CN111292283B - Carapace bone fragment conjugation method based on time sequence similarity calculation - Google Patents

Carapace bone fragment conjugation method based on time sequence similarity calculation Download PDF

Info

Publication number
CN111292283B
CN111292283B CN202010072172.5A CN202010072172A CN111292283B CN 111292283 B CN111292283 B CN 111292283B CN 202010072172 A CN202010072172 A CN 202010072172A CN 111292283 B CN111292283 B CN 111292283B
Authority
CN
China
Prior art keywords
edge curve
array
data
oracle bone
curve data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010072172.5A
Other languages
Chinese (zh)
Other versions
CN111292283A (en
Inventor
张重生
纵瑞星
凡高娟
史先进
莫伯峰
门艺
沈夏炯
夏瑞雪
余波
郑逢斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Henan University
Original Assignee
Henan University
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 Henan University filed Critical Henan University
Priority to CN202010072172.5A priority Critical patent/CN111292283B/en
Publication of CN111292283A publication Critical patent/CN111292283A/en
Application granted granted Critical
Publication of CN111292283B publication Critical patent/CN111292283B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/60—Rotation of whole images or parts thereof
    • G06T3/608—Rotation of whole images or parts thereof by skew deformation, e.g. two-pass or three-pass rotation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/181—Segmentation; Edge detection involving edge growing; involving edge linking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a carapace bone fragment conjugation method based on time sequence similarity calculation, which comprises the following steps: a: extracting an edge curve graph of each oracle bone rubbing image; b: obtaining time-series edge curve data of each oracle bone rubbing image; c: obtaining normalized time-series edge curve data T'; d, carrying out tilt correction on the oracle bone rubbing image to obtain new edge curve data T a '' put into set TS; e: calculating T a '' and T b ' difference array set ALL; f: calculating T a And T b Similarity of (2); g: and selecting the oracle bone rubbing image corresponding to other ten edge curve data with the highest edge curve similarity as the optimal conjugation result of the oracle bone rubbing image.

Description

Carapace bone fragment conjugation method based on time sequence similarity calculation
Technical Field
The invention relates to a split article image stitching method, in particular to a carapace bone fragment stitching method based on time series similarity calculation.
Background
Oracle conjugation is a term of art in the study of oracle and means that two or more oracle bone rubbings are pieced together, and the conjugation of the corresponding oracle bone fragments is generally accomplished by piecing together two or more oracle bone rubbings images. Many scholars working on oracle inscriptions are working on oracle conjugation studies to restore broken oracle bone fragments. In thousands of years from the time of burying the nail chips in the ground to the time of excavation, a large number of nail chips are broken into two or more pieces due to the influence of factors such as low formation pressure and excavation destruction. The broken oracle bone fragments are incomplete in character information and not beneficial to study of oracle characters, so that the broken oracle bone fragments are completely spliced, the meaning of the oracle bone fragments can be comprehensively understood, and events recorded on the oracle bone fragments can be understood, so that splicing or conjugation of the oracle bone fragments has very important significance for study of the oracle characters.
As the oracle bone fragments are generally stored in museums all over the world and are not allowed to be touched, the conjugation work of the oracle bone fragments is carried out on the basis of the oracle bone rubbing images. In a common conjugation method of an oracle bone rubbing image, the existing conjugation method tries to combine two or more oracle bone rubbing images by means of artificial association and memory, firstly, whether edge stubbles of the oracle bone rubbing images to be conjugated are aligned is investigated, and then, whether two or more oracle bone rubbing images can be conjugated is further determined by combining context sentences on the combined oracle bone to be analyzed.
However, the number of the oracle bone slices is hundreds of thousands, so that the number of combinations among different oracle bone rubbing images is huge, and the problems of low efficiency and incomplete conjugation results exist by singly relying on manual conjugation.
Therefore, a calculation method capable of automatically conjugating the image of the oracle bone rubbings is urgently needed to improve the conjugation efficiency of the oracle bone rubbings. The calculation method comprises the steps of automatically selecting ten candidate oracle bone rubbing images with the highest conjugation degree for each oracle bone rubbing image, and then handing the ten candidate oracle bone rubbing images of each oracle bone rubbing image to an oracle bone expert for further confirmation in a manual screening mode. The calculation method can ensure the integrity of the results of the oracle bone conjugation, and can greatly reduce the number of candidate conjugation objects needing manual confirmation by oracle bone experts.
According to the invention, the edge stitching degree of the two oracle bone rubbing images is calculated and evaluated through the similarity of the time sequences, so that in the invention, the similarity of the two edge curves of the two oracle bone rubbing images after the time sequences is corresponding to the two oracle bone rubbing images can directly reflect the stitching degree of the two oracle bone rubbing images. The similarity calculation between two time series is also known in computer terminology. Therefore, in the detailed description of the present invention, the similarity terms are used collectively.
Disclosure of Invention
The invention aims to provide a carapace bone fragment conjugation method based on time sequence similarity calculation, which can quickly and accurately judge whether two carapace bone rubbing images can be conjugated or not.
The invention adopts the following technical scheme:
a method of oracle bone fragment conjugation based on time series similarity calculation, comprising the steps of:
a: extracting an edge curve graph of each oracle bone rubbing image;
b: on an edge curve graph of each oracle bone rubbing image, sequentially reading the height value of the uppermost pixel point and the height value of the lowermost pixel point of each row of pixel data from left to right, obtaining an average value, taking the average value as the pixel position of an edge curve in the row of pixel data, and sequentially combining the pixel positions of each row of pixel data obtained in sequence to form time-series curve data T, T = { V } corresponding to the edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the pixel data of the ith column of the edge curve is represented; according to the method, time-series edge curve data of each oracle bone rubbing image are sequentially obtained;
c: corresponding time of each oracle bone rubbing imageNormalizing the serialized edge curve data T to obtain normalized time-serialized edge curve data T ', T' = { V = 1 ',V 2 ',V 3 '，...，V i ' }, i is a positive integer;
d, normalizing time-series edge curve data T corresponding to two oracle bone rubbing images a and b to be judged whether to be conjugated a ' and T b ', let T a ' Length is greater than T b ', data T of shorter curve b ' stride at T by 1 pixel a ' slide in sequence and at T b ' as reference basis T a ' corresponding oracle bone rubbing image a is subjected to inclination correction, and new edge curve data T is extracted from the oracle bone rubbing image a subjected to inclination correction every time a ", and will T a "put into the set TS;
e: for each edge curve T in the set TS obtained in the step D a ", calculating T a And T b ' difference array set ALL; then entering step F;
f: calculating T according to each difference array in the difference array set ALL obtained in the step E a And T b The similarity of (2); then entering step G;
g: according to the method of the step C, the step D, the step E and the step F in sequence, carrying out similarity calculation on the normalized time-series edge curve data corresponding to each oracle bone rubbing image and the normalized time-series edge curve data corresponding to each oracle bone rubbing image to be selected in sequence to obtain corresponding similarity values respectively; and finally, selecting the oracle bone rubbing image corresponding to other ten pieces of edge curve data with the highest edge curve similarity of each oracle bone rubbing image as the optimal conjugation result of the oracle bone rubbing image.
In the step A, a digital board and image editing processing software are utilized to carry out manual edge tracing processing on the upper edge or the lower edge of each oracle bone rubbing image by using a red painting brush with the thickness of 2 pixels, the image after edge tracing processing is stored, and the edge of each obtained oracle bone rubbing image is represented by a red curve; and then, automatically extracting all pixel points with the pixel value of (255,0,0) in each oracle bone rubbing image by using the color characteristics to obtain a red edge curve graph of each oracle bone rubbing image.
In the step B, on a red edge curve graph of each oracle bone rubbing image, sequentially reading the height value of the uppermost red pixel and the height value of the lowermost red pixel of each row of red pixel data from left to right and obtaining an average value, taking the average value as the pixel position of an edge curve on the row of red pixel data, and sequentially combining the pixel positions of each row of red pixel data obtained sequentially to form time-series curve data T, T = { V } corresponding to the red edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the red pixel data of the ith column of the edge curve is represented; and sequentially obtaining the time-series edge curve data T of each oracle bone rubbing image according to the method.
When normalization processing is performed in the step C, the minimum value min (T) in the time-series curve data T is first calculated, and then min (T) is subtracted from each data in the corresponding time-series curve data T, so as to obtain normalized time-series edge curve data T' of the edge curve graph.
The step D comprises the following specific steps:
d1: normalizing the normalized time-series edge curve data T corresponding to the two edge curve graphs to be compared a ' and T b The head and tail end points of the' are directly connected, so that the lengths of the formed line segments are respectively L a And L b (ii) a If L is a Is equal to L b Entering the step D2; otherwise, with L a And L b The larger median edge curve data is used as the reference, assuming L a Greater than L b Entering step D4;
d2: respectively calculating two normalized time-series edge curve data T a ' and T b Angle of inclination of `, let θ a Is T a Angle of inclination, θ b Is T b The angle of inclination of'; let T a ' head and tail end points P 1 And P 2 The coordinates in the image are respectively (S) x ,S y ) And (E) x ,E y ) Then theta a ＝arctan((E y -S y )/(E x -S x ) Arctan is an arctangent function; in the same way, according to T b ' head and tail end points P 3 And P 4 Coordinates in the image, calculating T b Angle of inclination theta of ` b (ii) a Finally, calculate T a ' and T b The difference of the tilt angle of ' θ ', θ ' = θ a -θ b Then enter D3;
d3: if θ' is greater than 0, then T is added a 'clockwise rotation of the corresponding oracle bone rubbing image by an angle theta'; if θ' is less than 0, then T a The corresponding oracle bone rubbing image rotates anticlockwise by an angle of theta ' |, and the theta ' | is the absolute value of theta '; for T a ' corresponding rotated oracle bone rubbing image, extracting new edge curve graph and normalized time-series edge curve data T sequentially through step B and step C a ", and new normalized time-sequenced edge curve data T a Putting in a set TS, and then entering the step E;
d4: let P 5 As a starting point, at an initial time P 5 Is equal to T a ' head end P 1 At T a ' from the starting point P 5 At the beginning, in the order of T from front to back a ' middle search Point P 6 Up to P 6 And P 5 Is equal to L b (ii) a Let T a ' in P 5 And P 6 Between the subplot data is T sub Calculating T using the method in step D2 sub And T b 'inclination angle difference θ', then using the method in step D3 for T a Rotating the corresponding oracle bone rubbing image, extracting a new edge curve graph from the rotated oracle bone rubbing image, and obtaining new normalized time-series edge curve data T a ", will T a "put into the set TS; then entering step D5;
d5: taking 1 pixel as a step, and dividing P 5 At T a ' slide backward in, repeat step D4Obtaining new normalized time-series curve data T a ", and will T a "put into the set TS; then entering step D6;
d6: repeating step D5 until P 6 Is equal to T a ' end point, finally obtaining T a Rotating the corresponding oracle bone rubbing image according to different angles and re-extracting a set TS formed by normalized time-series edge curve data; then step E is entered.
The step E comprises the following specific steps:
e1: reading curve data T from set TS a ", if T a And T b If the array lengths are consistent, entering step E2; if T a And T b If the array lengths are not consistent, step E3 is performed;
e2: calculating to obtain T a And T b The data difference values of the two curve data at each corresponding position form a difference value array d in sequence, minv is made to be equal to the minimum value in the difference value array d, then minv is subtracted from the value of each element in the difference value array d, finally the difference value array d is placed into a difference value array set ALL, and then the step E6 is carried out;
suppose that the two curve data sequences are T respectively a ”＝{T a1 ”,T a2 ”,T a3 ”，...，T ai "} and T b '＝{T b1 ',T b2 ',T b3 ',...，T bi ' }; calculating the difference value of the two curve data sequences to obtain a difference value array d, d = { | T a1 ”-T b1 '|,|T a2 ”-T b2 '|,|T a3 ”-T b3 '|,...，|T ai ”-T bi ' | }; let minv equal the minimum in the difference array d, and finally d = { | T a1 ”-T b1 '|-minv,|T a2 ”-T b2 '|-minv,|T a3 ”-T b3 '|-minv,...|T ai ”-T bi '|-minv}；
E3: let P 7 As a starting point, at an initial time P 7 Is equal to T a "the head end point; at T a "from the starting point P 7 At the beginning, T b ' array head end position and P 7 Align, calculate T a And T b ' data difference value for each corresponding position, up to T b The data in the array are all calculated to form a difference array d ', the obtained data differences are made into a difference array d ' in sequence, minv ' is made to be equal to the minimum value in the difference array d ', then minv ' is subtracted from the value of each element in the difference array d ', then the difference array d ' is put into a set S, and then the step E4 is carried out;
e4: taking 1 pixel as a step, and dividing P 7 At T a "slide backward in the middle, repeat step E3, and let T b ' array head end position and P 7 Align and calculate T a And T b 'the difference array d' is made to be equal to the minimum value in the difference array d ', then minv' is subtracted from the value of each element in the difference array d ', then the difference array d' is put into the set S, and then the step E5 is carried out;
e5: repeating step E4 until T a "end of array and T b ' the current positions of the tail end points of the array are the same, and finally T is obtained a And T b Putting S into a set ALL, and then entering a step E6;
e6: repeating the steps E1 to E5, and calculating the next curve data T in the set TS a And T b ' the new difference array set S is put into the set ALL until ALL the edge curve data in the TS are traversed; finally obtaining each edge curve data T in TS a And T b ' and then proceeds to step F.
In the step F, T is calculated by a maximum data difference method a And T b The specific calculation method of the maximum similarity H is as follows:
f11: for each difference array in the difference array set ALL obtained in the step E, calculating the maximum value m of the array 1 And m is 1 Put into the set M, let M a Equal to the minimum of M;
F12：using formulas
Figure BDA0002377570180000071
Calculating to obtain T a And T b The similarity H of (A); wherein m is a The smaller the H, the closer to 1,m it is a The larger, the closer H is to 0.
According to the invention, the oracle fragment edge is rotated, the inclination correction and normalization processing are carried out to obtain time series edge curve data, different oracle fragment combinations are graded through a time series similarity calculation method, and finally ten oracle fragment rubbing images which are optimally matched with each oracle fragment image are obtained, so that the conjugation of the oracle fragments can be simply, efficiently, quickly and accurately realized, a large amount of time cost and labor cost can be saved, and the method has positive significance for the research of oracle characters.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a graph of red edges of two oracle bone rubbing images;
FIG. 3 is a schematic diagram of a pair of oracle bone rubbing images with higher similarity of conjugation;
fig. 4 is a schematic diagram of a pair of oracle bone rubbing images with lower degrees of similarity of conjugation.
Detailed Description
The invention is described in detail below with reference to the following figures and examples:
as shown in fig. 1 to 4, the method for conjugating a thyroid fragment based on time series similarity calculation according to the present invention comprises the following steps:
a: extracting an edge curve graph of each oracle bone rubbing image;
in the invention, the edge curve graph of the oracle bone rubbing image can be extracted by various existing image processing modes; in the embodiment, an oracle expert uses a digitizer and image editing processing software to perform manual edge tracing processing on the upper edge or the lower edge of each oracle rubbing image by using a red painting brush with the thickness of 2 pixels, the image after edge tracing processing is stored, and the edge of each obtained oracle rubbing image is represented by a red curve; then, automatically extracting all pixel points with the pixel value of (255,0,0) in each oracle bone rubbing image by utilizing the color characteristics to obtain a red edge curve graph of each oracle bone rubbing image;
b: on an edge curve graph of each oracle bone rubbing image, sequentially reading the height value of the uppermost pixel point and the height value of the lowermost pixel point of each row of pixel data from left to right, obtaining an average value, taking the average value as the pixel position of an edge curve in the row of pixel data, and sequentially combining the pixel positions of each row of pixel data obtained in sequence to form time-series curve data T, T = { V } corresponding to the edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the pixel data of the ith column of the edge curve is represented; according to the method, time-series edge curve data of each oracle bone rubbing image are sequentially obtained;
in this embodiment, on the red edge curve graph of each oracle bone rubbing image, the height value of the uppermost red pixel and the height value of the lowermost red pixel of each row of red pixel data are sequentially read from left to right and averaged, the average value is used as the pixel position of the edge curve on the row of red pixel data, and then the pixel positions of each row of red pixel data sequentially obtained are sequentially combined together to form time-series curve data T, T = { V } corresponding to the red edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the red pixel data of the ith column of the edge curve is represented; according to the method, time-series edge curve data T of each oracle bone rubbing image is obtained in sequence;
c: normalizing the time-series edge curve data T corresponding to each oracle bone rubbing image to obtain normalized time-series edge curve data T ', T' = { V = (V) } 1 ',V 2 ',V 3 '，...，V i ' }, i is a positive integer;
in this embodiment, when normalization processing is performed, first, a minimum value min (T) in the time-series curve data T is calculated, and then min (T) is subtracted from each data in the corresponding time-series curve data T to obtain normalized time-series edge curve data T' of the edge curve graph;
d, normalizing time-series edge curve data T corresponding to two oracle bone rubbing images a and b to be judged whether to be conjugated a ' and T b ', suppose T a ' Length is greater than T b ', the shorter curve data T b ' stride at T by 1 pixel a ' slide in sequence and at T b ' as reference basis T a ' corresponding oracle bone rubbing image a is subjected to inclination correction, and new edge curve data T is extracted from the oracle bone rubbing image a subjected to inclination correction every time a ", and will T a "put into the set TS;
the step D comprises the following specific steps:
d1: normalizing the normalized time-series edge curve data T corresponding to the two edge curve graphs to be compared a ' and T b The head and tail end points of the' are directly connected, the lengths of the formed line segments are respectively L a And L b (ii) a If L is a Is equal to L b Entering the step D2; otherwise, with L a And L b The edge curve data with larger middle value is taken as a benchmark, and L is assumed a Greater than L b Entering step D4;
d2: respectively calculating two normalized time-series edge curve data T a ' and T b Angle of inclination of `, let θ a Is T a Angle of inclination, θ b Is T b The angle of inclination of'. Let T a ' end point P 1 And P 2 The coordinates in the image are respectively (S) x ,S y ) And (E) x ,E y ) Then theta a ＝arctan((E y -S y )/(E x -S x ) Arctan is an arctangent function; in the same way, according to T b ' head and tail end points P 3 And P 4 Coordinates in the image, calculating T b Angle of inclination theta of ` b . Finally, calculate T a ' and T b The difference of the tilt angle of ' θ ', θ ' = θ a -θ b Then enter D3;
d3: if θ' is greater than 0, then T is added a 'clockwise rotation of the corresponding oracle bone rubbing image by an angle theta'; if θ' is less than 0, then T a ' the corresponding oracle bone rubbing image is rotated counterclockwise by an angle | θ ' |, which is the absolute value of θ '. For T a ' corresponding rotated oracle bone rubbing image, extracting new edge curve graph and normalized time-series edge curve data T sequentially through step B and step C a ", and new normalized time-sequenced edge curve data T a Putting in a set TS, and then entering the step E;
d4: let P 5 As a starting point, at an initial time P 5 Is equal to T a ' head end P 1 . At T a ' from the starting point P 5 At the beginning, in the order of T from front to back a ' center search Point P 6 Up to P 6 And P 5 Is equal to L b . Let T a ' middle P 5 And P 6 Between the subplot data is T sub Calculating T using the method in step D2 sub And T b ' Tilt Angle Difference θ ' of ' then for T using the method in D3 a Rotating the corresponding oracle bone rubbing image, extracting a new edge curve graph from the rotated oracle bone rubbing image, and obtaining new normalized time-series edge curve data T a ", will T a "put into the set TS; then entering step D5;
d5: taking 1 pixel as a step, and dividing P 5 At T a ' slide backward in the process, repeat step D4, get new normalized time-series curve data T a ", and will T a "put into the set TS; then entering step D6;
d6: repeating step D5 until P 6 Is equal to T a ' end point, finally obtaining T a Rotating the corresponding oracle bone rubbing image according to different angles and re-extracting a set TS formed by normalized time-series edge curve data; then step E is entered.
E: for the set obtained in step DEach edge curve T in the combined TS a ", calculating T a And T b ' difference array set ALL; then entering step F;
in this embodiment, normalized time-series edge curve data T of a group of edge curves to be compared is considered a "the length of the corresponding curve data sequence may not be consistent, so the present invention proceeds as follows:
the step E comprises the following specific steps:
e1: reading curve data T from set TS a ", if T a And T b If the array lengths of' are consistent, entering step E2; if T is a And T b If the array lengths are not consistent, step E3 is performed;
e2: calculating to obtain T a And T b The data difference values of the two curve data at each corresponding position form a difference value array d in sequence, minv is made to be equal to the minimum value in the difference value array d, then minv is subtracted from the value of each element in the difference value array d, finally the difference value array d is placed into a difference value array set ALL, and then the step E6 is carried out;
suppose that the two curve data sequences are T respectively a ”＝{T a1 ”,T a2 ”,T a3 ”，...，T ai "} and T b '＝{T b1 ',T b2 ',T b3 ',...，T bi ' }; calculating the difference value of the two curve data sequences to obtain a difference value array d, d = { | T a1 ”-T b1 '|,|T a2 ”-T b2 '|,|T a3 ”-T b3 '|,...，|T ai ”-T bi ' | }; let minv equal the minimum in the difference array d, and finally d = { | T a1 ”-T b1 '|-minv,|T a2 ”-T b2 '|-minv,|T a3 ”-T b3 '|-minv,...|T ai ”-T bi '|-minv}。
E3: let P 7 As a starting point, at an initial time P 7 Is equal to T a "the head end point. At T a "from the starting point P 7 At the beginning, T b Of the' arrayHead end position and P 7 Aligning, calculating T a And T b ' data difference value for each corresponding position, up to T b The data in the array are all calculated to form a difference array d ', the obtained data differences are made into a difference array d ' in sequence, minv ' is made to be equal to the minimum value in the difference array d ', then minv ' is subtracted from the value of each element in the difference array d ', then the difference array d ' is put into a set S, and then the step E4 is carried out;
e4: taking 1 pixel as a step, and dividing P 7 At T a "slide in the middle backward direction, repeat step E3, and let T be b ' array head end position and P 7 Align and calculate T a And T b 'the difference array d' is made to be equal to the minimum value in the difference array d ', then minv' is subtracted from the value of each element in the difference array d ', then the difference array d' is put into the set S, and then the step E5 is carried out;
e5: repeating step E4 until T a "end of array and T b ' the current positions of the tail end points of the array are the same, and finally T is obtained a And T b ' putting S into a set ALL, and then entering a step E6;
e6: repeating the steps E1 to E5, and calculating the next curve data T in the set TS a And T b ' the new difference array set S is put into the set ALL until ALL the edge curve data in the TS are traversed; finally obtaining each edge curve data T in TS a And T b ' and then proceeds to step F.
F: calculating T according to each difference array in the difference array set ALL obtained in the step E a And T b The similarity of (2); then entering step G;
in the invention, T is calculated by a maximum data difference method a And T b The specific calculation method of the maximum similarity H is as follows:
f11: for each difference array in the difference array set ALL obtained in the step E, calculating the maximum value m of the array 1 And m is 1 Put into set M. Let m a Equal to the minimum of M;
f12: using formulas
Figure BDA0002377570180000121
Calculating to obtain T a And T b The similarity H of (A); wherein m is a The smaller the H, the closer to 1,m it is a The larger, the closer H is to 0.
G: and D, according to the method of the step C, the step D, the step E and the step F in sequence, carrying out similarity calculation on the normalized time-series edge curve data corresponding to each oracle bone rubbing image and the normalized time-series edge curve data corresponding to each oracle bone rubbing image to be selected in sequence to obtain corresponding similarity values respectively. And finally, selecting the oracle bone rubbing image corresponding to other ten pieces of edge curve data with the highest edge curve similarity of each oracle bone rubbing image as the optimal conjugation result of the oracle bone rubbing image.
The following will further illustrate the method for the conjugation of oracle bone fragments based on the calculation of similarity between time series according to the present invention with reference to the following embodiments:
a: and (3) carrying out manual edge tracing processing on the upper edge or the lower edge of each oracle bone rubbing image by using a digital board and image editing processing software by an oracle expert, storing the image subjected to edge tracing processing, and finally representing the edge of each obtained oracle bone rubbing image by a red curve. Automatically extracting a red edge curve graph of each oracle bone rubbing image by using the color characteristics;
the used oracle bone fragment image is an oracle bone fragment image obtained by carrying out manual edge tracing processing on the edge of the oracle bone fragment image by an oracle bone expert, and is shown as a red edge curve graph of two oracle bone fragment images in an attached figure 2. And then, by utilizing the color characteristics, extracting all pixel points with the pixel value of (255,0,0) in the oracle bone rubbing image according to the sequence from left to right to obtain a red edge curve graph of each oracle bone rubbing image.
B: in each frameOn a red edge curve graph of a bone rubbing image, sequentially reading the height value of the topmost red pixel and the height value of the bottommost red pixel of each row of red pixel data from left to right, obtaining an average value, taking the average value as the pixel position of an edge curve on the row of red pixel data, and sequentially combining the pixel positions of each row of red pixel data obtained in sequence to form time-series curve data T corresponding to the red edge curve graph, wherein T = { V = (V) } V (V) is formed 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the red pixel data of the ith column of the edge curve is represented; according to the method, time-series edge curve data T of each oracle bone rubbing image is obtained in sequence;
c: normalizing the time-series edge curve data T corresponding to each oracle bone rubbing image to obtain normalized time-series edge curve data T ', T' = { V = (V) = 1 ',V 2 ',V 3 '，...，V i ' }, i is a positive integer;
in the present example, the curve data T of the normalized time series of the edge profile a of the oracle bone fragment to be matched is calculated a ' ＝ {V a1 ，V a2 ，V a3 ，...，V a254 And normalized time-series curve data T of the edge graph b of the oracle bone fragment to be selected b '＝{V b1 ，V b2 ，V b3 ，...，V b238 }；
D, normalizing time-series edge curve data T corresponding to two oracle bone rubbing images a and b to be judged whether to be conjugated a ' and T b ', let T a ' Length is greater than T b ', the shorter curve data T b ' stride at T by 1 pixel a ' slide in sequence and at T b ' as reference basis T a ' corresponding oracle bone rubbing image a is subjected to inclination correction, and new edge curve data T is extracted from the oracle bone rubbing image a subjected to inclination correction every time a ", and will T a "put into the set TS;
the step D comprises the following specific steps:
d1: normalizing the normalized time-series edge curve data T corresponding to the two edge curve graphs to be compared a ' and T b The head and tail end points of the' are directly connected, so that the lengths of the formed line segments are respectively L a And L b (ii) a If L is a Is equal to L b Entering the step D2; otherwise, with L a And L b The larger median edge curve data is used as the reference, assuming L a Greater than L b Entering step D4;
in the present example, the curve data T of the normalized time series of the edge profile of the oracle bone fragment to be matched is calculated a Line length L of direct connection line between head and tail points a =351, normalized time series of curve data T of edge graph of a candidate oracle bone fragment b Line length L of direct connection line between head and tail points b =318; entering the step D4;
d4: let P 5 As a starting point, at an initial time P 5 Is equal to T a ' head end P 1 . At T a ' from the starting point P 5 At the beginning, in the order of T from front to back a ' center search Point P 6 Up to P 6 And P 5 Is equal to L b . Let T a ' in P 5 And P 6 Between the subplot data is T sub Calculating T using the method in step D2 sub And T b ' Tilt Angle Difference θ ' of ' then for T using the method in D3 a ' rotating the corresponding oracle bone rubbing image, extracting a new edge curve graph from the rotated oracle bone rubbing image, and obtaining new normalized time-series edge curve data T a ", will T a "put into the set TS; then entering step D5;
in this example, T is calculated sub Is at an inclination angle theta a ＝41.68，T b The angle of inclination of b =21.95, so T sub And T b The angular difference θ '=17.92 of's. So will T a ' corresponding oracle bone rubbing image rotates clockwise by 17.92 degreesAnd to T a ' corresponding rotated oracle bone rubbing image, extracting new edge curve graph and normalized time-series curve data T sequentially through step B and step C a ", then T a "put into the set TS;
d5: taking 1 pixel as a step, and dividing P 5 At T a ' slide backward in the process, repeat step D4, get new normalized time series curve data T a ", and will T a "put into the set TS; then entering step D6;
d6: repeating step D5 until P 6 Is equal to T a ' end point, finally obtaining T a Rotating the corresponding oracle bone rubbing image according to different angles and re-extracting a set TS formed by normalized time-series edge curve data; then step E is entered.
E: for each edge curve T in the set TS obtained in the step D a ", calculating T a And T b ' difference array set ALL; then entering step F;
the step E comprises the following specific steps:
e1: reading curve data T from set TS a ", if T a And T b If the array lengths are consistent, entering step E2; if T a And T b If the array lengths are not consistent, step E3 is performed;
in this example, T a "has an array length of 254,T b ' has an array length of 238, so step E3 is entered;
e3: let P 7 As a starting point, at an initial time P 7 Is equal to T a "the head end point. At T a "from the starting point P 7 Starting with T b ' array head end position and P 7 Align, calculate T a And T b ' data difference value for each corresponding position, up to T b The data in the array are all calculated to form a difference array d ', minv' is equal to the minimum value in the difference array d ', and the value of each element in the difference array d' is calculatedSubtracting minv ', then putting the difference value array d' into the set S, and then entering the step E4;
in this example, T is calculated a And T b ' data difference values to corresponding positions, which constitute a difference array d ', d ' = { | T a1 ”-T b1 '|,|T a1 ”-T b2 '|,|T a3 ”-T b3 '|,...|T a238 ”-T b238 '| }, minv' equals the minimum value in the difference array d ', d' = { | T a1 ”-T b1 '|-minv',|T a2 ”-T b2 '|-minv',|T a3 ”-T b3 '|-minv',...|T a238 ”-T b238 '| minv' }; then entering step E4;
e4: taking 1 pixel as a step, and dividing P 7 At T a "slide in the middle backward direction, repeat step E3, and let T be b ' array head end position and P 7 Align and calculate T a And T b Let minv ' be equal to the minimum value in the difference array d ', then subtract minv ' from the value of each element in the difference array d ', then put the difference array d ' into the set S, and then go to step E5;
e5: repeating step E4 until T a "end of array and T b ' the current positions of the tail end points of the array are the same, and finally T is obtained a And T b ' putting S into a set ALL, and then entering a step E6;
e6: repeating the steps E1 to E5, and calculating the next curve data T in the set TS a And T b ' the new difference array set S is put into the set ALL until ALL the edge curve data in the TS are traversed; finally obtaining each edge curve data T in TS a And T b ' and then proceeds to step F.
F: calculating T according to each difference array in the difference array set ALL obtained in the step E a And T b The similarity of (2); then entering step G;
in said step F, the maximum is passedData difference method for calculating T a And T b The specific calculation method of the maximum similarity H is as follows:
f11: for each difference array in the difference array set ALL obtained in the step E, calculating the maximum value m of the array 1 And m is 1 Put into set M. Let m a Equal to the minimum of M;
in the present example, the maximum value M of M a Is 45.
F12: using formulas
Figure BDA0002377570180000161
Calculating T a And T b The similarity H of (A); wherein m is a The smaller the H, the closer the H is to 1,m a The larger, the closer H is to 0.
In this example, H = 1/(1 + 45/10) =0.1818.
G: and D, according to the method of the step C, the step D, the step E and the step F in sequence, carrying out similarity calculation on the normalized time-series edge curve data corresponding to each oracle bone rubbing image and the normalized time-series edge curve data corresponding to each oracle bone rubbing image to be selected in sequence to obtain corresponding similarity values respectively. And finally, selecting the oracle bone rubbing image corresponding to other ten pieces of edge curve data with the highest edge curve similarity of each oracle bone rubbing image as the optimal conjugation result of the oracle bone rubbing image.
In this example, for a certain oracle bone rubbing image, the method of the present invention is used to calculate other ten oracle bone rubbings to be selected with the highest similarity to the edge curve thereof, and the corresponding edge curve similarity values are 0.68,0.66,0.63. The ten oracle bone rubbing images are the best conjugation result of the oracle bone rubbing image. Fig. 3 is an example of a pair of best-fit conjugated oracle bone rubbings images.

Claims (7)

1. A carapace bone fragment conjugation method based on time series similarity calculation is characterized by comprising the following steps:
a: extracting an edge curve graph of each oracle bone rubbing image;
b: on an edge curve graph of each oracle bone rubbing image, sequentially reading the height value of the uppermost pixel point and the height value of the lowermost pixel point of each row of pixel data from left to right, obtaining an average value, taking the average value as the pixel position of an edge curve in the row of pixel data, and sequentially combining the pixel positions of each row of pixel data obtained in sequence to form time-series curve data T, T = { V } corresponding to the edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the pixel data of the ith column of the edge curve is represented; according to the method, time-series edge curve data of each oracle bone rubbing image are sequentially obtained;
c: normalizing the time-series edge curve data T corresponding to each oracle bone rubbing image to obtain normalized time-series edge curve data T ', T' = { V = (V) } 1 ',V 2 ',V 3 '，...，V i ' }, i is a positive integer;
d, normalizing time-series edge curve data T corresponding to two oracle bone rubbing images a and b to be judged whether to be conjugated a ' and T b ', let T a ' Length is greater than T b ', the shorter curve data T b ' step by 1 pixel at T a ' slide in sequence and at T b ' as reference basis T a ' corresponding oracle bone rubbing image a is subjected to inclination correction, and new edge curve data T is extracted from the oracle bone rubbing image a subjected to inclination correction every time a ", and will T a "put into the set TS;
e: for each edge curve T in the set TS obtained in the step D a ", calculating T a And T b ' difference array set ALL; then entering step F;
f: calculating T according to each difference array in the difference array set ALL obtained in the step E a And T b The similarity of (2); then entering step G;
g: according to the method of the step C, the step D, the step E and the step F in sequence, carrying out similarity calculation on the normalized time-series edge curve data corresponding to each oracle bone rubbing image and the normalized time-series edge curve data corresponding to each oracle bone rubbing image to be selected in sequence to obtain corresponding similarity values respectively; and finally, selecting the oracle bone rubbing image corresponding to other ten pieces of edge curve data with the highest edge curve similarity of each oracle bone rubbing image as the optimal conjugation result of the oracle bone rubbing image.
2. The method for the conjugation of oracle bone fragments based on the calculation of similarity of time series according to claim 1, characterized in that: in the step A, a digital board and image editing processing software are utilized to carry out manual edge tracing processing on the upper edge or the lower edge of each oracle bone rubbing image by using a red painting brush with the thickness of 2 pixels, the image after edge tracing processing is stored, and the edge of each obtained oracle bone rubbing image is represented by a red curve; and then, automatically extracting all pixel points with the pixel value of (255,0,0) in each oracle bone rubbing image by using the color characteristics to obtain a red edge curve graph of each oracle bone rubbing image.
3. The method for the conjugation of oracle bone fragments based on the calculation of similarity of time series according to claim 2, characterized in that: in the step B, on a red edge curve graph of each oracle bone rubbing image, sequentially reading the height value of the uppermost red pixel and the height value of the lowermost red pixel of each row of red pixel data from left to right and obtaining an average value, taking the average value as the pixel position of an edge curve on the row of red pixel data, and sequentially combining the pixel positions of each row of red pixel data obtained sequentially to form time-series curve data T, T = { V } corresponding to the red edge curve graph 1 ，V 2 ，V 3 ，...，V i I is a positive integer, V i The pixel position of the red pixel data of the ith column of the edge curve is represented; according to the method, the time-sequence edges Qu Xianshu of each oracle bone rubbing image are obtained in sequenceAccording to the T.
4. The method for the conjugation of oracle bone fragments based on the calculation of similarity of time series according to claim 1, characterized in that: when the normalization processing is performed in the step C, first, the minimum value min (T) in the time-series curve data T is calculated, and then min (T) is subtracted from each data in the corresponding time-series curve data T to obtain the normalized time-series edge curve data T' of the edge curve graph.
5. The method for the conjugation of oracle bone fragments based on the calculation of similarity of time series according to claim 1, characterized in that said step D comprises the following specific steps:
d1: normalizing the normalized time-series edge curve data T corresponding to the two edge curve graphs to be compared a ' and T b The head and tail end points of the' are directly connected, so that the lengths of the formed line segments are respectively L a And L b (ii) a If L is a Is equal to L b Entering the step D2; otherwise, with L a And L b The larger median edge curve data is used as the reference, assuming L a Greater than L b Entering step D4;
d2: respectively calculating two normalized time-series edge curve data T a ' and T b Angle of inclination of `, let θ a Is T a Angle of inclination, θ b Is T b The angle of inclination of'; let T a ' head and tail end points P 1 And P 2 The coordinates in the image are respectively (S) x ,S y ) And (E) x ,E y ) Then theta a ＝arctan((E y -S y )/(E x -S x ) Where arctan is an arctan function; in a similar way, according to T b ' head and tail end points P 3 And P 4 Coordinates in the image, calculating T b Angle of inclination theta of ` b (ii) a Finally, calculate T a ' and T b The difference of the tilt angle of ' θ ', θ ' = θ a -θ b Then enter D3;
d3: if θ' is greater than 0, then T is added a 'clockwise rotation of the corresponding oracle bone rubbing image by an angle theta'; if θ' is less than 0, then T a The corresponding oracle bone rubbing image is anticlockwise rotated by an angle of theta ' |, wherein the angle of theta ' | is the absolute value of theta '; for T a ' corresponding rotated oracle bone rubbing image, extracting new edge curve graph and normalized time-series edge curve data T sequentially through step B and step C a ", and new normalized time-sequenced edge curve data T a Putting in a set TS, and then entering the step E;
d4: let P 5 As a starting point, at an initial time P 5 Is equal to T a ' head end P 1 At T a ' from the starting point P 5 At the beginning, in the order of T from front to back a ' center search Point P 6 Up to P 6 And P 5 Is equal to L b (ii) a Let T a ' middle P 5 And P 6 Between the subplot data is T sub Calculating T using the method in step D2 sub And T b 'inclination angle difference θ', then using the method in step D3 for T a Rotating the corresponding oracle bone rubbing image, extracting a new edge curve graph from the rotated oracle bone rubbing image, and obtaining new normalized time-series edge curve data T a ", will T a "put into the set TS; then entering step D5;
d5: taking 1 pixel as a step, and dividing P 5 At T a ' slide backward in the process, repeat step D4, get new normalized time series curve data T a ", and will T a "put into the set TS; then entering step D6;
d6: repeating step D5 until P 6 Is equal to T a ' end point, finally obtaining T a Rotating the corresponding oracle bone rubbing image according to different angles and re-extracting a set TS formed by normalized time-series edge curve data; then step E is entered.
6. The method for the conjugation of thyroid fragments based on the calculation of similarity of time series according to claim 5, wherein said step E comprises the following specific steps:
e1: reading curve data T from set TS a ", if T a And T b If the array lengths are consistent, entering step E2; if T a And T b If the array lengths of the' are not consistent, the step E3 is carried out;
e2: calculating to obtain T a And T b The data difference values of the two curve data at each corresponding position form a difference value array d in sequence, minv is made to be equal to the minimum value in the difference value array d, then minv is subtracted from the value of each element in the difference value array d, finally the difference value array d is placed into a difference value array set ALL, and then the step E6 is carried out;
suppose that the two curve data sequences are T respectively a ”＝{T a1 ”,T a2 ”,T a3 ”，...，T ai "} and T b '＝{T b1 ',T b2 ',T b3 ',...，T bi ' }; calculating the difference value of the two curve data sequences to obtain a difference value array d, d = { | T a1 ”-T b1 '|,|T a2 ”-T b2 '|,|T a3 ”-T b3 '|,...，|T ai ”-T bi ' | }; let minv equal the minimum in the difference array d, and finally d = { | T a1 ”-T b1 '|-minv,|T a2 ”-T b2 '|-minv,|T a3 ”-T b3 '|-minv,...|T ai ”-T bi '|-minv}；
E3: let P 7 As a starting point, at an initial time P 7 Is equal to T a "the head end point; at T a "from the starting point P 7 At the beginning, T b ' array head end position and P 7 Aligning, calculating T a And T b ' data difference value for each corresponding position, up to T b The data in the array are all calculated to form a difference array d ', the obtained data differences are made into a difference array d ' in sequence, minv ' is made to be equal to the minimum value in the difference array d ', then minv ' is subtracted from the value of each element in the difference array d ', then the difference array d ' is put into a set S, and then the step E4 is carried out;
e4: taking 1 pixel as a step, and dividing P 7 At T a "slide backward in the middle, repeat step E3, and let T b ' array of head position and P 7 Align and calculate T a And T b 'the difference array d' is made to be equal to the minimum value in the difference array d ', then minv' is subtracted from the value of each element in the difference array d ', then the difference array d' is put into the set S, and then the step E5 is carried out;
e5: repeating step E4 until T a "end of array and T b ' the current positions of the tail end points of the array are the same, and finally T is obtained a And T b ' putting S into a set ALL, and then entering a step E6;
e6: repeating the steps E1 to E5, and calculating the next curve data T in the set TS a And T b ' the new difference array set S is put into the set ALL until ALL the edge curve data in the TS are traversed; finally obtaining each edge curve data T in TS a And T b ' and then proceeds to step F.
7. The method for the conjugation of oracle bone fragments based on the calculation of similarity of time series according to claim 1, characterized in that: in the step F, T is calculated by a maximum data difference method a And T b The specific calculation method of the maximum similarity H is as follows:
f11: for each difference array in the difference array set ALL obtained in the step E, calculating the maximum value m of the array 1 And m is 1 Put into the set M, let M a Equal to the minimum of M;
f12: using formulas
Figure FDA0002377570170000061
Calculating to obtain T a And T b The similarity H of (A); wherein m is a The smaller the H, the closer to 1,m it is a The larger, the closer H is to 0.
CN202010072172.5A 2020-01-21 2020-01-21 Carapace bone fragment conjugation method based on time sequence similarity calculation Active CN111292283B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010072172.5A CN111292283B (en) 2020-01-21 2020-01-21 Carapace bone fragment conjugation method based on time sequence similarity calculation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010072172.5A CN111292283B (en) 2020-01-21 2020-01-21 Carapace bone fragment conjugation method based on time sequence similarity calculation

Publications (2)

Publication Number Publication Date
CN111292283A CN111292283A (en) 2020-06-16
CN111292283B true CN111292283B (en) 2023-02-07

Family

ID=71025642

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010072172.5A Active CN111292283B (en) 2020-01-21 2020-01-21 Carapace bone fragment conjugation method based on time sequence similarity calculation

Country Status (1)

Country Link
CN (1) CN111292283B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112070684B (en) * 2020-08-04 2023-05-23 辽宁师范大学 Method for repairing characters of a bone inscription based on morphological prior features
CN112837334B (en) * 2021-04-02 2022-07-05 河南大学 Automatic conjugation method of Chinese character and sketch image
CN112991185B (en) * 2021-04-23 2022-09-13 河南大学 Automatic conjugation method for Dunhuang relic image

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101872478A (en) * 2010-05-31 2010-10-27 徐州师范大学 Computer aided restoration method of oracle bone rubbing font
WO2015184764A1 (en) * 2014-11-17 2015-12-10 中兴通讯股份有限公司 Pedestrian detection method and device
CN110598030A (en) * 2019-09-26 2019-12-20 西南大学 Oracle bone rubbing classification method based on local CNN framework

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101872478A (en) * 2010-05-31 2010-10-27 徐州师范大学 Computer aided restoration method of oracle bone rubbing font
WO2015184764A1 (en) * 2014-11-17 2015-12-10 中兴通讯股份有限公司 Pedestrian detection method and device
CN110598030A (en) * 2019-09-26 2019-12-20 西南大学 Oracle bone rubbing classification method based on local CNN framework

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
一种计算机辅助甲骨文拓片缀合方法;张长青等;《电子设计工程》;20120905(第17期);全文 *
基于泊松分布和分形几何的甲骨拓片字形复原方法;顾绍通等;《中国科学:信息科学》;20110115(第01期);全文 *

Also Published As

Publication number Publication date
CN111292283A (en) 2020-06-16

Similar Documents

Publication Publication Date Title
CN111292283B (en) Carapace bone fragment conjugation method based on time sequence similarity calculation
Levi et al. LATCH: learned arrangements of three patch codes
Li et al. Fss-1000: A 1000-class dataset for few-shot segmentation
CN107480727B (en) Unmanned aerial vehicle image fast matching method combining SIFT and ORB
Chung et al. Jigsaw puzzle solver using shape and color
CN103578125B (en) Image processing equipment and image processing method
WO2019007004A1 (en) Image feature extraction method for person re-identification
US20050232490A1 (en) Red-eye detection device, red-eye detection method, and red-eye detection program
EP0534446A2 (en) System with approximation means for recognizing graphical elements in a drawing
CN107590772A (en) A kind of cultural relic fragments method for automatically split-jointing based on adaptive neighborhood matching
Li et al. Line segment matching: A benchmark
Shah et al. Multistage sfm: Revisiting incremental structure from motion
Wendel et al. Unsupervised facade segmentation using repetitive patterns
CN108009986A (en) Fragments mosaicing method and apparatus based on marginal information
CN103733225B (en) Characteristic point peer system, characteristic point counterpart method and record medium
CN111951152B (en) A-bone conjugation method taking primary side continuity and broken edge slag port matching degree into consideration
Shah et al. Multistage SfM: A coarse-to-fine approach for 3d reconstruction
Nomir et al. Hierarchical dental x-ray radiographs matching
Flores et al. Computer algorithm for archaeological projectile points automatic classification
CN105701500A (en) Single-sided English paper scrap splicing identification method
CN107437097B (en) Two-stage local contour matching method based on angular point description
Van Phan et al. A re-assembling scheme of fragmented Mokkan images
CN111626953A (en) Fragment splicing method and device based on multi-feature information
Kamran et al. An LCS-based 2D fragmented image reassembly algorithm
CN113947678A (en) Image alignment method suitable for text image

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant