022-06-03 | 京东方科技集团股份有限公司 | Image processing method and device |
| CN114584673B (en) * | 2020-12-01 | 2024-01-09 | 京东方科技集团股份有限公司 | Image processing method and device |
| CN112967310A (en) * | 2021-02-04 | 2021-06-15 | 成都国翼电子技术有限公司 | FPGA-based template matching acceleration method |
| CN112967310B (en) * | 2021-02-04 | 2023-07-14 | 成都国翼电子技术有限公司 | Template matching acceleration method based on FPGA |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110210346A (en) | A kind of optimization method that video template matching is handled in real time | |
| US4635292A (en) | Image processor | |
| WO2018040463A1 (en) | Data compression and decompression methods for demura table, and mura compensation method | |
| CN109974743B (en) | Visual odometer based on GMS feature matching and sliding window pose graph optimization | |
| CN110058841B (en) | General computing device and method for nonlinear function with symmetry | |
| CN108198196A (en) | The design and implementation methods of quantum Image Edge-Detection based on Sobel operators | |
| US8068673B2 (en) | Rapid and high precision centroiding method and system for spots image | |
| JP6970827B2 (en) | Arithmetic processing unit | |
| CN113407537A (en) | Data processing method and device and electronic equipment | |
| KR102697687B1 (en) | Method of merging images and data processing device performing the same | |
| CN113468469A (en) | Convolution processing method and device of feature graph executed by computer and electronic equipment | |
| Motten et al. | A binary adaptable window SoC architecture for a stereo vision based depth field processor | |
| CN114581454B (en) | Quantum image segmentation method, device and storage medium based on background difference method | |
| CN106775596B (en) | A kind of infrared image linear interpolation expansion hardware processing method | |
| TW202324299A (en) | Image processing method, image processing system, and non-transitory computer readable storage medium | |
| CN111860492B (en) | License plate inclination correction method and device, computer equipment and storage medium | |
| Zhu et al. | Efficient illumination insensitive object tracking by normalized gradient matching | |
| CN113592908A (en) | Template matching target tracking and system based on Otsu method and SAD-MCD fusion | |
| CN112837349A (en) | Target tracking method, target tracking equipment and computer-readable storage medium | |
| CN115210758A (en) | Motion blur robust image feature matching | |
| CN106548466A (en) | The method and apparatus of three-dimensional reconstruction object | |
| CN110889904A (en) | Image feature simplifying method | |
| TWM588288U (en) | Goldbach conjecture computing system | |
| JP3522714B2 (en) | Image generation method | |
| US11580617B2 (en) | Method of matching images to be merged and data processing device performing the same |
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 |
Effective date of abandoning: 20231103 |
|
| AD01 | Patent right deemed abandoned |