Method for identifying and searching oracle font matching based on graph structure dictionary

Abstract

The invention discloses a graph structure dictionary-based oracle font matching recognition and search method, which comprises the following steps of: an oracle skeleton and key point extraction stage: refining an oracle image to extract an oracle skeleton, extracting key points such as end points and cross points according to the skeleton, and adopting a neighbor key point elimination method to extract key points of the oracle; redundant key points which are too close to each other are eliminated; an oracle-bone text font structure extraction stage: according to connectivity between every two key points, generating a key point connectivity relation adjacency matrix as a structural feature of an oracle-bone text font, and generating a graph structure of the oracle-bone text font; and an oracle font structure matching recognition and search stage: establishing an oracle font graph structure dictionary, performing oracle font graph structure matching recognition and search, and removing different oracle characters with matched font structures. The oracle font matching recognition and search method based on the graph structure dictionary is large in font graph structure scale, high in font structure matching speed and high in recognition accuracy.

Description

Technical Field

[0001] The present invention relates to the technical fields of computer vision, pattern recognition and digital image processing, and in particular to an oracle bone script glyph matching recognition and search method based on a graph structure dictionary. Background Art

[0002] With the advent of the intelligent era, the intelligent processing and recognition technology of oracle bone script has been widely used. The full-character structure matching and recognition of the oracle bone script system has important application value in the fields of oracle bone script matching and recognition, natural language processing, etc. The recognition of oracle bone script isomorphism and similar structure is of great significance to the evolution of writing history and the deciphering of oracle bone script. At present, the oracle bone script structure matching and recognition has the disadvantages of small data scale, slow speed and low accuracy. Summary of the invention

[0003] 1. Technical issues to be resolved

[0004] In view of the deficiencies in the prior art, the present invention provides an oracle bone script glyph matching recognition and search method based on a graph structure dictionary, which solves the problems mentioned in the above background technology.

[0005] (II) Technical solution

[0006] To achieve the above objectives, the present invention is implemented by the following technical scheme: a method for matching, identifying and searching oracle bone script based on a graph structure dictionary, comprising the following steps:

[0007] S1, Oracle-bone script skeleton and key point extraction stage: It is proposed to extract the oracle-bone script skeleton after binarization of the oracle-bone script image, and extract key points based on the skeleton image, including endpoints and intersections, and use the nearest key point elimination method to merge redundant intersections that are too close;

[0008] S2, oracle bone script glyph structure extraction stage: according to the connectivity between the key points of the oracle bone script skeleton without passing through other intersections, an adjacency matrix of the key point connectivity relationship is generated, and the adjacency matrix is ​​used to store the adjacency relationship between all key points as the structural features of the oracle bone script glyph, and the graph structure of the oracle bone script glyph is generated;

[0009] S3, oracle bone script shape structure matching recognition and search stage: This stage is divided into three parts: first, establishing an oracle bone script shape graph structure dictionary; second, oracle bone script shape graph structure matching recognition and search method; third, eliminating different oracle bone scripts that have shape structure matching methods.

[0010] Preferably, the method for eliminating neighbor key points specifically includes the following steps:

[0011] Calculate the coordinates of the midpoint of the two intersections, take the midpoint as the center of the circle, and half the distance between the two intersections as the radius, draw a white solid circle on the binary image of the oracle bone script, and determine whether the total number of points with a pixel value of 0 before and after drawing the circle image changes. If it does not change, merge them into one intersection, otherwise do nothing. On this basis, calculate the distance between the two intersections, and use the circle diameter threshold method to determine whether the two intersections should be merged.

[0012] Preferably, the step of establishing the oracle bone script glyph structure dictionary specifically includes the following steps:

[0013] According to the graph structure method for generating oracle bone script glyphs described in S2, a complete graph structure of each oracle bone script glyph in the entire glyph library of the oracle bone script system is generated, and an oracle bone script glyph complete graph structure dictionary is established; according to the weakly connected graph structure of each oracle bone script glyph, an oracle bone script glyph weakly connected graph structure dictionary is established; according to the generalized subgraph structure of each oracle bone script glyph, an oracle bone script glyph generalized subgraph structure dictionary is established; according to the component graph structure of each oracle bone script glyph, an oracle bone script glyph component graph structure dictionary is established. In the graph structure dictionary, the graph structure maps a set of oracle bone script characters with the same graph structure.

[0014] Preferably, the oracle bone script complete graph structure matching and recognition method specifically comprises the following steps:

[0015] The oracle bone script to be recognized uses the adjacency matrix described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph to form a complete graph structure of the oracle bone script, which is matched and recognized with the established oracle bone script complete graph structure dictionary.

[0016] Preferably, the oracle bone inscriptions weakly connected graph structure matching and recognition method specifically comprises the following steps:

[0017] The oracle bone script to be recognized uses the adjacency matrix described in S2 to store the adjacency relationship between all key points, converts the adjacency matrix into a connected graph, and then divides the connected graph into several weakly connected graphs, which are matched and recognized with the established oracle bone script weakly connected graph structure dictionary.

[0018] Preferably, the oracle bone script generalized subgraph structure matching and recognition method specifically comprises the following steps:

[0019] The oracle bone script to be recognized is stored in an adjacency matrix as described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph. Several connected edges in the connected graph are taken to form several new connected graphs, which are matched and recognized with the established oracle bone script generalized subgraph structure dictionary.

[0020] Preferably, the oracle bone script component graph structure matching recognition and search method comprises the following steps:

[0021] Generate the graph structure of the oracle bone script to be recognized according to S2, split the oracle bone script according to its components, generate the graph structure of each component, and search for the oracle bone script that matches the graph structure of the oracle bone script component in the oracle bone script component graph structure dictionary according to the graph structure of the oracle bone script component.

[0022] Preferably, the method for eliminating different oracle bone inscriptions with matching glyph structures comprises the following steps:

[0023] After searching for two oracle bone inscriptions with matching glyph structures, the homography matrix between the two oracle bone inscription images is calculated based on the key points of the oracle bone inscription glyph structure matching. The degree of matching between the point coordinates is calculated through the orientation and distance between the points, which is used to eliminate unmatched point coordinates and text.

[0024] (III) Beneficial effects

[0025] The present invention provides an oracle bone script glyph matching recognition and search method based on a graph structure dictionary. Compared with the prior art, it has the following beneficial effects:

[0026] (1) The present invention has a high recognition accuracy rate: the matching recognition method of the glyph graph structure in the oracle bone script system is provided, which eliminates the redundant intersections with close distances in the oracle bone script glyph skeleton, and the oracle bone script glyph structure matching and positioning are accurate, so the recognition classification of the present invention is more specific.

[0027] (2) The data scale of the present invention is huge: the proposed oracle bone script glyph structure dictionary includes comprehensive glyph data in the large-scale oracle bone script system and collects comprehensive and concise oracle bone script glyph structure. Therefore, the search data scale of the present invention is large and the accuracy is high. The area under the test operation characteristic curve is 0.9807.

[0028] (3) The matching rate of the present invention is very fast: in the large-scale oracle bone script structure matching and recognition, according to the radical and other related graph structures of the oracle bone script, a strategy of establishing a dictionary to only search for the graph structures with related graph structures is reduced, so the search efficiency of the present invention is high and the real-time performance is good. The average time taken to search a single glyph is 0.6039 seconds.

[0029] (4) The method of the present invention has strong adaptability: it is suitable for the recognition of the glyph structure of characters such as oracle bone inscriptions, bronze inscriptions, large seal script, small seal script, official script, traditional Chinese characters and simplified Chinese characters, and is also suitable for the matching recognition and search of structures such as characters with topological structures, simple drawings, proteins and buildings, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a structural block diagram of an oracle bone script shape matching recognition and search method based on a graph structure dictionary of the present invention;

[0031] Figure 2 It is a schematic diagram of skeleton, contour and key point extraction in the present invention;

[0032] Figure 3 Schematic diagram of the method for eliminating neighboring key points in the present invention;

[0033] Figure 4 A schematic diagram of the oracle bone script glyph structure generated in the present invention;

[0034] Figure 5 A schematic diagram of the oracle bone script glyph structure dictionary established in the present invention;

[0035] Figure 6 It is a schematic diagram of the complete graph structure matching and recognition method of oracle bone script in the present invention;

[0036] Figure 7 It is a schematic diagram of the oracle bone script weakly connected graph structure matching and recognition method in the present invention;

[0037] Figure 8 It is a schematic diagram of the oracle bone script generalized subgraph structure matching and recognition method in the present invention;

[0038] Fig. 9 A schematic diagram of the oracle bone script component graph structure matching and recognition method of the present invention;

[0039] Fig.10 Schematic diagram of different oracle bone script methods for eliminating glyph structure matching in the present invention.

[0040] Fig.11 Schematic diagram of program operation in the present invention DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Example

[0043] A method for matching and recognizing oracle bone script based on graph structure dictionary, such as Figure 1 As shown, the following steps are included:

[0044] S1, Oracle-bone script skeleton and key point extraction stage: It is proposed to extract the oracle-bone script skeleton after binarization of the oracle-bone script image, and extract key points based on the skeleton image, including endpoints and intersections. The nearest key point elimination method is used to merge the redundant intersections that are too close;

[0045] S2, oracle bone script glyph structure extraction stage: according to the connectivity between the key points of the oracle bone script skeleton without passing through other intersections, an adjacency matrix of the key point connectivity relationship is generated, and the adjacency matrix is ​​used to store the adjacency relationship between all key points as the structural features of the oracle bone script glyph, and the graph structure of the oracle bone script glyph is generated;

[0046] S3, oracle bone script shape structure matching recognition and search stage: This stage is divided into three parts: first, establishing an oracle bone script shape graph structure dictionary; second, oracle bone script shape graph structure matching recognition and search method; third, eliminating different oracle bone scripts that have shape structure matching methods.

[0047] The method for eliminating neighboring key points in step S1 comprises the following steps:

[0048] Calculate the coordinates of the midpoint of the two intersections, take the midpoint as the center of the circle, and half the distance between the two intersections as the radius, draw a white solid circle on the binary image of the oracle bone script, and determine whether the total number of points with a pixel value of 0 before and after drawing the circle image changes. If it does not change, merge them into one intersection, otherwise do nothing. On this basis, calculate the distance between the two intersections, and use the circle diameter threshold method to determine whether the two intersections should be merged.

[0049] The step S3 of establishing the oracle bone script glyph structure dictionary includes the following steps:

[0050] According to the graph structure method for generating oracle bone script glyphs described in S2, a complete graph structure of each oracle bone script glyph in the entire glyph library of the oracle bone script system is generated, and an oracle bone script glyph complete graph structure dictionary is established; according to the weakly connected graph structure of each oracle bone script glyph, an oracle bone script glyph weakly connected graph structure dictionary is established; according to the generalized subgraph structure of each oracle bone script glyph, an oracle bone script glyph generalized subgraph structure dictionary is established; according to the component graph structure of each oracle bone script glyph, an oracle bone script glyph component graph structure dictionary is established. In the graph structure dictionary, the graph structure maps a set of oracle bone script characters with the same graph structure.

[0051] The oracle bone script complete graph structure matching and recognition method in step S3 comprises the following steps:

[0052] The oracle bone script to be recognized is stored with the adjacency matrix described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph, which is matched and recognized with the established oracle bone script complete graph structure dictionary.

[0053] The oracle bone script weakly connected graph structure matching and recognition method described in step S3 comprises the following steps:

[0054] The oracle bone script to be recognized uses the adjacency matrix described in S2 to store the adjacency relationship between all key points, converts the adjacency matrix into a connected graph, and then divides the connected graph into several weakly connected graphs, which are matched and recognized with the established oracle bone script weakly connected graph structure dictionary.

[0055] The oracle bone script generalized subgraph structure matching and recognition method in step S3 comprises the following steps:

[0056] The oracle bone script to be recognized is stored in an adjacency matrix as described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph. Several connected edges in the connected graph are taken to form several new connected graphs, which are matched and recognized with the established oracle bone script generalized subgraph structure dictionary.

[0057] The oracle bone script glyph component graph structure matching recognition and search method in step S3 comprises the following steps:

[0058] Generate the graph structure of the oracle bone script to be recognized according to S2, split the oracle bone script according to its components, generate the graph structure of each component, and search for the oracle bone script that matches the graph structure of the oracle bone script component in the oracle bone script component graph structure dictionary according to the component graph structure of the oracle bone script.

[0059] The method for eliminating different oracle bone inscriptions with matching glyph structures in step S3 comprises the following steps:

[0060] After searching for two oracle bone inscriptions with matching glyph structures, the homography matrix between the two oracle bone inscription images is calculated based on the key points of the oracle bone inscription glyph structure matching. The degree of matching between the point coordinates is calculated through the orientation and distance between the points, which is used to eliminate unmatched point coordinates and text.

[0061] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0062] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0063] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for matching, identifying and searching oracle bone script characters based on a graph structure dictionary, characterized in that: The following steps are involved: S1, Oracle-bone script skeleton and key point extraction stage: It is proposed to extract the skeleton of the oracle-bone script after binarization of the oracle-bone script image, and extract key points based on the skeleton image, including endpoints and intersections, and use the nearest key point elimination method to merge redundant key points that are too close; S2, oracle bone script glyph structure extraction stage: according to the connectivity between the key points of the oracle bone script skeleton without passing through other intersections, an adjacency matrix of the key point connectivity relationship is generated, and the adjacency matrix is ​​used to store the adjacency relationship between all key points as the structural features of the oracle bone script glyph, and the graph structure of the oracle bone script glyph is generated; S3, oracle bone script shape structure matching recognition and search stage: This stage is divided into three parts: first, establishing an oracle bone script shape graph structure dictionary; second, oracle bone script shape graph structure matching recognition and search method; third, eliminating different oracle bone scripts that have shape structure matching methods.

2. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1 is characterized by: The method for eliminating neighbor key points specifically comprises the following steps: Calculate the coordinates of the midpoint of the two intersections, take the midpoint as the center of the circle, and half the distance between the two intersections as the radius, draw a solid white circle on the binary image of the oracle bone script, and determine whether the total number of points with a pixel value of 0 before and after drawing the circle image changes. If it does not change, merge them into one intersection, otherwise do nothing. On this basis, calculate the distance between the two intersections, and use the circle diameter threshold method to determine whether redundant intersections should be merged.

3. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1, characterized in that: The establishment of the oracle bone script glyph structure dictionary is specifically The following steps are involved: According to the graph structure method for generating oracle bone script glyphs described in S2, a complete graph structure of each oracle bone script glyph in the entire glyph library of the oracle bone script system is generated, and a dictionary of complete graph structures of oracle bone script glyphs is established; according to the weakly connected graph structure of each oracle bone script glyph, a dictionary of weakly connected graph structures of oracle bone script glyphs is established; according to the generalized subgraph structure of each oracle bone script glyph, a dictionary of generalized subgraph structures of oracle bone script glyphs is established; according to the component graph structure of each oracle bone script glyph, a dictionary of component graph structures of oracle bone script glyphs is established. In the oracle-bone script graph structure dictionary, the graph structure maps a collection of oracle-bone script characters with the same graph structure.

4. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1, characterized in that: The method for matching and identifying the complete graph structure of oracle bone script has the following steps: The oracle bone script to be recognized uses the adjacency matrix described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph to form a complete graph structure of the oracle bone script, which is matched and recognized with the established complete graph structure dictionary of the oracle bone script.

5. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1, characterized in that: The oracle bone script weakly connected graph structure matching and recognition method comprises the following steps: The oracle bone script to be recognized uses the adjacency matrix described in S2 to store the adjacency relationship between all key points, converts the adjacency matrix into a connected graph, and then divides the connected graph into several weakly connected graphs to form an oracle bone script weakly connected graph structure, which is matched and recognized with the established oracle bone script weakly connected graph structure dictionary.

6. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1, characterized in that: The oracle bone script generalized subgraph structure matching and recognition method has the following steps: The oracle bone script to be recognized is stored in an adjacency matrix as described in S2 to store the adjacency relationship between all key points, and the adjacency matrix is ​​converted into a connected graph. Several connected edges in the connected graph are taken to obtain a new connected graph to form an oracle bone script glyph generalized subgraph structure, which is matched and recognized with the established oracle bone script glyph generalized subgraph structure dictionary.

7. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1 is characterized by: The oracle bone script glyph component graph structure matching and recognition method specifically comprises the following steps: The oracle bone script glyph graph structure to be recognized is generated according to S2, the oracle bone script glyph is split into components, and the graph structure of each component is generated, which is matched and recognized with the established oracle bone script glyph component graph structure dictionary.

8. The oracle bone script matching recognition and search method based on a graph structure dictionary according to claim 1, characterized in that: The specific method includes the following steps: After searching for two oracle bone inscriptions with matching glyph structures, the homography matrix between the two oracle bone inscription images is calculated based on the key points of the oracle bone inscription glyph structure matching. The degree of matching between the point coordinates is calculated through the orientation and distance between the points, which is used to eliminate unmatched point coordinates and text.

Images