WO2017101225A1 - 一种商标图形要素识别方法、装置、系统以及计算机存储介质 - Google Patents

一种商标图形要素识别方法、装置、系统以及计算机存储介质 Download PDF

Info

Publication number
WO2017101225A1
WO2017101225A1 PCT/CN2016/075871 CN2016075871W WO2017101225A1 WO 2017101225 A1 WO2017101225 A1 WO 2017101225A1 CN 2016075871 W CN2016075871 W CN 2016075871W WO 2017101225 A1 WO2017101225 A1 WO 2017101225A1
Authority
WO
WIPO (PCT)
Prior art keywords
trademark
image
feature information
sample
line
Prior art date
Application number
PCT/CN2016/075871
Other languages
English (en)
French (fr)
Inventor
徐庆
Original Assignee
徐庆
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 徐庆 filed Critical 徐庆
Priority to GB1811592.3A priority Critical patent/GB2561773A/en
Priority to US16/062,001 priority patent/US10430687B2/en
Publication of WO2017101225A1 publication Critical patent/WO2017101225A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/51Indexing; Data structures therefor; Storage structures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/22Matching criteria, e.g. proximity measures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75Organisation 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
    • G06V10/752Contour matching

Definitions

  • the invention relates to the field of trademark information retrieval, and in particular relates to a trademark graphic element identification method, device, system and computer storage medium.
  • Trademark search is an important task in the procedures of registration application, trademark review, trademark management, and trademark rights protection.
  • the traditional graphic trademark search basically achieves the purpose of retrieval by manually inputting the trademark graphic element code as a retrieval condition.
  • the trademark graphic element code is a trademark graphic element division tool produced according to the Vienna Agreement for the Establishment of International Classification of Graphical Elements of Marks. It consists of a list of trademark graphic elements classified by major categories, subcategories and groups, including trademark graphic element numbers. And the name of the trademark graphic element. Therefore, each trademark graphic element code represents the content meaning of the trademark graphic element.
  • trademark graphic element coding is mainly divided by a small number of examiners who have the professional level of trademark graphic element coding in the national trademark authority, and there is basically no intelligent tool or means.
  • the manual division method of the existing trademark graphic element coding can perform the division task of the trademark graphic element coding, it has obvious defects and drawbacks, which are mainly reflected in:
  • the present invention provides a method, a device and a system for identifying a trademark graphic element, and a computer storage medium, which utilizes a known big data resource analysis of a coded graphic element of a registered graphic trademark, and passes the image of the trademark to be identified. Matching and searching the feature information and the sample trademark image feature information to obtain the sample trademark with the highest degree of coincidence of the feature information and the recorded graphic element code, and coding the graphic element of the sample trademark as the graphic element code of the trademark to be identified, and realizing the trademark graphic Automated, standardized identification of feature coding.
  • the present invention provides a method for identifying a graphic graphic element, comprising: S101, establishing a sample trademark library and establishing a correspondence relationship between a sample trademark and a known coded image of the registered graphic graphic graphic element; S102, for the sample trademark Image feature information is extracted and processed, and a correspondence relationship between the sample trademark and the extracted image feature information is established; S103, image feature information of the trademark to be identified is extracted and processed; S104, image feature information of the trademark to be identified is used as a search condition Perform matching search to find the sample trademark with the highest degree of similarity to the feature image of the trademark to be identified and the corresponding trademark graphic element code; S105, encode the graphic graphic element corresponding to the sample trademark as the graphic element code of the trademark to be identified for output .
  • the present invention also provides a trademark graphic element identification device, comprising: a sample trademark library creation module, configured to establish a sample trademark library and establish a correspondence between a sample trademark and a known coded division data of a registered graphic trademark graphic element. Relationship; sample trademark map
  • the feature information extraction module is configured to extract and process the image feature information of the sample trademark, and establish a correspondence relationship between the sample trademark and the extracted image feature information; the feature image extraction module to be identified is used to identify the trademark
  • the image feature information is extracted and processed;
  • the matching retrieval module is configured to perform matching search by using the image feature information of the trademark to be identified as a search condition, and find the sample trademark and the corresponding trademark graphic element with the highest degree of similarity to the feature information of the trademark image to be identified.
  • the graphic element encoding output module is configured to output the graphic graphic element code corresponding to the sample trademark as a graphic element code of the trademark to be identified.
  • the present invention also provides a trademark graphic element identification system including a memory and a server configured to perform the following operations: creating a sample trademark library and establishing a sample trademark and a known applied-registered graphic trademark graphic Corresponding relationship between the element code division data; extracting and processing the image feature information of the sample trademark, and establishing the correspondence relationship between the sample trademark and the extracted image feature information; extracting and processing the image feature information of the identified trademark;
  • the image feature information of the trademark is matched and searched for the search condition, and the sample trademark with the highest degree of similarity with the feature information of the trademark image to be identified and the corresponding trademark graphic element code are found; the trademark graphic element corresponding to the sample trademark is encoded as the trademark to be identified.
  • the graphic element code is output.
  • the present invention also provides a storage medium comprising a computer readable program that performs the above-described method of identifying a trademark graphic element when the computer readable program in the storage medium is executed.
  • the embodiment of the invention realizes the big data resource divided by the existing known registered graphic trademark graphic element coding, and learns and automatically divides the trademark graphic element coding by the system, thereby solving the standardized identification of the coding and division of the trademark graphic element and the past.
  • the embodiment of the invention Compared with the traditional manual division of trademark graphic element coding, the embodiment of the invention has the advantages of high work efficiency and labor saving, and solves the problem that the traditional trademark graphic element coding is only manually divided, and the work efficiency is low and the work energy consumption is huge. The drawbacks.
  • the embodiment of the invention reduces the requirement of the professional level of the trademark graphic element coding of the trademark searcher, and is beneficial to the trademark retrieval technology of the trademark graphic element coding in a wider range, so as to overcome the traditional trademark graphic element coding only by manual division.
  • the technical defects or limitations of the classification criteria are difficult to be unified and subjectively divided, and the purpose of automatic identification and standardization identification of trademark graphic element coding can be effectively realized.
  • FIG. 1 is a flow chart of a method for identifying a trademark graphic element according to an embodiment of the present invention.
  • FIG. 2 is an exemplary image original view of an embodiment of the present invention.
  • FIG 3 is an exemplary contoured feature line diagram of an embodiment of the present invention.
  • FIG. 4 is a directional positioning diagram of natural reference positioning according to an embodiment of the present invention.
  • FIG. 5 is an extracted square positioning map of natural reference positioning according to an embodiment of the present invention.
  • Fig. 6 is a directional positioning diagram of a reference straight line positioning according to an embodiment of the present invention.
  • Fig. 7 is a directional positioning diagram of a reference straight line positioning according to another embodiment of the present invention.
  • FIG. 8 is an extracted square positioning diagram of a reference straight line positioning according to an embodiment of the present invention.
  • Figure 9 is an exemplary image original view of another embodiment of the present invention.
  • Figure 10 is an exemplary outline feature line diagram of another embodiment of the present invention.
  • FIG. 1 is a flow chart showing a method for identifying a trademark graphic element according to an embodiment of the present invention, comprising the following steps:
  • S104 performing matching search by using image feature information of the trademark to be identified as a search condition, and finding a sample trademark with the highest degree of similarity with the feature information of the trademark image to be identified and a corresponding trademark graphic element code;
  • FIG. 2 randomly gives an image of a plurality of sample trademarks.
  • the trademark graphic element coding refers to a trademark graphic element division tool produced according to the Vienna Agreement for the Establishment of International Classification of Graphical Elements of Marks, which consists of a list of trademark graphic elements classified by categories, sub-categories and groups, including trademark graphics. The element number and the trademark graphic element name are composed.
  • the sample mark image feature is identified and the image feature information is extracted.
  • the purpose of identifying and processing image feature image features is to find the same or most similar trademark by matching the image feature information.
  • the technical scheme adopts the Teh-Chin detection algorithm in the polygon approximation method to extract the key pixel feature information of the sample trademark image. , that is, the coordinate value of the pixel on the contour feature of the sample trademark image.
  • the key pixel feature information extraction of the sample trademark image may also adopt other known technical methods, including extracting skeleton feature information, extracting shape feature information, extracting template feature information, and the like.
  • the process of identifying and processing image feature image features is as follows:
  • a grayscale digital image is an image with only one sample color per pixel. Such images are typically displayed as grayscale from the darkest black to the brightest white, although in theory this sample can be any shade of any color, or even a different color on different brightness. Grayscale images are different from black and white images. In the computer image field, black and white images are only black and white. The grayscale image has many levels of color depth between black and white.
  • Image binarization is to set the gray value of the pixel on the image to 0 or 255, which is to show the whole image a distinct black and white effect.
  • the noise mainly refers to the rough part in the image generated by the charge-coupled element CCD (CMOS) as the received signal and output, and also refers to the external pixels that should not appear in the image, usually caused by electronic interference. It looks like the image is dirty and covered with some tiny rough spots. The digital photos we usually take may not be noticed if the high-quality images captured by the personal computer are reduced. However, if the original image is enlarged, then there will be a color (false color) that is not present. This false color is image noise, which can be removed by technical means.
  • CMOS charge-coupled element CCD
  • the key pixel feature of the image is extracted by the polygon approximation method, that is, the contour feature line, and the contour feature line is a set of pixel points on the contour line, and the pixel point sequence on the contour feature line of the sample trademark image can be generated.
  • Figure 3 shows the contour feature lines of several images, as can be seen, the contour feature lines include pixel points on the outer contour line and pixel points on the inner contour line. Pixels are the original features inherent in the image.
  • the orientation of the contour feature line and the extraction of the square are the unique positioning of the orientation of the contour feature line and the extraction of the square.
  • the specific purpose of directional positioning is: in order to achieve the directional matching of pixel points on different image contour features, it is necessary to place the sample trademark image in a uniform directional position to solve the direction and angle of the image.
  • the specific purpose of extracting square positioning is: in order to achieve the comparability of pixel points on the contour lines of different image contours in the extraction square, it is necessary to place the sample trademark image in the extraction square to solve the integrity of the image recognition range. Comparability and uniqueness when there are deformations or differences in the size, shape, position, etc. of the image. In this way, the complete coordinate values can be accurately extracted in the coordinate system, and the comparability of the same or approximate image coordinate values in terms of integrity can be achieved.
  • the embodiment of the present invention achieves the above object by using a reference positioning method, and the reference positioning includes: a natural reference positioning and a reference linear positioning. Only one type of positioning standard can be selected in the same processing system, otherwise it will destroy its comparability.
  • Fig. 4 shows the directional positioning points of the natural reference positioning
  • Fig. 5 shows the positioning points of the natural reference positioning as the circumscribed rectangle.
  • the natural reference positioning is to shift the sample mark image contour line to the coordinate system with the minimum value of the x-axis coordinate of the pixel on the contour feature line of the sample trademark image without changing the rotation direction, and the minimum value of the y-axis coordinate is 0. in.
  • the use of natural reference positioning has the advantage of simple and easy positioning.
  • FIG. 6 and 7 show the directional positioning points of the reference straight line positioning
  • FIG. 8 shows the positioning points of the extracted squares of the reference straight line positioning as circumscribed squares.
  • the reference linear positioning is to find the positioning line on the contour feature line of the sample trademark image, and to translate the sample trademark image contour line with the minimum rotation angle to the positioning line parallel to the x-axis or the y-axis (in this embodiment, it is parallel to the x-axis) And the minimum value of the x-axis coordinate of the pixel on the contour feature of the sample trademark image is 0, and the sample trademark image contour line is aligned and centered in the extraction square to the y-axis.
  • the advantage of using the reference linear positioning is that the positioning is accurate, and the positioning of the different angles of the image or the irregular image has a better unique positioning reference.
  • the two pixels with the largest distance on the contour feature line are detected and calculated, and the contour feature line is rotated and translated to the two largest distances.
  • the virtual line between the pixels is parallel to the x-axis, and the minimum value of the x-axis coordinate of the pixel on the image contour feature line is zero.
  • the embodiment of Figure 7 performs directional positioning using a virtual straight line between two pixel points having the largest distance.
  • the virtual line between the two pixels with the largest distance is parallel with the x-axis or y-axis.
  • the virtual straight line corresponding to the time is used as a reference straight line to perform direction positioning.
  • the positioning of the sample trademark image contour line may also adopt a different positioning strategy as described above, for example, using the circumscribed circle method of the sample trademark image contour feature line to make the circumscribed circle and the x-axis and The y-axis is tangent; other geometric methods of the sample mark image contour feature line may be used to make the geometric shape tangent to the x-axis and the y-axis.
  • the coordinate values G n (x n , y n ) of all the pixel points on the contour feature line are extracted in the coordinate system with a single pixel point as a coordinate scale.
  • the index n represents the nth pixel point
  • the extracted pixel points include all the pixel points on the outer contour line and the inner contour line.
  • the total number of pixels on the contour feature line can be counted.
  • the coordinate values G n (x n , y n ) are represented by relative numbers, and the relative coordinate values S n (x n , y n ) are obtained:
  • Coordinate value in a way to express the relative number of pixels it may be referred to as relative coordinate values, denoted calculated as S n (x n, y n ), in particular, the coordinate values of the outline feature points into a line of relative pixel coordinate value as follows:
  • Relative coordinate value S n (x n , y n ) G n (x n /h, y n /h),
  • x n is the x-axis coordinate value of the nth pixel point in the coordinate system
  • y n is the y-axis coordinate value of the nth pixel point in the coordinate system
  • h is the maximum straight side length of the extracted square
  • the values of x n and y n of Sn are represented by relative numbers (percentage), and the values of x n and y n of Gn are represented by absolute numbers.
  • the relative number coordinates refer to coordinates reflected by converting the absolute coordinate value with respect to the coordinate origin to the ratio of the absolute coordinate value to the maximum straight line side length of the image extraction square.
  • the difference in coordinate values caused by the size difference can be eliminated, so that even if the size ratios of the two images are greatly different, substantially the same image can be recognized.
  • the conversion deformation tolerance range should be reasonably determined during the process of converting relative coordinate values.
  • the deformation tolerance means that when the pixel is offset in any direction to the deformation tolerance parameter, the relative coordinate value of the pixel remains unchanged.
  • Embodiments of the present invention determine the deformation tolerance i of the relative coordinate values S n (x n , y n ) according to image analysis. After the coordinate value G n (x n , y n ) of the pixel on the contour feature line is converted into the relative coordinate value S n (x n , y n ), the uniqueness of the sample trademark image can be distinguished according to the slight difference of the relative coordinate values.
  • the deformation tolerance can solve the problem of the image matching within the deformation tolerance range.
  • the specific value of the deformation tolerance should be determined according to the needs of image analysis.
  • the deformation tolerance parameter generally takes a percentage, such as 1%, 2%, and so on. In an embodiment of the invention, the deformation tolerance parameter i takes a range of 0.5% to 10%.
  • the x-axis and y-axis coordinates of the pixel relative to the coordinate value should be equally divided into 100 reference relative coordinates, and the relative coordinate values that do not fall into the reference relative coordinate must be rounded according to the rounding rule. To fall into the reference coordinates.
  • the rounding rule of the embodiment according to the deformation tolerance parameter, the relative coordinate value S n (x n , y n ) of the pixel is rounded according to the rounding rule of “not half-half, over-half-in” to obtain the deformation tolerance
  • Embodiment 1 If the obtained coordinate value of a set of pixel points on a contour image of a sample trademark image is obtained, the relative coordinate value S n (x n , y n ) is as follows:
  • Z 1 (0%, 52%), Z 2 (0%, 52%), Z 3 (2%, 50%), Z 4 (4%, 50%), Z 5 (4%, 48%), Z 6 (6%, 46%), Z 7 (8%, 44%), Z 8 (10%, 42%), Z 9 (12%, 38%).
  • the rounded relative coordinate value Z n (x n , y n ) of the precisely rounded pixel is output and stored in the sample trademark library.
  • the calculation result may be output.
  • the rounded relative coordinate value Z n (x n , y n ) of all the pixels on the contour feature line of the sample trademark image calculated above is outputted in data form and stored in the sample trademark image database, and can be used for other trademark images to be identified.
  • the rounding relative coordinate values Z n (x n , y n ) are identified, matched, compared, and analyzed, and the degree of coincidence of the relative coordinate values Z n (x n , y n ) of the pixels is reflected by the similarity of different trademarks. degree.
  • each successive pixel dot array is a connected domain contour line; and then dividing it into a sub-picture card of the sample trademark image; Finally, the sub-picture card is used as the processing object, and the processing of the above 1-8 is repeated, and the image feature information of the sub-picture card of the sample trademark image is extracted, and the relative coordinate value s n (x n , y n of the sub-picture card is obtained. And rounding the relative coordinate values z n (x n , y n ).
  • connected domain contours may have re-segmented graphic features, or multiple connected domain contours may be combined to form a relatively independent graphical element
  • the data processing personnel may edit through the devices and systems in the technical solution. By subdividing or merging its composition features, you can correctly divide the sub-picture card of the sample trademark image.
  • the rounded relative coordinate values z n (x n , y n ) of all the pixels on the outline of the sample trademark subgraph image calculated in the foregoing are outputted in data form and saved under a certain trademark record in the sample trademark database. Compare and analyze with the rounded relative coordinate value Z n (x n , y n ) of the image to be inspected, and the degree of coincidence of the rounded relative coordinate values of the pixel points to reflect the similarity of the two images.
  • the feature of the image to be inspected is identified and the image feature information is extracted.
  • the trademark to be inspected is used as the processing object, and the image feature of the trademark to be inspected is identified and processed.
  • the image feature information is extracted, and the image feature information of the trademark main image and the sub-picture card are respectively extracted.
  • the image feature information extracted by the embodiment of the present invention is mainly the relative coordinate value S n (x n , y n ) of the pixel points on the contour feature line of the trademark image to be inspected and the rounded relative coordinate value Z n (x n , y n ).
  • the image feature information includes but is not limited to S n (x n , y n ) and Z n (x n , y n ), and other image feature information can be obtained by deriving and transforming according to the image feature information, which can be used for characterization. The information contained in the image itself.
  • the main purpose of the matching check of the trademark image feature information is to find the trademark with the highest degree of approximation and the trademark graphic element code of the recorded logo image by matching the image feature information.
  • the trademark image feature information form is a relative coordinate value S n (x n , y n ) of the pixel on the contour feature line of the trademark image and a rounded relative coordinate value Z n (x n , y n ).
  • the matching of the image feature information is checked to find the sample trademark image with the highest degree of similarity.
  • the search content of the trademark search in the sample trademark database by using the trademark image feature information as a retrieval condition includes: 1) matching check of the relative coordinate value Z n (x n , y n ), 2) each A sub-picture card of the trademark image or an exact match check of each successive pixel point array, 3) a check of the number of mismatched pixels that round the relative coordinate value Z n (x n , y n ).
  • the approximation rate of the sub-picture card of the trademark image or the array of consecutive pixel points is used to comprehensively evaluate the approximation degree of the two trademark images.
  • Approximate degree of two images sub-picture card of trademark image or approximate rate of continuous pixel point array * sub-picture card weight of trademark image + coincidence rate of relative coordinate value * coincidence relative coordinate value weight + pixel relative coordinate value
  • Approximate degree of two images sub-picture card of trademark image or approximate rate of continuous pixel point array * sub-picture card weight of trademark image + coincidence rate of relative coordinate value * coincidence relative coordinate value weight + pixel relative coordinate value
  • Non-coincidence rate* does not coincide with relative coordinate value weights.
  • the weight parameter is determined according to the image analysis requirements, and the weight is generally 5% to 60%.
  • each successive pixel dot array divided on the outline of the trademark image to be identified is matched with each successive pixel dot array segmented on the sample trademark image outline to find a matching array.
  • the second step is to calculate the approximation rate of the array of consecutive pixel points, and calculate the formula:
  • Approximate rate of sub-picture card or continuous pixel point array (matched first consecutive pixel point array + matching second consecutive pixel point array + matching third consecutive pixel point array +... + matching nth consecutive pixel Point array)
  • the total number of pixels of the relative coordinate value of the image is *100%.
  • the pixel relative point coordinate Z n (x n , y n ) on the contour feature line of the to-be-identified trademark image is rounded to the pixel coordinate point on the contour of the sample trademark image.
  • the relative coordinate value Z n (x n , y n ) Perform a matching check, and count the number of coincidence matches, then calculate the coincidence rate, the formula:
  • Coincidence ratio (the number of coincident rounded relative coordinate values Z n (x n , y n ) ⁇ the total number of pixel points on the contour feature line of the trademark image to be identified) * 100%.
  • the coincidence ratio is equal to 100%, it can be confirmed that the two trademarks are the same trademark, and when the coincidence ratio is less than 100%, it can be confirmed that the trademark to be identified has the same contour line segment as the sample trademark.
  • Pixel point relative coordinate value non-coincidence number pixel point on the contour of the image to be recognized: the total relative coordinate value Z n (x n , y n ) - the pixel point rounding relative coordinate value Z n (x n , y n ) Coincidence number
  • Non-coincidence rate (the number of pixels relative to the coordinate value does not coincide with the total number of pixels on the outline of the trademark image to be identified) *100%
  • the trademark graphic element code corresponding to the sample trademark is encoded as a graphic element code of the trademark to be identified and output.
  • the sample trademark image with the highest approximation of the image feature information is found, especially the approximation degree is reached.
  • the original drawings of the four graphic trademarks of Apple, Shell, Blue Ribbon and PetroChina of FIGS. 9 and 10 are taken as an embodiment, and the sample with the highest degree of approximation is found in the sample trademark database by using the processing of the embodiment of the present invention.
  • the trademarks are the graphic trademark No. 167364, the graphic trademark No. 180720, the BLUE RIBBON and graphic trademark No. 559294, and the Chinese petroleum and graphic trademark No. 4360587.
  • the graphic element codes recorded in the four graphic trademarks are:
  • the trademark image may be modified or replaced with a design image
  • the trademark graphic element code may be modified or replaced with the code in the international design classification table.
  • trademark image may be modified or replaced with a product image
  • trademark graphic element code may be modified or replaced with the code of the commodity
  • a sample trademark library creation module for establishing a sample trademark library and establishing a correspondence relationship between the sample trademark and the known coded image data of the registered graphic trademark graphic elements
  • the sample trademark image feature information extraction module is configured to extract and process image feature information of the sample trademark, and establish a correspondence relationship between the sample trademark and the extracted image feature information;
  • the image information feature extraction module to be identified is used for extracting and processing image feature information of the trademark to be recognized;
  • the matching retrieval module is configured to perform matching search by using image feature information of the trademark to be identified as a retrieval condition, and find a sample trademark with the highest degree of similarity with the feature information of the trademark image to be identified and a corresponding trademark graphic element code;
  • the graphic element coding output module is configured to output the graphic graphic element code corresponding to the sample trademark as a graphic element code of the trademark to be identified.
  • the sample trademark library creation module may include: a graphic element coding record sub-module, configured to record a sample trademark in a sample trademark library by using a known applied-registered graphic trademark graphic element coding division data. Graphical element coding.
  • the sample trademark image feature information extraction module and the to-be-identified trademark image feature information extraction module may respectively include:
  • a trademark pre-processing sub-module for performing at least one pre-processing on the trademark including: graying, binarization, denoising;
  • contour feature line extraction sub-module for extracting a contour feature line from the pre-processed trademark image; wherein the contour feature line includes a set of edge pixels of the trademark image, a set of outer contour pixels, and a set of inner contour pixels ;
  • a contour feature line positioning sub-module for locating the contour feature line in the coordinate system and extracting a square, wherein the coordinate system is a coordinate constructed by a single pixel of the image as a unit of measurement of the x-axis and the y-axis system;
  • a pixel point coordinate value extraction sub-module configured to extract coordinate values G n (x n , y n ) of pixel points in the contour feature line in the coordinate system, wherein the angle mark n represents an n-th pixel point;
  • a relative coordinate value obtaining sub-module for expressing the coordinate value G n (x n , y n ) as a relative number according to a preset rule, thereby obtaining a relative coordinate value S n (x n , y n ), wherein the x-axis and The coordinate values of the y-axis are expressed in relative numbers (percents);
  • a coordinate value data output processing sub-module for outputting and saving the acquired relative coordinate values S n (x n , y n ) and the rounded relative coordinate values Z n (x n , y n );
  • Sub-picture card creation sub-module for establishing a sub-picture card of the trademark image
  • the image feature information extraction sub-module is configured to extract and process the image feature information of the sub-picture card, and obtain the relative coordinate values s n (x n , y n ) of the sub-picture card and the rounded relative coordinate value z n ( x n, y n).
  • the image feature information may be key feature information of the trademark image, including coordinate values G n (x n , y n ) of the pixel points on the contour feature line of the trademark image extracted after the positioning, and relative The coordinate values S n (x n , y n ) and the rounded relative coordinate values Z n (x n , y n ).
  • the image feature information may further include skeleton feature information, shape feature information, and/or template feature information extracted by the known image key feature information extracting means.
  • the match retrieval module can include at least one of the following:
  • a first retrieval sub-module for retrieving the relative coordinate value Z n (x n , y n ) of the contour image of the trademark image
  • a second search sub-module for searching for each successive pixel dot array or trademark image sub-picture card information on the contour feature line of the trademark image
  • the third search sub-module is configured to retrieve the number of unmatched pixel points of the relative coordinate value Z n (x n , y n ) on the contour feature line of the trademark image.
  • the direction of the contour feature line locating sub-module and the positioning of the extracted square may include: shifting the contour feature line extracted by the contour feature line extraction sub-module to the condition without changing the rotation direction to The minimum value of the x-axis coordinate of the pixel on the contour feature line is 0, and the minimum value of the y-axis coordinate of the pixel on the contour feature line is 0; wherein the extracted square is an circumscribed square of the contour feature line.
  • the directional positioning performed by the contour feature line positioning sub-module may include: finding the longest straight line on the contour feature line, and rotating the contour feature line to the contour feature line with a minimum degree of rotation angle The longest straight line is parallel to the x-axis or the y-axis, and then the minimum x-axis coordinate of the pixel on the contour feature line is 0, and the contour feature line is aligned to the y-axis and centered in the extracted square.
  • the extracted square positioning performed by the contour feature line positioning sub-module may include: a square formed by a maximum x-axis or a y-axis coordinate value of the contoured feature line after the positioning; One edge of the square coincides with the x-axis and the other edge coincides with the y-axis.
  • the relative coordinate value obtaining submodule may calculate and obtain the relative coordinate value S n (x n , y n ) according to the following formula:
  • Relative coordinate value S n (x n , y n ) G n (x n /h, y n /h),
  • x n represents the x-axis coordinate value of the nth pixel point in the coordinate system
  • y n represents the y-axis coordinate value of the nth pixel point in the coordinate system
  • h represents the maximum of the extracted square
  • the rounded relative coordinate values Z n (x n, y n ) can be acquired by the sub-module relative coordinate value rounded by the relative coordinate value S n (x n, y n ) by deformation tolerance parameters and rounding rules in terms of acquisition, wherein the value of x n and y n Z n is expressed as relative numbers (percent).
  • the embodiment of the present invention further provides a trademark graphic element identification system, the system comprising a memory and a server; wherein the memory and the server are configured to perform the following operations:
  • the trademark graphic element corresponding to the sample trademark is encoded as a graphic element code of the trademark to be identified and output.
  • the specific technical implementation manner of the trademark graphic element identification device and the trademark graphic element identification system of the embodiment of the present invention adopts the same manner as in the above-mentioned trademark retrieval method.
  • All or part of the process of implementing the foregoing embodiments of the present invention may be completed by a computer program to instruct related hardware, and the program may be stored in a computer readable storage medium.
  • the flow of an embodiment of the methods as described above may be included.
  • the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM). Therefore, according to the above embodiments of the present invention, the present invention further provides a storage medium comprising a computer readable program, which can implement the present invention in any of the above manners when the computer readable program in the storage medium is executed Trademark graphic element identification method.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Library & Information Science (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种商标图形要素识别方法,包括:建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系(101);对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系(102);对待识别商标的图像特征信息进行提取及处理(103);以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码(104);将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出(105)。利用本方法能够对商标图形自动划分商标图形要素编码。

Description

一种商标图形要素识别方法、装置、系统以及计算机存储介质 技术领域
本发明涉及商标信息检索领域,具体涉及一种商标图形要素识别方法、装置、系统以及计算机存储介质。
背景技术
商标检索是注册申请、商标审查、商标管理、商标维权等程序中的重要工作。传统的图形商标检索基本上是通过手工输入商标图形要素编码作为检索条件而实现检索目的。商标图形要素编码是依据《建立商标图形要素国际分类维也纳协定》所产生一种的商标图形要素划分工具,由商标图形要素按大类、小类及组分类的一览表组成,其中包括商标图形要素编号和商标图形要素名称构成。因此,每一商标图形要素编码代表了商标图形要素的内容意义。
目前,在世界范围内,商标图形要素编码主要通过国家商标主管机构中享有商标图形要素编码专业水平的少数审查员承担人工划分,基本没有智能化的工具或手段。现有的商标图形要素编码的人工划分方法虽然可以执行商标图形要素编码的划分任务,但其存在明显的缺陷和弊端,主要体现在:
1)人工划分商标图形要素编码,划分工作效率的低下和工作精力的消耗巨大是显然易见的;
2)商标图形要素编码划分要求专业性较强,普通工作人员不易准确掌握商标图形要素编码方法,限制了图形商标检索的广泛应用。
3)即使是专业人员进行商标图形要素编码划分,因不同专业人员主观判断商标图像的意义会存在差异,这种差异会造成商标图形要素编码的不一致。
发明内容
有鉴于此,本发明提出一种商标图形要素识别方法、装置和系统,以及一种计算机存储介质,利用已知的已申请注册图形商标图形要素编码划分的大数据资源分析,通过待识别商标图像特征信息与样本商标图像特征信息的匹配检索来获取特征信息重合度最高的样本商标及其所记录的图形要素编码,将样本商标的图形要素编码作为待识别商标的图形要素编码,可实现商标图形要素编码的自动化、标准化识别。
一方面,本发明提供一种商标图形要素识别方法,包括:S101,建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;S102,对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;S103,对待识别商标的图像特征信息进行提取及处理;S104,以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;S105,将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
另一方面,本发明还提供一种商标图形要素识别装置,包括:样本商标库创建模块,用于建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;样本商标图 像特征信息提取模块,用于对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;待识别商标图像特征信息提取模块,用于对待识别商标的图像特征信息进行提取及处理;匹配检索模块,用于以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;图形要素编码输出模块,用于将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
再一方面,本发明还提供一种商标图形要素识别系统,包括存储器和服务器;所述存储器和服务器经配置执行以下操作:建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;对待识别商标的图像特征信息进行提取及处理;以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
再一方面,本发明还提供一种包含计算机可读程序的存储介质,当该存储介质中的计算机可读程序执行时,执行上述商标图形要素识别方法。
本发明的实施例实现了利用现有已知的已申请注册图形商标图形要素编码划分的大数据资源,由系统学习并自动划分商标图形要素编码,解决了商标图形要素编码划分的标准化识别与以往因不同专业人员主观划分存在差异的问题。本发明的实施例与传统的手工划分商标图形要素编码相比,具有工作效率高和节省工作精力的好处,解决了传统商标图形要素编码仅靠人工划分,划分工作效率的低下和工作精力消耗巨大的弊端。本发明的实施例降低了对商标检索人员的商标图形要素编码专业水平的要求,有利于在更广泛的范围使用商标图形要素编码的商标检索技术,从而达到克服传统商标图形要素编码仅靠人工划分、划分标准难统一、主观划分存在差异的技术缺陷或局限,能有效实现商标图形要素编码的自动识别和标准化识别的目的。
附图说明
图1是本发明实施例的商标图形要素识别方法的流程图。
图2是本发明实施例的示例性图像原图。
图3是本发明实施例的示例性轮廓特征线图。
图4是本发明实施例的自然基准定位的方向定位图。
图5是本发明实施例的自然基准定位的提取方形定位图。
图6是本发明一个实施例的基准直线定位的方向定位图。
图7是本发明另一实施例的基准直线定位的方向定位图。
图8是本发明实施例的基准直线定位的提取方形定位图。
图9是本发明另一实施例的示例性图像原图。
图10是本发明另一实施例的示例性轮廓特征线图。
具体实施方式
以下结合附图以及具体实施例,对本发明的技术方案进行详细描述。
图1示出了本发明实施例的商标图形要素识别方法的流程图,包括以下步骤:
S101,建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
S102,对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
S103,对待识别商标的图像特征信息进行提取及处理;
S104,以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
S105,将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
以下对上述各个步骤进行详细描述。
第一,建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系。
图2随机给出若干样本商标的图像,在本发明实施例中,就是要建立样本商标库和利用已知的已申请注册图形商标图形要素编码划分数据在样本商标库中记录样本商标的图形要素编码信息。
其中商标图形要素编码是指依据《建立商标图形要素国际分类维也纳协定》所产生的一种商标图形要素划分工具,由商标图形要素按大类、小类及组分类的一览表组成,其中包括商标图形要素编号和商标图形要素名称构成。
第二,对样本商标图像特征进行识别处理和提取图像特征信息。
对样本商标图像特征进行识别处理和提取图像特征信息的目的是通过图像特征信息的匹配可以找到相同或最近似的商标。
由于多边形逼近法中的Teh-Chin检测算法提取的图像轮廓线能较好地反映商标的显著特征,本技术方案采用多边形逼近法中的Teh-Chin检测算法提取样本商标图像的关键像素点特征信息,即样本商标图像轮廓特征线上像素点的坐标值。
样本商标图像的关键像素点特征信息提取也可以采用其他已知的技术方法,包括提取骨架特征信息,提取形状特征信息,提取模板特征信息等。
在本发明的实施例中,对样本商标图像特征进行识别处理和提取图像特征信息的过程如下:
①对样本商标进行以下各项预处理中的至少一项预处理:灰度化、二值化去噪处理:
灰度数字图像是每个像素只有一个采样颜色的图像。这类图像通常显示为从最暗的黑色到最亮的白色的灰度,尽管理论上这个采样可以是任何颜色的不同深浅,甚至可以是不同亮度上的不同颜色。灰度图像与黑白图像不同,在计算机图像领域中黑白图像只有黑白两种颜色,灰度图像在黑色与白色之间还有许多级的颜色深度。
图像二值化就是将图像上的像素点的灰度值设置为0或255,也就是将整个图像呈现出明显的黑白效果。
噪点主要是指电荷耦合元件CCD(CMOS)将光线作为接收信号并输出的过程中所产生的图像中的粗糙部分,也指图像中不该出现的外来像素,通常由电子干扰产生。看起来就像图像被弄脏了,布满一些细小的糙点。我们平时所拍摄的数码照片如果用个人电脑将拍摄到的高画质图像缩小以后再看的话,也许就注意不到。不过,如果将原图像放大,那么就会出现本来没有的颜色(假色),这种假色就是图像噪点,可以通过技术手段将这些噪点去除。
②对经灰度化、二值化、去噪处理的样本商标图像提取其轮廓特征线,轮廓特征线包括样本商标图像的外轮廓线和内轮廓线:
在本发明的实施例中,利用多边形逼近法提取图像的关键像素点特征,即轮廓特征线,轮廓特征线就是轮廓线上像素点的集合,可生成样本商标图像轮廓特征线上的像素点序列图像。图3示出了若干图像的轮廓特征线,可以看到,轮廓特征线包括外轮廓线上的像素点和内轮廓线上的像素点。像素点是该图像中固有的原始特征。
③在坐标系中对轮廓特征线进行方向与提取方形的定位:
对轮廓特征线进行方向与提取方形的定位就是将轮廓特征线的放置位置采用一定的方法进行方向和提取方形的唯一性定位。
在实际应用中,方向定位的具体目的是:为了实现不同的图像轮廓特征线上像素点在方向上的可比性,需要将样本商标图像置于统一的方向定位位置上,解决图像的方向、角度等有变形或差异时的可比性和唯一性问题。这样才能在坐标系里准确提取其坐标值,实现相同或近似图像坐标值的可比性。
在实际应用中,提取方形定位的具体目的是:为了实现不同的图像轮廓特征线上像素点在提取方形内的可比性,需要将样本商标图像置于提取方形内,解决图像识别范围的完整性和图像的大小、形状、位置等有变形或差异时的可比性和唯一性问题。这样才能在坐标系里准确提取其完整的坐标值,实现相同或近似图像坐标值在完整性上的可比性。
具体地,本发明实施例采用基准定位的方式实现上述目的,基准定位包括:自然基准定位和基准直线定位。在同一处理体系中只能选定一种定位标准,否则会破坏其可比性。
图4示出了自然基准定位的方向定位要点,图5示出了自然基准定位的提取方形为外接矩形的定位要点。自然基准定位是将样本商标图像轮廓特征线在不改变旋转方向的条件下平移至样本商标图像轮廓特征线上像素点的x轴坐标最小值为0,且y轴坐标最小值为0的坐标系中。采用自然基准定位具有简单、易快速定位的优点。
图6和图7示出了基准直线定位的方向定位要点,图8示出了基准直线定位的提取方形为外接正方形的定位要点。基准直线定位是在样本商标图像轮廓特征线上找出定位直线,将样本商标图像轮廓特征线以最小的旋转度角平移至定位直线与x轴或y轴平行(本实施例为与x轴平行),且样本商标图像轮廓特征线上像素点的x轴坐标最小值为0,样本商标图像轮廓特征线在提取方形内向y轴对齐并居中。采用基准直线定位的优点是定位精确,对放置不同角度的图像或不规则的图像具有较好的唯一性定位基准。
以下描述不同情况下基准直线定位的处理过程:
1、采用基准直线定位进行方向定位。首先,检测样本商标图像轮廓特征线上像素点的排列是否存在直线特征,如果有直线特征,找出外轮廓线上的最长的直线,然后,旋转样本商标图像轮廓特征线, 以最小的旋转度角使该外轮廓线上最长的直线与x轴或y轴平行,即可确定轮廓特征线在方向上进行了定位。图6实施例采用外轮廓线上最长直线实施方向定位。
方向定位时如果出现外轮廓线上最长的直线有两条或两条以上时,选取实现最小的旋转度角使该外轮廓线上最长的直线与x轴或y轴平行的直线作为基准直线。
方向定位时如果出现样本商标图像轮廓特征线上像素点的排列不存在直线特征时,检测并计算轮廓特征线上距离最大的两个像素点,将轮廓特征线旋转和平移至该距离最大的两个像素点之间的虚拟直线与x轴平行,且图像轮廓特征线上像素点的x轴坐标最小值为0。图7实施例采用距离最大的两个像素点之间的虚拟直线实施方向定位。
方向定位时如果出现样本商标图像轮廓特征线上距离最大的像素点超过两个时,取实现以最小的旋转度角使距离最大的两个像素点之间的虚拟直线与x轴或y轴平行时所对应的该虚拟直线作为基准直线实施方向定位。
2、采用基准直线进行提取方形定位。首先,检测样本商标图像轮廓特征线上像素点在坐标系中x轴或y轴最大值,取该最大值为边长做正方形,并使样本商标图像轮廓特征线在正方形向y轴对齐并居中,然后平移该正方形使其一条边线与x轴重合,另一条边线与y轴重合,如图8实施例所示。
经过上述定位处理,使不同的图形特征信息之间具备可比性,使不同的图像拥有共同的参照标准和统一的比对环境。
此外,在本发明其他实施例中,对样本商标图像轮廓特征线进行定位还可采取与上述不同的定位策略,例如利用样本商标图像轮廓特征线的外接圆法,使该外接圆与x轴和y轴相切;也可采用样本商标图像轮廓特征线的外接其他几何图形法,使该几何图形与x轴和y轴相切。
④在所述坐标系中提取轮廓特征线中像素点的坐标值Gn(xn,yn):
样本商标图像轮廓特征线经方向定位和提取方形定位后,以单个像素点为一个坐标刻度在坐标系内提取轮廓特征线上全部像素点的坐标值Gn(xn,yn),其中角标n代表第n个像素点,所提取的像素点包括外轮廓线和内轮廓线上的所有像素点。同时可统计轮廓特征线上的像素点总数。
需要注意的是,提取样本商标图像轮廓特征线上每个像素点的坐标值Gn(xn,yn)时,应基于同一方向定位和提取方形定位标准,否则会破坏其可比性。
⑤按预设的规则将坐标值Gn(xn,yn)以相对数表示,获得相对坐标值Sn(xn,yn):
对于视觉上相同的多个图像,因不同图像的大小存在差异,所以即使经过图像的方向定位和提取方形定位后,也很难保证其轮廓线上全部像素点的坐标值相同,这为图像的坐标匹配带来困难。为解决这个问题,对轮廓线上像素点的坐标值进行相对数转换,得到每个像素点的相对坐标值,可以有效解决图像在大小存在差异情况下的坐标匹配问题。以相对数方式表达像素点的坐标值,可称为相对坐标值,记为Sn(xn,yn),具体地,轮廓特征线上像素点的坐标值转换为相对坐标值的计算公式如下:
相对坐标值Sn(xn,yn)=Gn(xn/h,yn/h),
其中,xn为第n个像素点在所述坐标系中的x轴坐标值,
yn为第n个像素点在坐标系中的y轴坐标值,
h为提取方形的最大直线边长,
其中,Sn的xn和yn的值以相对数(百分数)表示,Gn的xn和yn的值以绝对数表示。
在本发明的实施例中,相对数坐标是指通过将相对于坐标原点的绝对坐标值转换为以绝对坐标值与图像提取方形的最大直线边长的比率反映的坐标。
采用相对坐标值表示轮廓特征线上的像素点特征,可以消除由尺寸差异导致的坐标值差异,使得即使两个图像的尺寸比例差别巨大,也可以识别出实质相同的图像。
⑥根据图像分析需求确定相对坐标值Sn(xn,yn)的变形公差参数。
在实际应用中需要注意:转换相对坐标值的过程中应合理确定转换变形公差范围。变形公差是指当像素点在任意方向上偏移至变形公差参数范围内时,该像素点的相对坐标值保持不变。本发明的实施例根据图像分析需要确定相对坐标值Sn(xn,yn)的变形公差i。轮廓特征线上像素点的坐标值Gn(xn,yn)转换为相对坐标值Sn(xn,yn)后,能根据相对坐标值微小的差异区分样本商标图像的唯一性,但坐标匹配时则可能很难找到相匹配的图像,采用变形公差可以解决图像在变形公差范围内的坐称匹配问题。变形公差具体的取值应根据图像分析的需求确定。变形公差参数一般取百分数,如1%,2%等等。在本发明的实施例中,变形公差参数i取0.5%至10%的范围。
⑦按变形公差参数和“未过半舍,过半入”的取整规则对相对坐标值Sn(xn,yn)进行取整,以获取经变形公差精确取整后的像素点的取整相对坐标值Zn(xn,yn):
当变形公差参数取1%时,像素点相对坐标值的x轴和y轴坐标应平均划分为100个基准相对坐标,凡不落入基准相对坐标的相对坐标值须按取整规则取整,以落入基准坐标。
本实施例的取整规则:根据变形公差参数对像素点的相对坐标值Sn(xn,yn)按“未过半舍,过半入”的取整规则进行取整,以获取经变形公差精确后的像素点的取整相对坐标值Zn(xn,yn)。
结合本发明的如下实施例说明取整的具体处理过程:
实施例一:假如获得的某一样本商标图像轮廓特征线上一组像素点的相对坐标值Sn(xn,yn)如下:
S1(0%,52.321%),S2(0.891%,51.567%),S3(2.189%,50.463%),S4(3.986%,49.646%),S5(4.895%,47.347%),S6(6.263%,45.396%),S7(8.231%,43.373%),S8(9.172%,41.502%),S9(11.265%,38.674%)。
以变形公差参数i取2%为例,按“未过半舍,过半入”的取整规则进行取整,获取以上像素点经变形公差精确后的取整相对坐标值Zn(xn,yn)如下:
Z1(0%,52%),Z2(0%,52%),Z3(2%,50%),Z4(4%,50%),Z5(4%,48%),Z6(6%,46%),Z7(8%,44%),Z8(10%,42%),Z9(12%,38%)。
⑧将经精确取整后的像素点的取整相对坐标值Zn(xn,yn)输出,保存于样本商标库。
在本发明的实施例中,在计算得到样本商标图像轮廓特征线上像素点的取整相对坐标值Zn(xn,yn)之后,可将计算结果输出。前述计算出的样本商标图像轮廓特征线上所有像素点的取整相对坐标值Zn(xn,yn),以数据形式输出并保存在样本商标图像数据库,可用于与其他待识别商标图像的取整相对坐标值Zn(xn,yn)进行识别、匹配、比较、分析,通过像素点的取整相对坐标值Zn(xn,yn)的重合程度反映不同商标的相似度。
⑨建立样本商标图像的子图分卡并提取子图分卡的图像特征信息:
首先,检测已提取样本商标图形轮廓线中相对独立部分的连通域轮廓线,即每一个连续像素点数组为一个连通域轮廓线;然后,将其分割作为该样本商标图像的子图分卡;最后,以子图分卡为处理对象, 重复前述①-⑧的处理,提取样本商标图像的子图分卡的图像特征信息,获取子图分卡的相对坐标值sn(xn,yn)和取整相对坐标值zn(xn,yn)。
其中,样本商标图像的子图分卡的处理过程:
1)对样本图像轮廓线上像素点取整相对坐标值Zn(xn,yn)进行排序,像素点取整相对坐标值Zn(xn,yn)排序规则:第1排序按x轴的值升序排序,第2排序按y轴的值升序排序。也可降序排序或混合排序。
2)计算和检测连续像素点数组,即检测像素点间的排列是否具有连续排列的特征,检测方法:自x轴为0的像素点起检测相邻像素点,当排序后的相邻像素点x轴或y轴坐标数增加或减少数超过1个刻度值时,视为断点,断点前的一组像素点为第1个连续像素点数组;自断点后的像素点起再检测相邻像素点,当相邻像素点x轴或y轴坐标数增加或减少数超过1个刻度值时,视为又一断点,断点前的一组像素点分割为第2个连续像素点数组;如此类推,直至检测完轮廓特征线上全部像素点。
3)将每一连续像素点数组保存为一个样本商标图像的子图分卡。
对于一些连通域轮廓线可能具备再分割的图形要素时,或多个连通域轮廓线可能组合构成一个相对独立的图形要素时,可以由数据处理人员通过本技术方案中的装置和系统进行编辑,自定义分割或合并其构图要素,即可正确划分样本商标图像的子图分卡。
⑩输出并保存从样本商标图像子图分卡所提取的相对坐标值sn(xn,yn)和取整相对坐标值zn(xn,yn)。
前述计算出的样本商标子图图像轮廓线上所有像素点的取整相对坐标值zn(xn,yn),以数据形式输出并保存在样本商标数据库的某一商标记录之下,可用于与待检商标图像的取整相对坐标值Zn(xn,yn)匹配比较分析,通过像素点的取整相对坐标值的重合度多少去反映两图像的相似度。
第三,对待检商标图像特征进行识别处理和提取图像特征信息。
在本发明的实施例中,参照前述第二点中“对样本商标图像特征进行识别处理和提取图像特征信息”的处理过程,以待检商标为处理对象,对待检商标图像特征进行识别处理和提取图像特征信息,分别提取待检商标主图和子图分卡的图像特征信息。
本发明实施例所提取的图像特征信息主要是待检商标图像轮廓特征线上像素点的相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)。注意:图像特征信息包含但不限于Sn(xn,yn)和Zn(xn,yn),根据此图像特征信息进行推导和变型也可得到其他图像特征信息,均可用于表征图像本身包含的信息。
第四,以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码(即样本商标库中记录的该样本商标对应的图形要素编码)。
商标图像特征信息的匹配检查主要目的是通过图像特征信息的匹配检查找出近似度最高的商标及所记录标志图像的商标图形要素编码。
本发明实施例采用商标图像特征信息形式是商标图像轮廓特征线上像素点的相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn),通过商标图像特征信息的匹配检查,找出近似度最高的样本商标图像。
本发明实施例中以商标图像特征信息为检索条件在样本商标数据库中进行商标检索的检索内容包括:1)取整相对坐标值Zn(xn,yn)的匹配检查,2)每一商标图像的子图分卡或每一连续像素点数组的 完全匹配检查,3)取整相对坐标值Zn(xn,yn)的不匹配像素点数检查。
虽然两商标图像轮廓特征线上像素点的取整相对坐标值Zn(xn,yn)完全重合一般会构成相同或最近似的商标。但在不完全重合的情况下,说明仅在部分图像轮廓特征线上像素点重合,不重合部分可能影响商标的显著特征,从而导致两商标不一定构成相同或最近似商标的结果。
本发明实施例采用了商标图像的子图分卡或连续像素点数组的近似率、像素点相对坐标值重合率、像素点相对坐标值不重合率综合评价两商标图像的近似程度。计算公式:
两图像的近似度=商标图像的子图分卡或连续像素点数组的近似率*商标图像的子图分卡权数+相对坐标值重合率*重合相对坐标值权数+像素点相对坐标值不重合率*不重合相对坐标值权数。
权数参数根据图像分析需求确定,权数一般可取5%至60%。
以上计算公式中各变量获取计算过程如下:
a)商标图像的子图分卡或连续像素点数组的近似率计算:
商标图像的子图分卡或连续像素点数组的近似率计算处理步骤如下:
第一步,以待识别商标图像轮廓线上所分割的每一连续像素点数组与样本商标图像轮廓线上所分割的每一连续像素点数组进行匹配,找出匹配的数组。
第二步,计算连续像素点数组的近似率,计算公式:
子图分卡或连续像素点数组的近似率=(匹配的第1连续像素点数组+匹配的第2连续像素点数组+匹配的第3连续像素点数组+……+匹配的第n连续像素点数组)÷本图像的相对坐标值的像素点总数*100%。
b)像素点相对坐标值重合率:
以待识别商标图像轮廓特征线上像素点取整相对坐标值Zn(xn,yn)逐个与样本商标图像轮廓特征线上像素点取整相对坐标值Zn(xn,yn)进行匹配检查,并统计出重合匹配的数量,然后计算其重合率,公式:
重合率=(重合的取整相对坐标值Zn(xn,yn)数÷待识别商标图像轮廓特征线上像素点总数)*100%。
当重合率等于100%时,可以确认两商标是相同的商标,当重合率小于100%时,可以确认待识别商标与样本商标有一部分轮廓特征线段相同。
c)像素点相对坐标值不重合率:
在两商标匹配比对过程中大多数商标图像像素点相对坐标值不会重合,匹配过程应考虑不重合的像素点取整相对坐标值Zn(xn,yn)对两商标图像的近似程度的影响,本发明实施例采用了像素点相对坐标值不重合率的评价,计算公式如下:
像素点相对坐标值不重合数=待识别商标图像轮廓线上像素点取整相对坐标值Zn(xn,yn)总数-像素点取整相对坐标值Zn(xn,yn)重合数
不重合率=(像素点相对坐标值不重合数÷待识别商标图像轮廓线上像素点总数)*100%
在图像特征信息匹配和近似度评价中找出近似度最高的样本商标及所记录的商标图形要素编码。
第五,将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
经待识别商标主图及子图分卡图像特征信息与样本商标主图及子图分卡图像特征信息的匹配检查,找出了图像特征信息近似度最高的样本商标图像,尤其是近似度达到100%时,可以确认两商标图像是 相同的商标,当近似度接近100%时,可以确认两商标图像是高度近似的商标。因此,可以将上述找出的近似度最高的样本商标及所记录的商标图形要素编码视为待识别商标的图形要素编码并进行输出,以用于本发明实施例的商标检索系统。
本发明实施例以图9和图10的苹果、壳牌、蓝带、中国石油四件图形商标原图为实施例,采用本发明实施例的处理,在样本商标数据库中找出近似度最高的样本商标分别是第167364号图形商标、第180720号图形商标、第559294蓝带BLUE RIBBON及图形商标、第4360587号中国石油及图形商标,可以获得该四件图形商标所记录的图形要素编码分别是:
第167364号苹果图形商标的图形要素编码:5.7.13;
第180720号壳牌图形商标的图形要素编码:3.8.18;
第559294蓝带BLUE RIBBON及图形商标的图形要素编码:24.5.20;25.1.6;
第4360587号中国石油及图形商标的图形要素编码:1.3.1;25.1.25;29.1.13;5.5.20;A1.3.15。
上述内容以商标的检索应用为例描述了本发明的实施例。其实,本发明的技术方案还可以用在其他类似的应用场合中,例如:
1、在外观设计专利申请的检索应用中,可将上述商标图像变型或替换为外观设计图像,将商标图形要素编码变型或替换为国际外观设计分类表中的编码。
2、在相同或相似商品的检索应用中,可将上述商标图像变型或替换为商品图像,将商标图形要素编码变型或替换为商品的编码。
一个实施例中本发明提供的商标图形要素识别装置,包括有:
样本商标库创建模块,用于建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
样本商标图像特征信息提取模块,用于对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
待识别商标图像特征信息提取模块,用于对待识别商标的图像特征信息进行提取及处理;
匹配检索模块,用于以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
图形要素编码输出模块,用于将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
其中,在一个具体示例中,所述样本商标库创建模块可以包括:图形要素编码记录子模块,用于利用已知的已申请注册图形商标图形要素编码划分数据在样本商标库中记录样本商标的图形要素编码。
所述样本商标图像特征信息提取模块和待识别商标图像特征信息提取模块分别可以包括:
商标预处理子模块,用于对所述商标进行包括以下各项的至少一项预处理:灰度化、二值化、去噪处理;
轮廓特征线提取子模块,用于对经上述预处理的商标图像提取轮廓特征线;其中所述轮廓特征线包括商标图像的边缘像素的集合、外轮廓线像素的集合和内轮廓线像素的集合;
轮廓特征线定位子模块,用于在坐标系中对轮廓特征线进行方向与提取方形的定位,其中,所述坐标系是以图像的单个像素为x轴和y轴的计量单位而构建的坐标系;
像素点坐标值提取子模块,用于在所述坐标系中提取轮廓特征线中像素点的坐标值Gn(xn,yn),其中,角标n代表第n个像素点;
相对坐标值获取子模块,用于按预设的规则将坐标值Gn(xn,yn)以相对数表示,从而获得相对坐标值Sn(xn,yn),其中x轴和y轴的坐标值以相对数(百分数)表示;
取整相对坐标值获取子模块,用于按预设的变形公差参数和“未过半舍,过半入”的取整规则对相对坐标值Sn(xn,yn)进行取整,从而得到取整相对坐标值Zn(xn,yn);
坐标值数据输出处理子模块,用于对获取的相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)进行输出和保存;
子图分卡创建子模块,用于建立商标图像的子图分卡;
图像特征信息提取子模块,用于对子图分卡的图像特征信息进行提取及处理,获取子图分卡的相对坐标值sn(xn,yn)和取整相对坐标值zn(xn,yn)。
在此情况下,所述图像特征信息可以是商标图像的关键特征信息,其包括经过所述定位之后提取的商标图像轮廓特征线上像素点的坐标值Gn(xn,yn)、相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)。另一方面,所述图像特征信息还可以包括采用已知的图像关键特征信息提取装置所提取的骨架特征信息、形状特征信息和/或模板特征信息。
在一个具体示例中,所述匹配检索模块可以包括以下各项的至少一项:
第一检索子模块,用于对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的检索;
第二检索子模块,用于对商标图像轮廓特征线上每一连续像素点数组或商标图像子图分卡信息的检索;
第三检索子模块,用于对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的不匹配像素点数的检索。
在一个具体示例中,所述轮廓特征线定位子模块进行的方向与提取方形的定位可以包括:将所述轮廓特征线提取子模块所提取的轮廓特征线在不改变旋转方向的条件下平移至轮廓特征线上像素点的x轴坐标最小值为0,且轮廓特征线上像素点的y轴坐标最小值为0的位置;其中,所述提取方形是轮廓特征线的外接方形。
在另一个具体示例中,所述轮廓特征线定位子模块进行的方向定位可以包括:找出轮廓特征线上最长的直线,以最小的旋转度角将轮廓特征线旋转至使轮廓特征线上最长的直线与x轴或y轴平行,再平移至轮廓特征线上像素点的x轴坐标最小值为0,轮廓特征线在提取方形内向y轴对齐并居中位置。
在另一个具体示例中,所述轮廓特征线定位子模块进行的提取方形定位可以包括:以定位后的轮廓特征线上最大的x轴或y轴坐标值为边长所构成的方形,所述方形的一条边线与x轴重合,另一条边线与y轴重合。
其中,所述相对坐标值获取子模块可以按如下公式计算取得所述相对坐标值Sn(xn,yn):
相对坐标值Sn(xn,yn)=Gn(xn/h,yn/h),
其中,xn表示第n个像素点在所述坐标系中的x轴坐标值,yn表示第n个像素点在所述坐标系中的y轴坐标值,h表示所述提取方形的最大直线边长,其中,Sn的xn和yn的值以相对数(百分数)表示,Gn的xn和yn的值以绝对数表示。
在一个具体示例中,所述取整相对坐标值Zn(xn,yn)可以是由所述取整相对坐标值获取子模块由相对坐标值Sn(xn,yn)按变形公差参数和取整规则的换算取得,其中,Zn的xn和yn的值以相对数(百分数)表示。
根据上述商标图形要素识别方法和商标图形要素识别装置,本发明实施例还提供一种商标图形要素识别系统,该系统包括存储器和服务器;其中,所述存储器和服务器经配置执行以下操作:
建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
对待识别商标的图像特征信息进行提取及处理;
以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
本领域普通技术人员可以理解的是,本发明实施例的商标图形要素识别装置、商标图形要素识别系统的具体技术实现方式采用与上述商标检索方法中相同的方式。而实现上述本发明实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的程序可存储于一计算机可读取存储介质中,该程序在执行时,可包括如上述各方法的实施例的流程。其中,所述的存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory,ROM)或随机存储记忆体(Random Access Memory,RAM)等。因此,根据上述本发明实施例方案,本发明还提供一种包含计算机可读程序的存储介质,当该存储介质中的计算机可读程序执行时,可以实现上述任何一种方式中的本发明的商标图形要素识别方法。
以上,结合具体实施例对本发明的技术方案进行了详细介绍,所描述的具体实施例用于帮助理解本发明的思想。本领域技术人员在本发明具体实施例的基础上做出的推导和变型也属于本发明保护范围之内。

Claims (24)

  1. 一种商标图形要素识别方法,其特征在于,包括:
    S101,建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
    S102,对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
    S103,对待识别商标的图像特征信息进行提取及处理;
    S104,以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
    S105,将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
  2. 根据权利要求1所述的方法,其特征在于,其中S101所述建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系包括:利用已知的已申请注册图形商标图形要素编码划分数据在样本商标库中记录样本商标的图形要素编码。
  3. 根据权利要求1所述的方法,其特征在于,其中在S102和S103中,对所述商标的图像特征信息进行提取及处理包括:
    ①对所述商标进行以下各项的至少一项预处理:灰度化、二值化、去噪处理;
    ②对经上述预处理的商标图像提取轮廓特征线;其中所述轮廓特征线包括商标图像的边缘像素的集合、外轮廓线像素的集合和内轮廓线像素的集合;
    ③在坐标系中对轮廓特征线进行方向与提取方形的定位,其中,所述坐标系是以图像的单个像素为x轴和y轴的计量单位而构建的坐标系;
    ④在所述坐标系中提取轮廓特征线中像素点的坐标值Gn(xn,yn),其中,角标n代表第n个像素点;
    ⑤按预设的规则将坐标值Gn(xn,yn)以相对数表示,从而获得相对坐标值Sn(xn,yn),其中,Sn的xn和yn的值以相对数(百分数)表示,Gn的xn和yn的值以绝对数表示;
    ⑥按预设的变形公差参数和“未过半舍,过半入”的取整规则对相对坐标值Sn(xn,yn)进行取整,从而得到取整相对坐标值Zn(xn,yn);
    ⑦建立商标图像的子图分卡;
    ⑧对子图分卡的图像特征信息进行提取及处理,获取子图分卡的相对坐标值sn(xn,yn)和取整相对坐标值zn(xn,yn)。
  4. 如权利要求3所述的方法,其特征在于,其中所述图像特征信息是商标图像的关键特征信息,其包括经过所述定位之后提取的商标图像轮廓特征线上像素点的坐标值Gn(xn,yn)、相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)。
  5. 如权利要求4所述的方法,其特征在于,其中所述图像特征信息还包括采用已知的图像关键特征信息提取方法所提取的骨架特征信息、形状特征信息和/或模板特征信息。
  6. 如权利要求1所述的方法,其特征在于,其中S104所述以待识别商标的图像特征信息为检 索条件进行匹配检索包括以下各项的至少一项:
    对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的检索;
    对商标图像轮廓特征线上每一连续像素点数组或商标图像子图分卡信息的检索;
    对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的不匹配像素点数的检索。
  7. 如权利要求3所述的方法,其特征在于,其中方向与提取方形的定位包括:将步骤①所提取的轮廓特征线在不改变旋转方向的条件下平移至轮廓特征线上像素点的x轴坐标最小值为0,且轮廓特征线上像素点的y轴坐标最小值为0的位置;其中,所述提取方形是轮廓特征线的外接方形。
  8. 如权利要求3所述的方法,其特征在于,其中方向定位包括:找出轮廓特征线上最长的直线,以最小的旋转度角将轮廓特征线旋转至使轮廓特征线上最长的直线与x轴或y轴平行,再平移至轮廓特征线上像素点的x轴坐标最小值为0,轮廓特征线在提取方形内向y轴对齐并居中位置。
  9. 如权利要求3所述的方法,其特征在于,其中提取方形定位包括:以步骤③定位后的轮廓特征线上最大的x轴或y轴坐标值为边长所构成的方形,所述方形的一条边线与x轴重合,另一条边线与y轴重合。
  10. 如权利要求3所述的方法,其特征在于,其中所述相对坐标值Sn(xn,yn)按如下公式计算取得:
    相对坐标值Sn(xn,yn)=Gn(xn/h,yn/h),
    其中,xn表示第n个像素点在所述坐标系中的x轴坐标值,yn表示第n个像素点在所述坐标系中的y轴坐标值,h表示所述提取方形的最大直线边长,其中,Sn的xn和yn的值以相对数(百分数)表示,Gn的xn和yn的值以绝对数表示。
  11. 如权利要求3所述的方法,其特征在于,其中所述取整相对坐标值Zn(xn,yn)是由相对坐标值Sn(xn,yn)按变形公差参数和取整规则的换算取得,其中,Zn的xn和yn的值以相对数(百分数)表示。
  12. 一种商标图形要素识别装置,其特征在于,包括:
    样本商标库创建模块,用于建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
    样本商标图像特征信息提取模块,用于对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
    待识别商标图像特征信息提取模块,用于对待识别商标的图像特征信息进行提取及处理;
    匹配检索模块,用于以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
    图形要素编码输出模块,用于将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
  13. 根据权利要求12所述的装置,其特征在于,其中所述样本商标库创建模块包括:图形要素编码记录子模块,用于利用已知的已申请注册图形商标图形要素编码划分数据在样本商标库中记录样本商标的图形要素编码。
  14. 根据权利要求12所述的装置,其特征在于,其中所述样本商标图像特征信息提取模块和待识别商标图像特征信息提取模块分别包括:
    商标预处理子模块,用于对所述商标进行包括以下各项的至少一项预处理:灰度化、二值化、去噪处理;
    轮廓特征线提取子模块,用于对经上述预处理的商标图像提取轮廓特征线;其中所述轮廓特征线包括商标图像的边缘像素的集合、外轮廓线像素的集合和内轮廓线像素的集合;
    轮廓特征线定位子模块,用于在坐标系中对轮廓特征线进行方向与提取方形的定位,其中,所述坐标系是以图像的单个像素为x轴和y轴的计量单位而构建的坐标系;
    像素点坐标值提取子模块,用于在所述坐标系中提取轮廓特征线中像素点的坐标值Gn(xn,yn),其中,角标n代表第n个像素点;
    相对坐标值获取子模块,用于按预设的规则将坐标值Gn(xn,yn)以相对数表示,从而获得相对坐标值Sn(xn,yn),其中x轴和y轴的坐标值以相对数(百分数)表示;
    取整相对坐标值获取子模块,用于按预设的变形公差参数和“未过半舍,过半入”的取整规则对相对坐标值Sn(xn,yn)进行取整,从而得到取整相对坐标值Zn(xn,yn);
    坐标值数据输出处理子模块,用于对获取的相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)进行输出和保存;
    子图分卡创建子模块,用于建立商标图像的子图分卡;
    图像特征信息提取子模块,用于对子图分卡的图像特征信息进行提取及处理,获取子图分卡的相对坐标值sn(xn,yn)和取整相对坐标值zn(xn,yn)。
  15. 如权利要求14所述的装置,其特征在于,其中所述图像特征信息是商标图像的关键特征信息,其包括经过所述定位之后提取的商标图像轮廓特征线上像素点的坐标值Gn(xn,yn)、相对坐标值Sn(xn,yn)和取整相对坐标值Zn(xn,yn)。
  16. 如权利要求15所述的装置,其特征在于,其中所述图像特征信息还包括采用已知的图像关键特征信息提取装置所提取的骨架特征信息、形状特征信息和/或模板特征信息。
  17. 如权利要求12所述的装置,其特征在于,其中所述匹配检索模块包括以下各项的至少一项:
    第一检索子模块,用于对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的检索;
    第二检索子模块,用于对商标图像轮廓特征线上每一连续像素点数组或商标图像子图分卡信息的检索;
    第三检索子模块,用于对商标图像轮廓特征线上取整相对坐标值Zn(xn,yn)的不匹配像素点数的检索。
  18. 如权利要求14所述的装置,其特征在于,其中所述轮廓特征线定位子模块进行的方向与提取方形的定位包括:将所述轮廓特征线提取子模块所提取的轮廓特征线在不改变旋转方向的条件下平移至轮廓特征线上像素点的x轴坐标最小值为0,且轮廓特征线上像素点的y轴坐标最小值为0的位置;其中,所述提取方形是轮廓特征线的外接方形。
  19. 如权利要求14所述的装置,其特征在于,其中所述轮廓特征线定位子模块进行的方向定位包括:找出轮廓特征线上最长的直线,以最小的旋转度角将轮廓特征线旋转至使轮廓特征线上最长的直线与x轴或y轴平行,再平移至轮廓特征线上像素点的x轴坐标最小值为0,轮廓特征线在提取方形内向y轴对齐并居中位置。
  20. 如权利要求14所述的装置,其特征在于,其中所述轮廓特征线定位子模块进行的提取方形定位包括:以定位后的轮廓特征线上最大的x轴或y轴坐标值为边长所构成的方形,所述方形的一条边线与x轴重合,另一条边线与y轴重合。
  21. 如权利要求14所述的装置,其特征在于,其中所述相对坐标值获取子模块按如下公式计算取得所述相对坐标值Sn(xn,yn):
    相对坐标值Sn(xn,yn)=Gn(xn/h,yn/h),
    其中,xn表示第n个像素点在所述坐标系中的x轴坐标值,yn表示第n个像素点在所述坐标系中的y轴坐标值,h表示所述提取方形的最大直线边长,其中,Sn的xn和yn的值以相对数(百分数)表示,Gn的xn和yn的值以绝对数表示。
  22. 如权利要求14所述的装置,其特征在于,其中所述取整相对坐标值Zn(xn,yn)是由所述取整相对坐标值获取子模块由相对坐标值Sn(xn,yn)按变形公差参数和取整规则的换算取得,其中,Zn的xn和yn的值以相对数(百分数)表示。
  23. 一种商标图形要素识别系统,其特征在于,包括存储器和服务器;所述存储器和服务器经配置执行以下操作:
    建立样本商标库并建立样本商标与已知的已申请注册图形商标图形要素编码划分数据的对应关系;
    对样本商标的图像特征信息进行提取及处理,并建立样本商标与所提取的图像特征信息的对应关系;
    对待识别商标的图像特征信息进行提取及处理;
    以待识别商标的图像特征信息为检索条件进行匹配检索,找出与待识别商标图像特征信息近似度最高的样本商标及对应的商标图形要素编码;
    将所述样本商标对应的商标图形要素编码作为待识别商标的图形要素编码进行输出。
  24. 一种包含计算机可读程序的存储介质,其特征在于,当该存储介质中的计算机可读程序执行时,执行上述权利要求1至11任意一项中的商标图形要素识别方法。
PCT/CN2016/075871 2015-12-15 2016-03-08 一种商标图形要素识别方法、装置、系统以及计算机存储介质 WO2017101225A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
GB1811592.3A GB2561773A (en) 2015-12-15 2016-03-08 Trademark graph element identification method, apparatus and system, and computer storage medium
US16/062,001 US10430687B2 (en) 2015-12-15 2016-03-08 Trademark graph element identification method, apparatus and system, and computer storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510945834.4 2015-12-15
CN201510945834.4A CN105574161B (zh) 2015-12-15 2015-12-15 一种商标图形要素识别方法、装置和系统

Publications (1)

Publication Number Publication Date
WO2017101225A1 true WO2017101225A1 (zh) 2017-06-22

Family

ID=55884292

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/075871 WO2017101225A1 (zh) 2015-12-15 2016-03-08 一种商标图形要素识别方法、装置、系统以及计算机存储介质

Country Status (4)

Country Link
US (1) US10430687B2 (zh)
CN (1) CN105574161B (zh)
GB (1) GB2561773A (zh)
WO (1) WO2017101225A1 (zh)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3411828A4 (en) 2016-02-01 2019-09-25 See-Out Pty Ltd. CLASSIFICATION AND LABELING OF IMAGES
CA3020845C (en) * 2016-04-14 2023-06-20 Ader Bilgisayar Hizmetleri Ve Ticaret A.S. Content based search and retrieval of trademark images
CN106649851A (zh) * 2016-12-30 2017-05-10 徐庆 近似商标查询结果排序方法、装置及其商标服务器
CN109240572B (zh) * 2018-07-20 2021-01-05 华为技术有限公司 一种获取图片的方法、对图片进行处理的方法及装置
CN109857897B (zh) * 2019-02-14 2021-06-29 厦门一品威客网络科技股份有限公司 一种商标图像检索方法、装置、计算机设备及存储介质
CN110674328A (zh) * 2019-09-27 2020-01-10 长城计算机软件与系统有限公司 一种商标图像检索方法、系统、介质及设备
CN112132783B (zh) * 2020-08-21 2022-09-20 成都飞机工业(集团)有限责任公司 一种基于数字图像处理技术的零件识别方法
CN113553980A (zh) * 2021-07-30 2021-10-26 徐庆 一种图片的商标图形要素编码生成方法、系统及装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101464951A (zh) * 2007-12-21 2009-06-24 北大方正集团有限公司 图像识别方法及系统
CN101894154A (zh) * 2010-07-13 2010-11-24 重庆大学 一种图像文档关键图案的提取方法
CN104156413A (zh) * 2014-07-30 2014-11-19 中国科学院自动化研究所 一种基于商标密度的个性化商标匹配识别方法
CN104199931A (zh) * 2014-09-04 2014-12-10 厦门大学 一种商标图像一致语义提取方法及商标检索方法
CN104298990A (zh) * 2014-09-15 2015-01-21 西安电子科技大学 一种基于骨架图的快速图形匹配与识别的方法
US9058543B2 (en) * 2010-11-01 2015-06-16 Raf Technology, Inc. Defined data patterns for object handling
CN104809142A (zh) * 2014-01-29 2015-07-29 北京瑞天科技有限公司 商标查询系统和方法

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS60200376A (ja) 1984-03-26 1985-10-09 Hitachi Ltd 部分パタ−ンマツチング方式
US5524182A (en) * 1994-12-21 1996-06-04 Hewlett-Packard Company System and method for compressing and decompressing fonts based upon font stroke regularities
US6134340A (en) * 1997-12-22 2000-10-17 Trw Inc. Fingerprint feature correlator
JP4332649B2 (ja) * 1999-06-08 2009-09-16 独立行政法人情報通信研究機構 手の形状と姿勢の認識装置および手の形状と姿勢の認識方法並びに当該方法を実施するプログラムを記録した記録媒体
US6801661B1 (en) * 2001-02-15 2004-10-05 Eastman Kodak Company Method and system for archival and retrieval of images based on the shape properties of identified segments
US7496228B2 (en) * 2003-06-13 2009-02-24 Landwehr Val R Method and system for detecting and classifying objects in images, such as insects and other arthropods
US20060045312A1 (en) * 2004-08-27 2006-03-02 Bernstein Daniel B Image comparison device for providing real-time feedback
US7646918B2 (en) * 2005-09-27 2010-01-12 Wisconsin Alumni Research Foundation Systems and methods for recognizing objects in an image
US8441440B2 (en) * 2005-09-27 2013-05-14 Tamura Corporation Position information detection device, position information detection method, and position information detection program
JP4738469B2 (ja) * 2008-10-29 2011-08-03 株式会社東芝 画像処理装置、画像処理プログラムおよび画像処理方法
JP2010237598A (ja) * 2009-03-31 2010-10-21 Fujitsu Semiconductor Ltd データ検証方法、データ検証装置、およびデータ検証プログラム
US8892594B1 (en) 2010-06-28 2014-11-18 Open Invention Network, Llc System and method for search with the aid of images associated with product categories
CN102541954B (zh) * 2010-12-29 2014-11-05 北京大学 一种商标检索方法及系统
US9269023B2 (en) 2012-05-31 2016-02-23 Sharp Laboratories Of America, Inc. Edge based location feature index matching
WO2014000261A1 (zh) * 2012-06-29 2014-01-03 中国科学院自动化研究所 一种基于空间连通域预定位的商标检测方法
CN103258037A (zh) * 2013-05-16 2013-08-21 西安工业大学 一种针对多组合内容的商标识别检索方法
CN104680393A (zh) * 2013-12-02 2015-06-03 章文贤 一种基于图像内容及匹配的交互广告方法
CN104462380A (zh) * 2014-12-11 2015-03-25 北京中细软移动互联科技有限公司 商标检索方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101464951A (zh) * 2007-12-21 2009-06-24 北大方正集团有限公司 图像识别方法及系统
CN101894154A (zh) * 2010-07-13 2010-11-24 重庆大学 一种图像文档关键图案的提取方法
US9058543B2 (en) * 2010-11-01 2015-06-16 Raf Technology, Inc. Defined data patterns for object handling
CN104809142A (zh) * 2014-01-29 2015-07-29 北京瑞天科技有限公司 商标查询系统和方法
CN104156413A (zh) * 2014-07-30 2014-11-19 中国科学院自动化研究所 一种基于商标密度的个性化商标匹配识别方法
CN104199931A (zh) * 2014-09-04 2014-12-10 厦门大学 一种商标图像一致语义提取方法及商标检索方法
CN104298990A (zh) * 2014-09-15 2015-01-21 西安电子科技大学 一种基于骨架图的快速图形匹配与识别的方法

Also Published As

Publication number Publication date
CN105574161A (zh) 2016-05-11
GB2561773A (en) 2018-10-24
GB201811592D0 (en) 2018-08-29
US10430687B2 (en) 2019-10-01
CN105574161B (zh) 2017-09-26
US20180373960A1 (en) 2018-12-27

Similar Documents

Publication Publication Date Title
WO2017101224A1 (zh) 一种商标检索方法、装置、系统以及计算机存储介质
WO2017101225A1 (zh) 一种商标图形要素识别方法、装置、系统以及计算机存储介质
CN107145829B (zh) 一种融合纹理特征和尺度不变特征的掌静脉识别方法
Dibeklioglu et al. 3D facial landmarking under expression, pose, and occlusion variations
CN108829711B (zh) 一种基于多特征融合的图像检索方法
Roy et al. Script identification from handwritten document
CN110738216A (zh) 基于改进surf算法的药品识别方法
CN103839042A (zh) 人脸识别方法和人脸识别系统
Rane et al. Chartreader: Automatic parsing of bar-plots
WO2022148396A1 (zh) 芯片采集方法和芯片定位方法
CN114581646A (zh) 文本识别方法、装置、电子设备及存储介质
Musicco et al. Automatic point cloud segmentation for the detection of alterations on historical buildings through an unsupervised and clustering-based Machine Learning approach
CN117474929A (zh) 基于机器视觉的托盘外形尺寸检测方法及系统
CN114776304A (zh) 一种深海矿产异常区识别的方法、装置、设备及介质
CN110378337B (zh) 金属切削刀具图纸标识信息视觉输入方法及系统
CN115169375B (zh) 基于ar与枪球联动的高位物料可视化方法
CN104504385A (zh) 手写粘连数字串的识别方法
Mishchenko et al. Model-Based Recognition and Extraction of Information from Chart Images.
Souza et al. MICR automated recognition based on paraconsistent artificial neural networks
CN112990064A (zh) 一种基于颜色分割与概率模型的表盘指针识别方法
RP et al. Image feature extraction and recognition of abstractionism and realism style of indonesian paintings
CN116452587B (zh) 一种环保建筑结构钢板缺陷识别方法
CN113255657B (zh) 票据表面刮痕检测方法、装置、电子设备、机器可读介质
CN117119253B (zh) 一种针对目标对象的高质量视频抽帧方法
Qu et al. Eliminating lining seams in tunnel concrete crack images via line Segments’ translation and expansion

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16874291

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 201811592

Country of ref document: GB

Kind code of ref document: A

Free format text: PCT FILING DATE = 20160308

WWE Wipo information: entry into national phase

Ref document number: 1811592.3

Country of ref document: GB

122 Ep: pct application non-entry in european phase

Ref document number: 16874291

Country of ref document: EP

Kind code of ref document: A1