CN108845998A - A kind of trademark image retrieval matching process - Google Patents
A kind of trademark image retrieval matching process Download PDFInfo
- Publication number
- CN108845998A CN108845998A CN201810292274.0A CN201810292274A CN108845998A CN 108845998 A CN108845998 A CN 108845998A CN 201810292274 A CN201810292274 A CN 201810292274A CN 108845998 A CN108845998 A CN 108845998A
- Authority
- CN
- China
- Prior art keywords
- window
- image
- sliding window
- retrieved
- similar
- 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
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services; Handling legal documents
- G06Q50/184—Intellectual property management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Technology Law (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to field of image search more particularly to a kind of trademark image retrieval matching process, include the following steps：Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature is extracted, described image feature includes scale feature and space characteristics；Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all windows for meeting similar possibility in contrast images, similarity distance is calculated；Erroneous matching is eliminated using the method for scale-space consistency；Go out similar area according to adaptive threshold fuzziness；The sequence that search result is carried out according to the similitude of each contrast images similar area, exports the contrast images of search result.The method of the present invention accuracy is high, and omission factor is low.
Description
Technical field
The invention belongs to field of image search more particularly to a kind of trademark image retrieval matching process.
Background technique
Trade mark is one of important logo of commodity, and the registration of trade mark is for protecting the lawful right of trade mark holder very heavy
It wants, thus must first be retrieved in trade mark library before new trade mark registration, to guarantee that it has obviously with other registered trade marks
It is different.
Although content-based image retrieval technology has had a great development, it is still unable to satisfy the inspection of people
It asks for and asks, maximum difficulty is exactly：The high-rise language used when image bottom content characteristic and user search that system extracts
It can not be mapped between justice, that is to say, that characteristics of image is beyond expression the high-level semantic of user at all, therefore the inspection based on content
Hitch fruit is often unsatisfactory.
Application No. is 201410292820.2 Chinese patents, disclose a kind of shape table for trademark image retrieval
Show and matching process, by " calculating the minimum circumscribed circle of target object, mark region homalographic divides with one heart, concentric annular region
Bulk divides " and etc., realize the matching of trademark image retrieval.But the application limitation of this method is larger, for shape
Complexity, the shape and structure of especially several object compositions together, matching effect is not very good.
Summary of the invention
The present invention provides a kind of trademark image retrieval matching process, carries out matching to image local feature and similitude is arranged
Sequence, accuracy is high, and omission factor is low.
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics；
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all symbols in contrast images
The window for closing similar possibility, is calculated similarity distance；
S3:Erroneous matching is eliminated using the method for scale-space consistency；
S4:Go out similar area according to adaptive threshold fuzziness；
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Compare image.
Present invention segmented image region by the way of multiple dimensioned sliding window, combines the overlapping of traditional prior art
(overlap) and the advantages of non-overlap (non-overlap) two kinds of segmentation strategies, both it had been avoided that some important objects were divided into
Two blocks, calculation amount and not excessive, fast speed.
Characteristic window matches between the present invention first carries out global scale, and omission factor is extremely low.
The method that the present invention uses scale-space consistency increases compared to traditional simple consistency of scale method
Corresponding relationship, combines the feature in graphical rule and space while having added the corresponding relationship and scale-space in space,
The matching of mistake can be effectively eliminated, fallout ratio is reduced.
The present invention is partitioned into similar area using adaptive threshold method, improves the robust of trade mark search matching method
Property.
Preferably, a kind of scheme, step S1, the size and sliding step of sliding window are according to the fixation of image actual size
Ratio calculates, and the ratio that the size of specific sliding window accounts for image actual size can be from 0.01 to 1, and sliding step accounts for image reality
The ratio of size can be from 0.002 to 0.5.
Preferably, the size of a kind of scheme, step S1, sliding window is calculated according to the fixed proportion of image actual size, specifically
The ratio that the size of sliding window accounts for image actual size can be from 0.01 to 1；The step-length application variable-step self-adaptive of sliding window sliding is calculated
Method, specially the Runge-Kutta method of variable step.There is no the white space of feature, the change that step-length can be adaptive in image
Long, in the close quarters of characteristics of image, step-length adaptive can shorten.
Preferably, step S1, the quantity of sliding window are greater than 1, and the quantity of sliding window is more, images match it is more accurate, but
Be calculate data volume it is also bigger, the quantity of sliding window is preferably 3 to 8, more preferably 4 to 6.
Preferably, a kind of scheme, step S2 slide the mode of sliding window in image to be retrieved, for from a left side for image to be retrieved
Upper angle to the lower right corner successively from top to bottom, is from left to right slided, and a series of images feature is obtained.
Preferably, a kind of scheme, step S2 slide the mode of sliding window in image to be retrieved, for from image to be retrieved
The heart slides around.
Under normal circumstances, important and significant feature, it is bigger in the central part possibility of image to be retrieved, from
Center is slided around, and recall precision can improve.
Preferably, step S2, similarity distance are calculated by Hamming distance (Hamming Distance)；It is similar away from
FromWherein, feature binary string of the image to be retrieved after coding is fi, contrast images are by compiling
Feature binary string after code is gj, fi kIndicate binary string fiKth position, gj kIndicate binary string gjKth position,It indicates
The value of xor operation, α is equal to feature binary string fiWith gjThe inverse of length sum.
Preferably, step S2, meeting the condition that similar possibility needs meet is：(1) center of contrast images window
Near image sliding window window center position to be retrieved, permission transformation range is u for position, and the value range of u arrives for 0.4
0.6；(2) contrast images window and image sliding window window to be retrieved have similar length-width ratio, the ratio of described two length-width ratios
Range is 0.2 to 5, preferably 0.5 to 2.
Preferably, step S3 is retained on scale and spatial position using random sampling consistency (RANSAC) algorithm
On all it is consistent matching pair, the matching pair of debug.
Preferably, step S3, specific algorithm are：If a pair of of match window { (x of image to be retrieved and contrast images1,
y1),(x1′,y1′)}:{(x2,y2),(x2′,y2') (wherein, (x1,y1)、(x1′,y1') respectively indicate image window to be retrieved
The upper left corner and bottom right angular coordinate, (x2,y2)、(x2′,y2') respectively indicate contrast images window the upper left corner and the lower right corner sit
Mark), then there are space transform modelsSo that L can be solved,
Middle ɑ 1, ɑ 2 are the relevant zooming parameter of specific matching window, and tx, ty are translation parameters relevant to specific matching window；To sky
Between transformation model L use random sampling consistency (RANSAC) algorithm, be retained on scale on spatial position all have it is consistent
The matching pair of property, the matching pair of debug.
Preferably, correct match window is carried out quantitative weighted superposition, counts each similar window by step S4
Number, according to adaptive threshold Factorization algorithm similar area.
The number in more similar region, similar window is more.
Adaptivity threshold value obtains the size and number of similar window to the initial threshold matrix of each position and screening
It is related.
It is further preferred that the center of step S1 sliding window is defined as structure positioning point (anchor point),
Step S4, statistics cover the number of the similar window of each structure positioning point (anchor point).
More similar region, the number for covering the similar window of the regional structure anchor point (anchor point) are more.
It is further preferred that the weighted superposition of step S4, is the weight of each pair of match window by similarity distance dijIt determines,
Similarity distance is smaller, and the weight given is bigger, and similarity distance is bigger, and the weight given is smaller, and overall average weight is 1.
The meaning of weighting processing is that the higher match window of similarity degree has relatively high correct probability, to be greater than
1 weight is overlapped, and the lower match window of similarity degree has relatively low correct probability, is carried out with the weight less than 1
Superposition.In this way, obtained result has more accuracy.
Divide similar area and is based on a understanding：The scale of the similar window of similar area is wide, number is more.So according to
The ratio of each similar window defines threshold matrix to be split, and threshold matrix can be rule of thumb modified in actual treatment.
It is further preferred that setting T0For initial threshold matrix, the gross area of all similar windows is s, then adaptive threshold
Value matrix T=κ T0(s/(100))α, wherein κ, α are experience numerical constant, as the running parameter of sliding window specification should be fitted
The adjustment of answering property.
Preferably, step S5, the similitude of each contrast images similar area, passes through Hamming distance (Hamming
Distance it) is calculated and is indicated.
Beneficial effect：
1, present invention segmented image region by the way of multiple dimensioned sliding window, combines overlapping (overlap) and non-overlap
(non-overlap) the advantages of two kinds of segmentation strategies, both it had been avoided that some important objects were divided into two blocks, calculation amount is again
It is not excessive, fast speed.
2, characteristic window matches between the present invention first carries out global scale, and omission factor is extremely low.
3, the method that the present invention uses scale-space consistency, compared to traditional simple consistency of scale method,
Corresponding relationship while the corresponding relationship and scale-space in space is increased, the spy in graphical rule and space is combined
Sign can effectively eliminate the matching of mistake, reduce fallout ratio, and accuracy is high.
4, the present invention is partitioned into similar area using adaptive threshold method, improves the Shandong of trade mark search matching method
Stick.
5, the present invention goes out similar area according to adaptive threshold fuzziness, and by way of weighted superposition, search result is had more
There is accuracy.
Specific embodiment
The technical solution in the embodiment of the present invention will be clearly and completely described below, it is clear that described
Embodiment is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ability
Domain those of ordinary skill every other embodiment obtained without making creative work, belongs to the present invention
The range of protection.
Embodiment 1
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics；The size and sliding step of sliding window are according to the fixation ratio of image actual size
Example calculates, and the ratio that the size of specific sliding window accounts for image actual size can be from 0.01 to 1, and it is actually big that sliding step accounts for image
Small ratio can be from 0.002 to 0.5.The quantity of sliding window is 3. S2:Characteristic window matches between global scale, figure to be retrieved
Sliding window is slided as in, traverses all windows for meeting similar possibility in contrast images, similarity distance is calculated；It is to be retrieved
The mode of sliding window is slided in image, from the upper left corner of image to be retrieved to the lower right corner, successively from top to bottom, from left to right to slide
It is dynamic, obtain a series of images feature.
S3:Erroneous matching is eliminated using the method for scale-space consistency；Using random sampling consistency (RANSAC)
Algorithm, be retained on scale and on spatial position all it is consistent matching pair, the matching pair of debug.
S4:Go out similar area according to adaptive threshold fuzziness；Correct match window is subjected to quantitative weighted superposition,
The number for counting each similar window, according to adaptive threshold Factorization algorithm similar area；By the center of step S1 sliding window
It is defined as structure positioning point (anchor point), in step S4, statistics covers each structure positioning point (anchor point)
Similar window number；For each pair of match window weight by similarity distance dijIt determines, similarity distance is smaller, the power given
Again bigger, similarity distance is bigger, and the weight given is smaller, and overall average weight is 1.
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Compare image；The similitude of each contrast images similar area carries out calculating and table by Hamming distance (Hamming Distance)
Show.
Similar Window match in region, lookup method are local neighborhood lookup.
In the similar area of image to be retrieved, by any sliding of sliding window, institute in the similar area of contrast images is traversed
There is the sliding window window for meeting similar possible condition, similarity distance is calculated, the smallest similarity distance is most like window
Mouthful.
The similitude of sliding window is in image to be retrieved and contrast images similar area with the structure positioning point (anchor of sliding window
Point similitude) replaces, and similarity distance is by all windows centered on the structure positioning point (anchor point)
The mean value of similarity distance is corresponded to calculate.
The similarity distance d of image similar area to be retrieved and contrast images similar areaAB, specific algorithm is：Wherein, nAFor in image similar area to be retrieved include structure positioning point (anchor
Point number), nBFor the number in contrast images similar area including structure positioning point (anchor point), (u, v) is
The coordinate of structure positioning point (anchor point), dAUVFor the similar of image similar area structure positioning point (u, v) to be retrieved
Distance, dBUVFor the similarity distance of contrast images similar area structure positioning point (u, v), λ is similar area parameters and nA、nBAt
Inverse ratio, the similar area gross area is bigger, and λ is smaller.
The similarity distance of all contrast images, is retrieved according to similarity distance in image more to be retrieved and image library
As a result the sequence of Similar contrasts' image.Similarity distance is smaller, and image to be retrieved is more similar to contrast images, sorts more forward.
Embodiment 2
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics；The size of sliding window is calculated according to the fixed proportion of image actual size, specific sliding
The ratio that the size of window accounts for image actual size can be from 0.01 to 1；The step-length application variable-step self-adaptive of sliding window sliding is calculated
Method, specially the Runge-Kutta method of variable step.There is no the white space of feature, the change that step-length can be adaptive in image
Long, in the close quarters of characteristics of image, step-length adaptive can shorten.The quantity of sliding window is 8.
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all symbols in contrast images
The window for closing similar possibility, is calculated similarity distance；The mode that sliding window is slided in image to be retrieved is from figure to be retrieved
It slides around at the center of picture.
Similarity distance is calculated by Hamming distance (Hamming Distance)；Similarity distance
Wherein, feature binary string of the image to be retrieved after coding is fi, feature binary string of the contrast images after coding
For gj, fi kIndicate binary string fiKth position, gj kIndicate the kth position of binary string gj,Indicate xor operation, the value of α
Equal to feature binary string fiWith gjThe inverse of length sum.
Meeting the condition that similar possibility needs meet is：(1) center of contrast images window is in figure to be retrieved
Near picture sliding window window center position, permission transformation range is u, and the value range of u is 0.4 to 0.6；(2) contrast images window
Mouth has similar length-width ratio with image sliding window window to be retrieved, and the ratio range of described two length-width ratios is 0.2 to 5, preferably
It is 0.5 to 2.
S3:Erroneous matching is eliminated using the method for scale-space consistency.
S4:Go out similar area according to adaptive threshold fuzziness.
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Compare image.
Embodiment 3
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics；The quantity of sliding window is 6.
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all symbols in contrast images
The window for closing similar possibility, is calculated similarity distance.
S3:Erroneous matching is eliminated using the method for scale-space consistency；Using random sampling consistency (RANSAC)
Algorithm, be retained on scale and on spatial position all it is consistent matching pair, the matching pair of debug.
Specific algorithm is：If a pair of of match window { (x of image to be retrieved and contrast images1,y1),(x1′,y1′)}:
{(x2,y2),(x2′,y2') (wherein, (x1,y1)、(x1′,y1') respectively indicate the upper left corner and bottom right of image window to be retrieved
Angular coordinate, (x2,y2)、(x2′,y2') respectively indicate the upper left corner and the bottom right angular coordinate of contrast images window), then there is space
Transformation modelSo that L can be solved, wherein ɑ 1, ɑ 2 are specific
The relevant zooming parameter of match window, tx, ty are translation parameters relevant to specific matching window；Space transform models L is adopted
It with random sampling consistency (RANSAC) algorithm, is retained on scale and all consistent matching pair on spatial position, row
Except the matching pair of mistake.
S4:Go out similar area according to adaptive threshold fuzziness.
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Than image, the similitude of each contrast images similar area carries out calculating and table by Hamming distance (Hamming Distance)
Show.
Embodiment 4
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics.
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all symbols in contrast images
The window for closing similar possibility, is calculated similarity distance.
S3:Erroneous matching is eliminated using the method for scale-space consistency.
S4:Go out similar area according to adaptive threshold fuzziness；Correct match window is subjected to quantitative weighted superposition,
The number for counting each similar window, according to adaptive threshold Factorization algorithm similar area；By the center of step S1 sliding window
It is defined as structure positioning point (anchor point), in step S4, statistics covers each structure positioning point (anchor point)
Similar window number；For each pair of match window weight by similarity distance dijIt determines, similarity distance is smaller, the power given
Again bigger, similarity distance is bigger, and the weight given is smaller, and overall average weight is 1；If T0It is all similar for initial threshold matrix
The gross area of window is s, then adaptive threshold matrix T=κ T0(s/(100))α, wherein κ, α are experience numerical constant, with
The running parameter of sliding window specification should carry out the adjustment of adaptability.
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Compare image.
Embodiment 5
A kind of trademark image retrieval matching process, includes the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, the figure are extracted
As feature includes scale feature and space characteristics；
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverses all symbols in contrast images
The window for closing similar possibility, is calculated similarity distance；
S3:Erroneous matching is eliminated using the method for scale-space consistency；
S4:Go out similar area according to adaptive threshold fuzziness；
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports pair of search result
Compare image.
Finally it should be noted that：The foregoing is only a preferred embodiment of the present invention, is not limited to this hair
It is bright, although the present invention is described in detail referring to the foregoing embodiments, for those skilled in the art, according to
It is so possible to modify the technical solutions described in the foregoing embodiments, or part of technical characteristic is equal
Replacement, all within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should be included in this
Within the protection scope of invention.
Claims (10)
1. a kind of trademark image retrieval matching process, which is characterized in that include the following steps：
S1:Using the sliding window segmented image region of multiple and different scales, sliding window image in window feature, described image feature are extracted
Including scale feature and space characteristics；
S2:Characteristic window matches between global scale, and sliding window is slided in image to be retrieved, traverse in contrast images it is all meet it is similar
The window of possibility, is calculated similarity distance；
S3:Erroneous matching is eliminated using the method for scale-space consistency；
S4:Go out similar area according to adaptive threshold fuzziness；
S5:The sequence that search result is carried out according to the similitude of each contrast images similar area, exports the comparison diagram of search result
Picture.
2. trademark image retrieval matching process according to claim 1, which is characterized in that the step S1, sliding window it is big
Small and sliding step is calculated according to the fixed proportion of image actual size, and the size of specific sliding window accounts for the ratio of image actual size
For example from 0.01 to 1, the ratio that sliding step accounts for image actual size can be from 0.002 to 0.5.
3. trademark image retrieval matching process according to claim 1, which is characterized in that the step S1, sliding window it is big
The small fixed proportion according to image actual size calculates, and the size of specific sliding window accounts for the ratio of image actual size from 0.01 to 1；
The step-length application variable step size adaptive algorithm of sliding window sliding, specially the Runge-Kutta method of variable step.
4. trademark image retrieval matching process according to claim 2 or 3, which is characterized in that the step S2, it is to be retrieved
The mode of sliding window is slided in image, from the upper left corner of image to be retrieved to the lower right corner, successively from top to bottom, from left to right to slide
It is dynamic.
5. trademark image retrieval matching process according to claim 2 or 3, which is characterized in that the step S2, it is to be retrieved
The mode of sliding window is slided in image, to slide around from the center of image to be retrieved.
6. trademark image retrieval matching process according to any one of claims 1 to 5, which is characterized in that the step S2,
Similarity distance is calculated by Hamming distance (Hamming Distance)；Similarity distanceWherein,
Feature binary string of the image to be retrieved after coding is fi, feature binary string of the contrast images after coding is gj,
fi kIndicate binary string fiKth position, gj kIndicate binary string gjKth position,Indicate xor operation, the value of α is equal to spy
Levy binary string fiWith gjThe inverse of length sum.
7. trademark image retrieval matching process according to any one of claims 1 to 5, which is characterized in that the step S2,
Meeting the condition that similar possibility needs meet is：(1) center of contrast images window is in image sliding window window to be retrieved
Near mouth center, permission transformation range is u, and the value range of u is 0.4 to 0.6；(2) contrast images window with it is to be checked
Rope image sliding window window has similar length-width ratio, and the ratio range of described two length-width ratios is 0.2 to 5, preferably 0.5 to 2.
8. trademark image retrieval matching process according to any one of claims 1 to 5, which is characterized in that the step S3,
Using random sampling consistency (RANSAC) algorithm, be retained on scale and on spatial position all it is consistent matching pair,
The matching pair of debug；Specific algorithm is：If a pair of of match window { (x of image to be retrieved and contrast images1,y1),(x1′,
y1′)}:{(x2,y2),(x2′,y2') (wherein, (x1,y1)、(x1′,y1') respectively indicate image window to be retrieved the upper left corner and
Bottom right angular coordinate, (x2,y2)、(x2′,y2') respectively indicate the upper left corner and the bottom right angular coordinate of contrast images window), then there is sky
Between transformation modelSo thatL can be solved, wherein ɑ 1, ɑ 2 are spy
Determine the relevant zooming parameter of match window, tx, ty are translation parameters relevant to specific matching window；To space transform models L
Using random sampling consistency (RANSAC) algorithm, be retained on scale and on spatial position all it is consistent matching pair,
The matching pair of debug.
9. trademark image retrieval matching process according to any one of claims 1 to 8, which is characterized in that including following skill
Art feature at least one：
Correct match window is carried out quantitative weighted superposition, counts the number of each similar window by the step S4, according to
Adaptive threshold Factorization algorithm similar area；
The center of step S1 sliding window is defined as structure positioning point, in step S4, statistics covers each structure positioning point
The number of similar window；
The weighted superposition of the step S4 is the weight of each pair of match window by similarity distance dijIt determines, similarity distance is smaller, gives
With weight it is bigger, similarity distance is bigger, and the weight given is smaller, overall average weight be 1；The step S4, if T0It is initial
Threshold matrix, the gross area of all similar windows are s, then adaptive threshold matrix T=κ T0(s/(100))α, wherein κ, α are
Experience numerical constant, as the running parameter of sliding window specification should carry out the adjustment of adaptability.
10. trademark image retrieval matching process according to claim 9, which is characterized in that the step S1, the number of sliding window
Amount is 3 to 8；The step S5, the similitude of each contrast images similar area, passes through Hamming distance (Hamming
Distance it) is calculated and is indicated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810292274.0A CN108845998B (en) | 2018-04-03 | 2018-04-03 | Trademark image retrieval and matching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810292274.0A CN108845998B (en) | 2018-04-03 | 2018-04-03 | Trademark image retrieval and matching method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108845998A true CN108845998A (en) | 2018-11-20 |
| CN108845998B CN108845998B (en) | 2022-04-29 |
Family
ID=64211930
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810292274.0A Active CN108845998B (en) | 2018-04-03 | 2018-04-03 | Trademark image retrieval and matching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108845998B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934739A (en) * | 2019-02-27 | 2019-06-25 | 浪潮软件集团有限公司 | A kind of trademark infringement method for early warning and system |
| CN110084298A (en) * | 2019-04-23 | 2019-08-02 | 北京百度网讯科技有限公司 | Method and device for detection image similarity |
| CN110648339A (en) * | 2019-09-27 | 2020-01-03 | 广东溢达纺织有限公司 | Trademark cutting method and device, computer equipment and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101082987A (en) * | 2007-06-28 | 2007-12-05 | 复旦大学 | Column diagram comparability measurement method based on average difference between windows |
| US20080152225A1 (en) * | 2004-03-03 | 2008-06-26 | Nec Corporation | Image Similarity Calculation System, Image Search System, Image Similarity Calculation Method, and Image Similarity Calculation Program |
| CN101576956A (en) * | 2009-05-11 | 2009-11-11 | 天津普达软件技术有限公司 | On-line character detection method based on machine vision and system thereof |
| US20100290708A1 (en) * | 2009-05-12 | 2010-11-18 | Canon Kabushiki Kaisha | Image retrieval apparatus, control method for the same, and storage medium |
| US20130208997A1 (en) * | 2010-11-02 | 2013-08-15 | Zte Corporation | Method and Apparatus for Combining Panoramic Image |
| CN103383700A (en) * | 2013-06-26 | 2013-11-06 | 西安电子科技大学 | Image retrieval method based on margin directional error histogram |
| CN105139023A (en) * | 2015-07-24 | 2015-12-09 | 福州大学 | Seed identification method based on multi-scale feature fusion and extreme learning machine |
| CN105160654A (en) * | 2015-07-09 | 2015-12-16 | 浙江工商大学 | Towel label defect detecting method based on feature point extraction |
-
2018
- 2018-04-03 CN CN201810292274.0A patent/CN108845998B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080152225A1 (en) * | 2004-03-03 | 2008-06-26 | Nec Corporation | Image Similarity Calculation System, Image Search System, Image Similarity Calculation Method, and Image Similarity Calculation Program |
| CN101082987A (en) * | 2007-06-28 | 2007-12-05 | 复旦大学 | Column diagram comparability measurement method based on average difference between windows |
| CN101576956A (en) * | 2009-05-11 | 2009-11-11 | 天津普达软件技术有限公司 | On-line character detection method based on machine vision and system thereof |
| US20100290708A1 (en) * | 2009-05-12 | 2010-11-18 | Canon Kabushiki Kaisha | Image retrieval apparatus, control method for the same, and storage medium |
| US20130208997A1 (en) * | 2010-11-02 | 2013-08-15 | Zte Corporation | Method and Apparatus for Combining Panoramic Image |
| CN103383700A (en) * | 2013-06-26 | 2013-11-06 | 西安电子科技大学 | Image retrieval method based on margin directional error histogram |
| CN105160654A (en) * | 2015-07-09 | 2015-12-16 | 浙江工商大学 | Towel label defect detecting method based on feature point extraction |
| CN105139023A (en) * | 2015-07-24 | 2015-12-09 | 福州大学 | Seed identification method based on multi-scale feature fusion and extreme learning machine |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934739A (en) * | 2019-02-27 | 2019-06-25 | 浪潮软件集团有限公司 | A kind of trademark infringement method for early warning and system |
| CN110084298A (en) * | 2019-04-23 | 2019-08-02 | 北京百度网讯科技有限公司 | Method and device for detection image similarity |
| CN110648339A (en) * | 2019-09-27 | 2020-01-03 | 广东溢达纺织有限公司 | Trademark cutting method and device, computer equipment and storage medium |
| CN110648339B (en) * | 2019-09-27 | 2022-08-19 | 广东溢达纺织有限公司 | Trademark cutting method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108845998B (en) | 2022-04-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108549891B (en) | Multi-scale diffusion well-marked target detection method based on background Yu target priori | |
| CN107808143B (en) | Dynamic gesture recognition method based on computer vision | |
| CN111104898B (en) | Image scene classification method and device based on target semantics and attention mechanism | |
| Yin et al. | Large scale remote sensing image segmentation based on fuzzy region competition and Gaussian mixture model | |
| CN111652321B (en) | Marine ship detection method based on improved YOLOV3 algorithm | |
| CN106296695B (en) | Adaptive threshold natural target image segmentation extraction algorithm based on conspicuousness | |
| CN109344701A (en) | A kind of dynamic gesture identification method based on Kinect | |
| Lin et al. | Ship detection with superpixel-level Fisher vector in high-resolution SAR images | |
| CN111191622A (en) | Posture recognition method and system based on thermodynamic diagram and offset vector and storage medium | |
| CN110032925B (en) | Gesture image segmentation and recognition method based on improved capsule network and algorithm | |
| CN110276780A (en) | A kind of multi-object tracking method, device, electronic equipment and storage medium | |
| Pons-Moll et al. | Metric regression forests for correspondence estimation | |
| CN112784736B (en) | Character interaction behavior recognition method based on multi-modal feature fusion | |
| CN108845998A (en) | A kind of trademark image retrieval matching process | |
| CN107133496B (en) | Gene feature extraction method based on manifold learning and closed-loop deep convolution double-network model | |
| CN108388874A (en) | Prawn morphological parameters method for automatic measurement based on image recognition and cascade classifier | |
| CN108664838A (en) | Based on the monitoring scene pedestrian detection method end to end for improving RPN depth networks | |
| CN109598234A (en) | Critical point detection method and apparatus | |
| CN110413816A (en) | Colored sketches picture search | |
| CN109886128A (en) | A kind of method for detecting human face under low resolution | |
| CN109886153A (en) | A kind of real-time face detection method based on depth convolutional neural networks | |
| CN110008844A (en) | A kind of long-term gesture tracking method of KCF merging SLIC algorithm | |
| CN107292896A (en) | Contour extraction method based on Snake models | |
| CN107169962A (en) | The gray level image fast partition method of Kernel fuzzy clustering is constrained based on space density | |
| CN108629762A (en) | A kind of stone age evaluation and test model reduces the image pre-processing method and system of interference characteristic |
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 |