Search

CN-116681909-B - Cultural relic fragment similarity measurement method based on multi-feature information

CN116681909BCN 116681909 BCN116681909 BCN 116681909BCN-116681909-B

Abstract

The invention discloses a cultural relic fragment similarity measurement method based on multi-feature information, which comprises the steps of collecting cultural relic fragment data, clustering cultural relic fragment images, selecting one type of clustered cultural relic fragment images to be converted into binary images, extracting contour lines of the binary images one by one, identifying corner points of the contour lines and identifying similarity of two cultural relic fragment images. According to the invention, the ceramic fragments with different characteristics are clustered before the fragments are spliced according to the color information and the texture characteristics of the ceramic fragments, the edge superpixel characteristics and the contour curvature characteristics of the ceramic fragments are fused, the similarity degree between the fragments is calculated by using a difference and distance measurement formula, the time and complexity of the fragment splicing are effectively reduced, and the accuracy of the similarity comparison of the fragments is improved.

Inventors

  • ZHAO SHULIANG
  • WANG XINPING
  • Liang Senwei

Assignees

  • 河北师范大学

Dates

Publication Date
20260512
Application Date
20230609

Claims (7)

  1. 1. A cultural relic fragment similarity measurement method based on multi-feature information is characterized by comprising the following steps: step 1, acquiring data of cultural relic fragments, namely acquiring more than 1 cultural relic fragment image, and recording the image as Pm= { p 1 ,p 2 ,...,p mm }, wherein mm is the number of cultural relic fragments; step 2, clustering the cultural relic fragment images, namely clustering the clustered categories Where nn is the number of categories after clustering, The ith fragment belongs to the jth class; Step 3, selecting a class of clustered cultural relic fragment images to be converted into a binary image; Step 4, extracting contour lines of the binarized images one by one; step 5, identifying angular points of the contour lines, and dividing the contour lines into contour line segments according to the angular points; and 6, identifying the similarity of two cultural relic fragment images, wherein the similarity comprises the following specific steps of: step 6-1, predicting the abrasion of the edge of the fragment of the cultural relic, and defining fragment edge pixels along the contour line according to a preset prediction bandwidth; Step 6-2 extraction of the first Super-pixels of fragment edge pixels of the individual cultural relic fragment image form the first Edge superpixel set of individual cultural relic fragments Wherein Represents the first The first cultural relic fragment Edge superpixel, in is the first The number of edge superpixels of the individual cultural relic fragments; Step 6-3, extracting the feature point set on each contour line by using a vertical distance method Combining two feature points spaced 1 feature point one by one into feature point pairs, fitting feature lines where the feature point pairs are located, calculating vertical distances between the middle feature points and the corresponding feature lines, deleting the middle feature points if the vertical distances are smaller than a preset distance threshold value, and screening the feature point sets to obtain the feature point sets Every 3 continuous feature points after screening form feature segments, and respectively calculating curvature sets of each feature segment Wherein Represents the first The first of the contour lines A curvature; Step 6-4, calculating chord lengths of characteristic segments formed by every 3 continuous screened characteristic points: (1) Building chord length sets ; Step 6-5, calculating the difference and the SoD distance between the superpixel sets of the two contour lines: (2) (3) (4) (5) and 7, judging the similarity, namely when the SOD is smaller than a similarity threshold value, determining that the two contour feature segments are similar.
  2. 2. The multi-feature information-based cultural relic fragment similarity measurement method according to claim 1, wherein: The step 5 specifically comprises the following steps: step 5-1, approximating the outline shape of the cultural relic fragments by using a polygonal fitting function, fitting the shapes of the cultural relic fragments into polygons, and extracting key points of the fitted polygons as corner points of the cultural relic fragments Dividing the fragment contour line to obtain a contour line segment, wherein nc represents a total fitting number; step 5-2, according to the corner points Dividing the contour line, wherein the set of the divided contour line segments is expressed as: L={l1,l2,......,lnl} (6) representing the segmented i-th contour segment, 。
  3. 3. The method for measuring the similarity of cultural relic fragments based on multi-feature information according to claim 2, wherein step 5-1 uses a B-spline interpolation function to fit the contour.
  4. 4. The method for measuring the similarity of cultural relic fragments based on multi-feature information according to claim 2, wherein step 5-1 is used for fitting a contour line by using a third-order spline interpolation function.
  5. 5. The method for measuring the similarity of cultural relic fragments based on multi-feature information according to claim 2, wherein step 5-1 is used for fitting a contour line by using a fourth-order spline interpolation function.
  6. 6. The method for measuring the similarity of cultural relic fragments based on multi-feature information according to claim 2, wherein step 5-1 is used for fitting a contour line by using a fifth-order spline interpolation function.
  7. 7. The method for measuring the similarity of cultural relic fragments based on multi-feature information according to claim 2, wherein step 5-1 is used for fitting a contour line by using a sixth-order spline interpolation function.

Description

Cultural relic fragment similarity measurement method based on multi-feature information Technical Field The invention relates to a cultural relic fragment similarity measurement method, in particular to a cultural relic fragment similarity measurement method based on multi-feature information, and belongs to the technical field of computer vision. Background Cultural relics remain as historic substances and are increasingly attracting attention. Archaeologists have discovered a large number of ceramic fragments in archaeological sites. Archaeologists need to integrate the fragments. After the archaeological technician collects the fragments, the fragments are cleaned, classified, and recombined according to their own experience. However, the manual splicing of the pieces of the cultural relics may face the risk of re-breakage of the cultural relics, and the manual splicing of a large number of ceramic pieces is time consuming and costly. With the development of computer technology, image processing and other technologies, archaeologists can cluster fragments and identify the similarity by means of a computer so as to prepare for subsequent splicing work. At present, many scholars have studied fragment splicing, and Ji Zhoujiang in 2009 published paper "a new two-dimensional fragment contour matching method" in "computer application research" volume 26, 8, a two-dimensional fragment contour matching algorithm for representing a contour by making the position relationship between each point and six adjacent points is proposed. The method is very simple in calculation processing, and can effectively improve the matching speed of the outlines of the fragments, but the algorithm cannot deal with the situation when the fragments rotate. Liao Haibo in 2014 published in journal of laser, volume 35, 12, a two-dimensional patch concatenation based on contour corner points and gray scale, a method using angles and distances between corner points as matching features is proposed. However, the method only extracts angular points as features to match, and does not consider the situation that the contour features are not obvious. Efthymia Tsamoura in 2010, paper "Automatic color based reassembly of FRAGMENTED IMAGES AND PAINTINGS" published in "IEEE Transactions on Image Processing", volume 19, 3, proposes automatic color oil painting fragments splicing based on color information, but this method only uses the color information of the outline, and does not use the shape information of the outline. Yuan Jie in 2018 published paper "relic fragment splicing algorithm based on contour line bidirectional distance field" in volume 44 and 6 of computer engineering "provides that the characteristic points of the textures of fragments are obtained by extracting the display ridge lines of the relic fragments, and the method can achieve better splicing effect on the relic fragments with clearer textures. However, this method does not take into account the fact that the chips are buried deeply under the ground for a long time, and the surface texture may be worn. Disclosure of Invention The invention aims to provide a cultural relic fragment similarity measurement method based on multi-feature information. In order to solve the technical problems, the invention adopts the following technical scheme: A cultural relic fragment similarity measurement method based on multi-feature information comprises the following steps: Step 1, acquiring data of cultural relic fragments, namely acquiring more than 1 cultural relic fragment image, and recording the image as Pm= { p 1,p2,...,pmm }, wherein mm is the number of cultural relic fragments; Step 2, clustering the cultural relic fragment images, wherein the clustered categories C= { C 1,C2,...,Cnn }, wherein nn is the number of clustered categories, The ith fragment belongs to the jth class; Step 3, selecting a class of clustered cultural relic fragment images to be converted into a binary image; Step 4, extracting contour lines of the binarized images one by one; step 5, identifying angular points of the contour lines, and dividing the contour lines into contour line segments according to the angular points; and 6, identifying the similarity of two cultural relic fragment images, wherein the similarity comprises the following specific steps of: step 6-1, predicting the abrasion of the edge of the fragment of the cultural relic, and defining fragment edge pixels along the contour line according to a preset prediction bandwidth; Step 6-2 extraction of the first Super-pixels of fragment edge pixels of the individual cultural relic fragment image form the firstEdge superpixel set of individual cultural relic fragmentsWhereinRepresents the firstThe first cultural relic fragmentEdge superpixel, in is the firstThe number of edge superpixels of the individual cultural relic fragments; Step 6-3, extracting the feature point set on each contour line by using a vertical distance method Combining two fe