CN108845997A - A kind of trade mark searching system and method - Google Patents
A kind of trade mark searching system and method Download PDFInfo
- Publication number
- CN108845997A CN108845997A CN201810291407.2A CN201810291407A CN108845997A CN 108845997 A CN108845997 A CN 108845997A CN 201810291407 A CN201810291407 A CN 201810291407A CN 108845997 A CN108845997 A CN 108845997A
- Authority
- CN
- China
- Prior art keywords
- image
- unit
- trade mark
- memory
- pixel
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 123
- 230000015654 memory Effects 0.000 claims abstract description 77
- 238000000605 extraction Methods 0.000 claims abstract description 33
- 238000001914 filtration Methods 0.000 claims description 42
- 238000010606 normalization Methods 0.000 claims description 37
- 230000002708 enhancing effect Effects 0.000 claims description 34
- 238000012937 correction Methods 0.000 claims description 28
- 238000007781 pre-processing Methods 0.000 claims description 16
- 230000002452 interceptive effect Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims 1
- 238000003708 edge detection Methods 0.000 description 11
- 238000010276 construction Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
- 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
- G06T7/11—Region-based segmentation
-
- 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)
- General Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of trade mark searching system and methods, including WEB server, search condition analytical unit, image matching unit and isomeric data memory.Trade mark searching system of the invention carries out pretreatment and feature extraction for trade mark to be retrieved, is matched with the mass image data in internet, is not limited to the trade mark library of standard, range of search is more comprehensive.Since trade mark to be retrieved has carried out pretreatment and feature extraction, matching characteristic point is more accurate, and identification error is small, substantially increases the accuracy and efficiency of retrieval.
Description
Technical field
The present invention relates to a kind of searching systems, and in particular to a kind of trade mark searching system and method.Belong to picture search skill
Art field.
Background technique
Trade mark is the mark of company, product or service, and melting with the commercial quality of enterprise, service quality, management is one
Body plays very important effect in industry and commerce society, is an important attribute of company and product.In order to enable trade mark by
To legal protection, generally require to trademark office's official register.With the quickening of China's expanding economy and globalization process, trade mark
Quantity cumulative year after year.Preventing repeated registration or similar brand from registering is the key problem of trade-mark administration.In order to protect registered trademark
Legitimate rights and interests, hit counterfeit illegal activities for usurping registered trademark, need to retrieve trade mark to be registered, and it is registered
Trade mark be compared, determine both it is not identical or not similar, just have registration and qualification.
Trademark image demand is continuously increased in recent years, and traditional trade mark retrieval is typically based on classification code and with a large amount of people
Power is cost, and retrieval accuracy and efficiency are all extremely low, does not adapt to the requirement of current a large amount of trade mark registrations increasingly.
If carrying out manual retrieval to trade mark needs to expend a large amount of manpowers, the mass data of relative interconnections net, speed right and wrong
It is often slow.And current Internal retrieval, such as Baidu, google etc., or based on keyword.Even existing some bases
In the trade-mark searching method of image, same and similar trade mark is also mainly searched in application in trade mark library.Figure in trade mark library
Seem by standardized, only includes trade mark, without other backgrounds.And trademark image is clear, proper.And quotient is searched in internet
In the case where mark, trade mark reappears in the picture, often generates scale, rotation angle, illumination, the variation at visual angle, shape.
The Internet images are generally compressed more for the ease of transmission, and picture quality is not also high.This is method used in inquiry trade mark library
It is insurmountable.
Summary of the invention
The purpose of the present invention is to overcome above-mentioned the deficiencies in the prior art, provide a kind of trade mark searching system and method.
To achieve the above object, the present invention adopts the following technical solutions：
A kind of trade mark searching system, including：
WEB server is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit, the figure of characteristics of image and image in isomeric data memory for extracting trade mark to be retrieved
As feature is matched, and matching result is fed back into WEB server；
Isomeric data memory, for storing image data.
One of as a preferred technical scheme, the search condition analytical unit includes sequentially connected image preprocessing mould
Block, the first arithmetic device for executing feature extraction algorithm and the first memory for storing first arithmetic device operation result.
As further preferred one of technical solution, described image preprocessing module include Image geometry transform unit,
Image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Described image geometrical transformation unit determines the gray value of each pixel of correction space using interpolation method three times；
Described image denoises unit using nonlinear filtering method removal background noise and reduces in image transmitting process
The noise of doping；
Described image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Described image enhancement unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Described image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
One of as a preferred technical scheme, described image matching unit include execute the second arithmetic device of matching algorithm with
And the second memory for storing second arithmetic device operation result, second arithmetic device also respectively with first memory and isomery number
It is connected according to memory.
As further preferred one of technical solution, the second memory is connect with WEB server.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory, it will
Matching result feeds back to WEB server and shows.
One of as a preferred technical scheme, it is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
One of as a preferred technical scheme, step S2 extract be characterized in not being illuminated by the light, color, scale and rotationally-varying
The invariant feature of influence, specific extracting method are：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
One of as a preferred technical scheme, feature extraction includes color characteristic and edge shape feature in step S2.
As further preferred one of technical solution, color characteristic is for describing scape corresponding to image or image-region
The surface nature of object, extracting method include color histogram, color set, color moment, color convergence vector, color correlogram etc.
Deng.
As further preferred one of technical solution, edge shape feature refers to that its surrounding pixel gray scale is jumpy
The set of those pixels, edge are present between target, background and region, are the most basic features of image, extracting method can
It is realized using any one of following edge detection algorithm：Sobel operator edge detection, Roberts operator edge detection,
Prewitt operator edge detection, Laplacian operator edge detection and Canny operator edge detection.
One of as a preferred technical scheme, the characteristics of image extracted in step S2 is stored to first memory.
One of as a preferred technical scheme, the matching result in step S3 is stored to second memory, second memory
It is connect with WEB server.
Beneficial effects of the present invention：
1, trade mark searching system of the invention carries out in pretreatment and feature extraction, with internet for trade mark to be retrieved
Mass image data is matched, and the trade mark library of standard is not limited to, and range of search is more comprehensive.Since trade mark to be retrieved carries out
Pretreatment and feature extraction, matching characteristic point is more accurate, and identification error is small, substantially increase the accuracy and effect of retrieval
Rate.
2, the extracted feature of trade mark to be retrieved separately stores, when carrying out matching operation, can directly transfer feature with it is different
The characteristics of image of image is matched in structure data storage, improves retrieval accuracy, and matching result individually stores, for WEB
Server calls avoid repeatability retrieval work, improve recall precision.
3, the pretreatment and feature extraction of trade mark to be retrieved are very crucial, and image pre-processing module of the invention includes image
Geometrical transformation unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit, realize quotient
The denoising and normalization of logo image are conducive to the accuracy for improving retrieval.
4, the present invention extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, match pair
As not by image irradiation, color, scale and it is rotationally-varying influenced, search result is more comprehensive.
Detailed description of the invention
Fig. 1 is system structure diagram of the invention；
Wherein, 1 is WEB server, and 2 be search condition analytical unit, and 21 be image pre-processing module, and 22 be the first operation
Device, 23 be first memory, and 3 be image matching unit, and 31 be second arithmetic device, and 32 be second memory, and 4 deposit for isomeric data
Reservoir.
Specific embodiment
The present invention will be further elaborated with reference to the accompanying drawings and examples, it should which explanation, following the description is only
It is not to be defined to its content to explain the present invention.
Embodiment 1：
A kind of trade mark searching system as shown in Figure 1, including：
WEB server 1 is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit 2 pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit 3, the characteristics of image for extracting trade mark to be retrieved and image in isomeric data memory
Characteristics of image is matched, and matching result is fed back to WEB server 1；
Isomeric data memory 4, for storing image data.
Search condition analytical unit 2 includes sequentially connected image pre-processing module 21, executes the of feature extraction algorithm
One arithmetic unit 22 and first memory 23 for storing 22 operation result of first arithmetic device.Image pre-processing module 21 includes
Image geometry transform unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Image matching unit 3 includes executing the second arithmetic device 31 of matching algorithm and transporting for storing second arithmetic device 31
The second memory 32 of result is calculated, second arithmetic device 31 is also connect with first memory 23 and isomeric data memory 4 respectively.The
Two memories 32 are connect with WEB server 1.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server 1；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory 4, it will
Matching result feeds back to WEB server 1 and shows.
It is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Step S2 extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, it is specific to extract
Method is：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
Feature extraction includes color characteristic and edge shape feature in step S2.Color characteristic is for describing image or image
The surface nature of scenery corresponding to region, extracting method are color histogram.Edge shape feature refers to its surrounding pixel
The set of those of gray scale change dramatically pixel, edge are present between target, background and region, are the most basic spies of image
The realization of Sobel operator edge detection can be used in sign, extracting method.The characteristics of image extracted in step S2 is stored to the first storage
Device 23.
Matching result in step S3 is stored to second memory 32, and second memory 32 is connect with WEB server 1.
Embodiment 2：
A kind of trade mark searching system as shown in Figure 1, including：
WEB server 1 is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit 2 pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit 3, the characteristics of image for extracting trade mark to be retrieved and image in isomeric data memory
Characteristics of image is matched, and matching result is fed back to WEB server 1；
Isomeric data memory 4, for storing image data.
Search condition analytical unit 2 includes sequentially connected image pre-processing module 21, executes the of feature extraction algorithm
One arithmetic unit 22 and first memory 23 for storing 22 operation result of first arithmetic device.Image pre-processing module 21 includes
Image geometry transform unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Image matching unit 3 includes executing the second arithmetic device 31 of matching algorithm and transporting for storing second arithmetic device 31
The second memory 32 of result is calculated, second arithmetic device 31 is also connect with first memory 23 and isomeric data memory 4 respectively.The
Two memories 32 are connect with WEB server 1.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server 1；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory 4, it will
Matching result feeds back to WEB server 1 and shows.
It is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Step S2 extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, it is specific to extract
Method is：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
Feature extraction includes color characteristic and edge shape feature in step S2.Color characteristic is for describing image or image
The surface nature of scenery corresponding to region, extracting method are color set.Edge shape feature refers to its surrounding pixel gray scale
The set of those of change dramatically pixel, edge are present between target, background and region, are the most basic features of image,
The realization of Roberts operator edge detection can be used in extracting method.The characteristics of image extracted in step S2 is stored to first memory
23。
Matching result in step S3 is stored to second memory 32, and second memory 32 is connect with WEB server 1.
Embodiment 3：
A kind of trade mark searching system as shown in Figure 1, including：
WEB server 1 is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit 2 pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit 3, the characteristics of image for extracting trade mark to be retrieved and image in isomeric data memory
Characteristics of image is matched, and matching result is fed back to WEB server 1；
Isomeric data memory 4, for storing image data.
Search condition analytical unit 2 includes sequentially connected image pre-processing module 21, executes the of feature extraction algorithm
One arithmetic unit 22 and first memory 23 for storing 22 operation result of first arithmetic device.Image pre-processing module 21 includes
Image geometry transform unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Image matching unit 3 includes executing the second arithmetic device 31 of matching algorithm and transporting for storing second arithmetic device 31
The second memory 32 of result is calculated, second arithmetic device 31 is also connect with first memory 23 and isomeric data memory 4 respectively.The
Two memories 32 are connect with WEB server 1.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server 1；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory 4, it will
Matching result feeds back to WEB server 1 and shows.
It is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Step S2 extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, it is specific to extract
Method is：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
Feature extraction includes color characteristic and edge shape feature in step S2.Color characteristic is for describing image or image
The surface nature of scenery corresponding to region, extracting method are color moment.Edge shape feature refers to its surrounding pixel gray scale
The set of those of change dramatically pixel, edge are present between target, background and region, are the most basic features of image,
The realization of Prewitt operator edge detection can be used in extracting method.The characteristics of image extracted in step S2 is stored to first memory
23。
Matching result in step S3 is stored to second memory 32, and second memory 32 is connect with WEB server 1.
Embodiment 4：
A kind of trade mark searching system as shown in Figure 1, including：
WEB server 1 is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit 2 pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit 3, the characteristics of image for extracting trade mark to be retrieved and image in isomeric data memory
Characteristics of image is matched, and matching result is fed back to WEB server 1；
Isomeric data memory 4, for storing image data.
Search condition analytical unit 2 includes sequentially connected image pre-processing module 21, executes the of feature extraction algorithm
One arithmetic unit 22 and first memory 23 for storing 22 operation result of first arithmetic device.Image pre-processing module 21 includes
Image geometry transform unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Image matching unit 3 includes executing the second arithmetic device 31 of matching algorithm and transporting for storing second arithmetic device 31
The second memory 32 of result is calculated, second arithmetic device 31 is also connect with first memory 23 and isomeric data memory 4 respectively.The
Two memories 32 are connect with WEB server 1.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server 1；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory 4, it will
Matching result feeds back to WEB server 1 and shows.
It is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Step S2 extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, it is specific to extract
Method is：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
Feature extraction includes color characteristic and edge shape feature in step S2.Color characteristic is for describing image or image
The surface nature of scenery corresponding to region, extracting method are color convergence vector.Edge shape feature refers to picture around it
The set of those of plain gray scale change dramatically pixel, edge are present between target, background and region, are the most basic spies of image
The realization of Laplacian operator edge detection can be used in sign, extracting method.The characteristics of image extracted in step S2 is stored to first
Memory 23.
Matching result in step S3 is stored to second memory 32, and second memory 32 is connect with WEB server 1.
Embodiment 5：
A kind of trade mark searching system as shown in Figure 1, including：
WEB server 1 is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit 2 pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
Image matching unit 3, the characteristics of image for extracting trade mark to be retrieved and image in isomeric data memory
Characteristics of image is matched, and matching result is fed back to WEB server 1；
Isomeric data memory 4, for storing image data.
Search condition analytical unit 2 includes sequentially connected image pre-processing module 21, executes the of feature extraction algorithm
One arithmetic unit 22 and first memory 23 for storing 22 operation result of first arithmetic device.Image pre-processing module 21 includes
Image geometry transform unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Image matching unit 3 includes executing the second arithmetic device 31 of matching algorithm and transporting for storing second arithmetic device 31
The second memory 32 of result is calculated, second arithmetic device 31 is also connect with first memory 23 and isomeric data memory 4 respectively.The
Two memories 32 are connect with WEB server 1.
A kind of corresponding trade-mark searching method of above system, including step：
S1. trade mark to be retrieved is uploaded to WEB server 1；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory 4, it will
Matching result feeds back to WEB server 1 and shows.
It is pretreated in step S2 that the specific method is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
Step S2 extract be characterized in not being illuminated by the light, the invariant feature of color, scale and rotationally-varying influence, it is specific to extract
Method is：
S2-1. it is made of multiple frequency ranges, successive bands ruler using sampling and Gaussian convolution structural map as gaussian pyramid
Degree differs 50%, utilizes the multiple sublayers of Gaussian convolution construction in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature on scale space in neighborhood
The processing result value of detective operators, if the end value on the pixel is maximum value or minimum value in its neighborhood, just by it
As candidate feature point, the frequency range that it occurs, sublayer serial number, the coordinate information in image are recorded；
S2-4. duplicate point in candidate feature point is removed, weak contrast and adjacent edges in candidate feature point are then removed
Point, obtain invariant feature point.
Feature extraction includes color characteristic and edge shape feature in step S2.Color characteristic is for describing image or image
The surface nature of scenery corresponding to region, extracting method are color correlogram.Edge shape feature refers to its surrounding pixel
The set of those of gray scale change dramatically pixel, edge are present between target, background and region, are the most basic spies of image
The realization of Canny operator edge detection can be used in sign, extracting method.The characteristics of image extracted in step S2 is stored to the first storage
Device 23.
Matching result in step S3 is stored to second memory 32, and second memory 32 is connect with WEB server 1.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, based on the technical solutions of the present invention, those skilled in the art are not needed to make the creative labor and can be done
Various modifications or changes out are still within protection scope of the present invention.
Claims (10)
1. a kind of trade mark searching system, which is characterized in that including：
WEB server is equipped with man-machine interactive interface, for uploading trade mark to be retrieved, receiving search result and showing；
Search condition analytical unit pre-processes for trade mark to be retrieved, and carries out feature extraction and storage；
The image of image matching unit, characteristics of image and image in isomeric data memory for extracting trade mark to be retrieved is special
Sign is matched, and matching result is fed back to WEB server；
Isomeric data memory, for storing image data.
2. a kind of trade mark searching system according to claim 1, which is characterized in that the search condition analytical unit includes
Sequentially connected image pre-processing module, execute feature extraction algorithm first arithmetic device and for store first arithmetic device fortune
Calculate the first memory of result.
3. a kind of trade mark searching system according to claim 2, which is characterized in that described image preprocessing module includes figure
As geometrical transformation unit, image denoising unit, image restoration unit, image enhancing unit and image normalization unit；
Described image geometrical transformation unit determines the gray value of each pixel of correction space using interpolation method three times；
Described image denoising unit is removed background noise and is reduced in image transmitting process and adulterated using nonlinear filtering method
Noise；
Described image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Described image enhancement unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Described image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
4. a kind of trade mark searching system according to claim 1, which is characterized in that described image matching unit includes executing
The second arithmetic device of matching algorithm and second memory for storing second arithmetic device operation result, second arithmetic device also divide
It is not connect with first memory and isomeric data memory.
5. a kind of trade mark searching system according to claim 4, which is characterized in that the second memory and WEB service
Device connection.
6. a kind of corresponding trade-mark searching method of system described in any one of Claims 1 to 5, which is characterized in that including step：
S1. trade mark to be retrieved is uploaded to WEB server；
S2. trade mark to be retrieved is pre-processed, feature extraction and storage；
S3. the characteristics of image that step S2 is extracted is matched with the characteristics of image of image in isomeric data memory, will be matched
As a result it feeds back to WEB server and shows.
7. a kind of trade-mark searching method according to claim 6, which is characterized in that pretreated specific method in step S2
It is as follows：
Image geometry transform unit determines the gray value of each pixel of correction space using interpolation method three times；
Image denoising unit is made an uproar using what is adulterated in nonlinear filtering method removal background noise and reduction image transmitting process
Sound；
Image restoration unit is degenerated using image caused by Wiener filtering restoring method correction a variety of causes；
Image enhancing unit is selectively reinforced and is inhibited to the information in image using Gabor filtering enhancing method；
Image normalization unit obtains the image with invariance using the normalization algorithm based on image pixel.
8. a kind of trade-mark searching method according to claim 6, which is characterized in that step S2, which is extracted, is characterized in not light
According to, the invariant feature of color, scale and rotationally-varying influence, specific extracting method is：
S2-1. it is made of multiple frequency ranges, successive bands scale phase using sampling and Gaussian convolution structural map as gaussian pyramid
Poor 50%, multiple sublayers are constructed using Gaussian convolution in each frequency range；
S2-2. each tomographic image is handled using various features detective operators；
S2-3. in each frequency range, to each pixel in each sublayer, compare the feature detection on scale space in neighborhood
The processing result value of operator, if the end value on the pixel is maximum value or minimum value in its neighborhood, just as
Candidate feature point records the frequency range that it occurs, sublayer serial number, the coordinate information in image；
S2-4. duplicate point in candidate feature point is removed, the point of weak contrast and adjacent edges in candidate feature point are then removed,
Obtain invariant feature point.
9. a kind of trade-mark searching method according to claim 6, which is characterized in that feature extraction includes color in step S2
Feature and edge shape feature.
10. a kind of trade-mark searching method according to claim 6, which is characterized in that the matching result storage in step S3
To second memory, second memory is connect with WEB server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810291407.2A CN108845997A (en) | 2018-04-03 | 2018-04-03 | A kind of trade mark searching system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810291407.2A CN108845997A (en) | 2018-04-03 | 2018-04-03 | A kind of trade mark searching system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108845997A true CN108845997A (en) | 2018-11-20 |
Family
ID=64211892
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810291407.2A Withdrawn CN108845997A (en) | 2018-04-03 | 2018-04-03 | A kind of trade mark searching system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108845997A (en) |
-
2018
- 2018-04-03 CN CN201810291407.2A patent/CN108845997A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106981077B (en) | Infrared image and visible light image registration method based on DCE and LSS | |
| CN107688806B (en) | Affine transformation-based free scene text detection method | |
| CN107103314B (en) | A kind of fake license plate vehicle retrieval system based on machine vision | |
| Fantoni et al. | Accurate and automatic alignment of range surfaces | |
| US20060126943A1 (en) | Geometric hashing method for model-based recognition of an object | |
| CN101593205B (en) | Video-based method for searching three-dimension model | |
| CN108664556A (en) | Parallel search system and method suitable for associated mark | |
| CN103617420A (en) | Commodity fast recognition method and system based on image feature matching | |
| CN104991954A (en) | Shopping searching system based on image search and realization method thereof | |
| CN112396643A (en) | Multi-mode high-resolution image registration method with scale-invariant features and geometric features fused | |
| Wang et al. | Static gesture segmentation technique based on improved Sobel operator | |
| CN105095446A (en) | Medicine search processing method, server and terminal device | |
| CN108491542A (en) | A kind of trade mark searching system and method based on database building | |
| Li et al. | RIFT2: Speeding-up RIFT with a new rotation-invariance technique | |
| Yang et al. | A research of feature-based image mosaic algorithm | |
| CN108470073A (en) | Searching system and method suitable for associated mark | |
| Nawaz et al. | Image authenticity detection using DWT and circular block-based LTrP features | |
| Huang et al. | Robust simultaneous localization and mapping in low‐light environment | |
| Lengauer et al. | A benchmark dataset for repetitive pattern recognition on textured 3d surfaces | |
| CN106651756B (en) | Image registration method based on SIFT and verification mechanism | |
| CN108416068A (en) | A kind of trade mark parallel search system and method | |
| EP4089644A1 (en) | Image matching system | |
| CN108845997A (en) | A kind of trade mark searching system and method | |
| Wang et al. | An Improved Canny‐Zernike Subpixel Detection Algorithm | |
| CN108446405A (en) | A kind of trade mark parallel search system and method based on database building |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20181120 |
|
| WW01 | Invention patent application withdrawn after publication |