CN113112570A - Vectorization effect evaluation method based on perception drive - Google Patents

Vectorization effect evaluation method based on perception drive Download PDF

Info

Publication number
CN113112570A
CN113112570A CN202110515383.6A CN202110515383A CN113112570A CN 113112570 A CN113112570 A CN 113112570A CN 202110515383 A CN202110515383 A CN 202110515383A CN 113112570 A CN113112570 A CN 113112570A
Authority
CN
China
Prior art keywords
vectorization
vector diagram
raster
evaluation method
effect evaluation
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.)
Granted
Application number
CN202110515383.6A
Other languages
Chinese (zh)
Other versions
CN113112570B (en
Inventor
赵海英
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.)
Beijing University of Posts and Telecommunications
Original Assignee
Beijing University of Posts and Telecommunications
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 Beijing University of Posts and Telecommunications filed Critical Beijing University of Posts and Telecommunications
Priority to CN202110515383.6A priority Critical patent/CN113112570B/en
Publication of CN113112570A publication Critical patent/CN113112570A/en
Application granted granted Critical
Publication of CN113112570B publication Critical patent/CN113112570B/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
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/40—Filling a planar surface by adding surface attributes, e.g. colour or texture
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Abstract

The invention relates to a perception-driven image vectorization effect quantitative evaluation method, which comprises the following steps: analyzing the vector diagram, acquiring a document object model of the vector diagram and traversing a document tree to obtain detailed information of a vector path; then, quantifying the visual effect of the vector diagram by combining the form tower psychology, accurately reflecting that the visual error between the original diagram and the vector diagram is measured through the global pixel error, continuously reflecting that the fluency of the curve is measured through the curvature change of the splicing part of the statistical curve, simply reflecting that the complexity of the pattern is weighted by linear path and polynomial path proportion, area number, color number and the like to obtain a measurement value; and finally, giving different weights to the quantized values of different dimensions according to the materials to obtain the comprehensive evaluation measurement of the vector diagram. The embodiment of the invention can realize automatic evaluation, sequencing and screening of the vectorization result of the traditional texture pattern and help a user to obtain a vector diagram with higher quality.

Description

Vectorization effect evaluation method based on perception drive
Technical Field
The invention relates to the technical field of computer image processing, in particular to a method for quantitatively evaluating the vectorization effect of a vector diagram.
Background
The traditional ornamentation pattern contains lucky connotation which holds the beauty pursuit of people and is an important component of the traditional Chinese culture. The method for digitalizing and recycling the cultural heritage is an effective inheritance protection method, a material library is constructed, Chinese cultural elements and marks are integrated into content creation production and creative design, and the traditional ornamentation can be revitalized. The vector format picture has the advantages of independence on equipment resolution, undistorted zooming, capability of being edited by a user, small file storage volume and the like, and is widely applied to the design and publishing industries. The vectorization research on the texture patterns can provide powerful technical support and tool support for the digitization and the reuse of cultural resources.
The existing process of constructing the vector material library still has a plurality of challenges, on one hand, the sources of the texture patterns are widely and largely appeared in traditional clothes, ceramics, bronze ware and murals, and the obtained original image has complex texture, which results in poor vectorization effect; on the other hand, the existing vectorization algorithm has poor robustness, and the quality of the vector image obtained by applying the vectorization algorithm to the pattern is uneven. Therefore, when the pattern is put in storage, a large amount of manpower is needed to screen and sort the vector diagrams automatically generated by the algorithm.
Disclosure of Invention
The invention aims to solve the problems in the image vectorization process, and provides a vectorization effect quantitative evaluation method based on perception drive, which quantifies the visual preference of human eyes, and performs quantitative analysis on three visual characteristics of the accuracy, continuity and simplicity of a pattern, thereby completing the automatic evaluation of a vector diagram. Therefore, the method helps a user to quickly discriminate the high-quality vector diagrams, can also sequence the vector diagrams in the material library, and provides quantitative indexes for analyzing the advantages and the disadvantages of different vectorization algorithms.
The invention provides
The vectorization effect evaluation method based on perception driving comprises the following steps:
step 1, collecting raster images needing vectorization processing in a material library, and converting the raster images into vector diagrams, wherein each vector diagram has a raster image as an original image corresponding to the raster image;
step 2, analyzing the vector diagram, wherein the vector diagram is represented as a tree T (V, E), a node V of the tree is an n-element array, each element is a Key-Value Key Value pair, and attributes of the nodes and values of the attributes are represented; the edge E of the tree represents the hierarchical relationship in the vector diagram; traversing the tree T to obtain a closed region of the vector diagram, and filled color and geometric information, wherein the geometric information comprises a plurality of path paths, and the path paths comprise a plurality of curves;
step 3, calculating the following three indexes for each vector diagram:
3.1, accuracy, calculated by the following formula:
Figure BDA0003061729910000021
Figure BDA0003061729910000022
faccuracy(I) indicating the accuracy of vectorization of the grating pattern I, | I | indicates the number of pixels of the grating pattern I, Δ (x, y) is an intermediate quantity, I (x, y) indicates a component of the RGB color space corresponding to the grating pattern I at coordinates (x, y), I ' (x, y) indicates a component of the RGB color space at coordinates (x, y) of a bitmap I ' obtained by rasterization after vectorization of the corresponding grating pattern I | I ' (x, y) -I (x, y) |2Representing the Euclidean distance between I (x, y) and I' (x, y), wherein tolerance represents the visual tolerance and is a preset threshold;
3.2, continuity, calculated by the following formula:
Figure BDA0003061729910000023
fcontinuiy(T) denotes the continuity of the vector diagram corresponding to tree T, Cx, Cy are two curves joined end-to-end in path, angleCx,CyRepresenting the angle at which the curve Cx intersects the curve Cy; llangIs a preset continuity parameter;
3.3, simplicity, calculated by the following formula: :
fsimplicty(I)＝(∑e∈E1+deg(e))+count(regions,I)+3count(color,I)，
wherein f issimplicty(I) Representing the simplicity of vectorization of the raster pattern I, E represents that the raster pattern I corresponds to a curve unit in the vector diagram, E represents that the raster pattern I corresponds to all curves in the vector diagram, deg (E) represents the highest item times of the curve unit E, and count (regions) represents the number of closed areas of the tree T corresponding to the raster pattern I; with count (color, I) representing the tree T for the raster pattern IThe number of colors of the closed area;
3.4, comprehensively weighting the accuracy, the continuity and the simplicity:
fperceive(I)＝αw(i)+βw(j)+γw(k)i,j,k∈N，
fperceive(I) a comprehensive weighted value representing the vectorization effect evaluation of the raster image I; i represents the sequencing of the accuracy of the raster image I vectorization in the material library; j represents the sequence of the continuity of the raster image I vectorization in the material library; k represents the sequence of the simplicity of the raster image I vectorization in the material library; w (×) represents the distribution function, α, β, γ are the weights of the three dimensions;
and 3.5, sequencing the comprehensive weighted values of all the raster patterns, and evaluating the vectorization effect through the comprehensive weighted values.
The method can realize automatic evaluation, sequencing and screening of the vectorization result of the traditional texture pattern, and help a user to obtain a vector diagram with higher quality.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a vectorization effect quantitative evaluation method provided in an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, a flowchart of a vectorization effect quantitative evaluation method based on perceptual driving according to an embodiment of the present invention is shown. The principle of the method is as follows: and quantifying the accuracy, continuity and simplicity of three important factors influencing the visual perception effect of the vector diagram through the psychological analysis of the format tower to obtain indexes of three dimensions of the vector diagram and sort the indexes in a library.
Specifically, the vectorization effect quantitative evaluation method based on perceptual driving in this embodiment includes the following steps:
s110, collecting raster image materials needing vectorization processing in a material library, taking the materials as input, obtaining corresponding vector diagrams by using the existing vectorization algorithm, wherein each vector diagram has a raster image as an original image, and setting the weight of vector diagram accuracy measurement obtained by manual tracing of a designer to be 0.
And S120, analyzing the vector diagram.
The format of the vector diagram is generally compatible with a document object model, so that the vector diagram can be represented as a tree T (V, E), a node V of the tree is an n-element array, each element is a Key-Value Key Value pair, and attributes of the node and values of the attributes are represented; the edge E of the tree represents the hierarchical relationship in the vector diagram; the traversal tree T acquires the closed regions of the vector image, and the filled colors and geometric information, wherein the geometric information contains a plurality of path paths containing a plurality of curves.
S130, calculating the following three indexes for each vector diagram:
and 3.1, accuracy, the capability of retaining original image information in the vectorization process is reflected, and the error between the bitmap obtained by the vector diagram through a rasterization equation and the original image with the same resolution is used for measurement. Given by:
Figure BDA0003061729910000041
Figure BDA0003061729910000051
faccuracy(I) Indicating the accuracy of vectorization of the grating pattern I, | I | indicates the number of pixels of the grating pattern I, Δ (x, y) is an intermediate quantity, I (x, y) indicates a component of the RGB color space corresponding to the grating pattern I at coordinates (x, y), I ' (x, y) indicates a component of the RGB color space at coordinates (x, y) of a bitmap I ' obtained by rasterization after vectorization of the corresponding grating pattern I | I ' (x, y) -I (x, y) |2Representing Euclidean distance between I (x, y) and I' (x, y), wherein tolerance represents visual tolerance, is a preset threshold, and the tolerance value range of the color image is [10 ]2,302]The tolerance of the binary image is taken to be [50 ]2,1002]。
And calculating errors between the reconstructed image and the original image one by one pixel point in the RGB space, marking the point which is in line with expectation as 1, otherwise marking the point as 0, and finally counting the pixel points which meet the accuracy. The reason why the visual tolerance must be given is that a closed region with consistent colors is used to represent discrete pixel points of the original image in the vectorization process, and colors in the same region of the original image may be multiple, so that the visual tolerance can be well balanced.
3.2, continuity reflects the fitting condition of the vectorization process to the region boundary, the sensing of human eyes can highlight the inflection point with a smaller included angle at the curve splicing part, and the curve with a larger span can be smooth.
The continuity is calculated by the following formula:
Figure BDA0003061729910000052
fcontinuiy(T) denotes the continuity of the vector diagram corresponding to tree T, Cx, Cy are two curves joined end-to-end in path, angleCx,CyRepresenting the angle at which the curve Cx intersects the curve Cy; llangFor a predetermined continuity parameter,/, in this examplelangIf a large number of angles smaller than 45 ° occur in the vector path, the formula indicates that the vectorization forms a large number of inflection points.
3.3, the simplicity reflects the condition of the vector diagram component elements, and the calculation is carried out by counting the path information of different nodes through traversing the DOM tree of the whole vector diagram:
fsimplicty(I)＝(∑e∈E1+deg(e))+count(regions,I)+3count(color,I)，
wherein f issimplicty(I) Representing the simplicity of vectorization of the raster pattern I, E represents that the raster pattern I corresponds to a curve unit in the vector diagram, E represents that the raster pattern I corresponds to all curves in the vector diagram, deg (E) represents the highest item times of the curve unit E, and count (regions) represents the number of closed areas of the tree T corresponding to the raster pattern I; count (color, I) denotes the number of colors of the closed area of the raster pattern I corresponding to the tree T.
The complexity of the curve is represented by the number of times of the curve, the weight of the straight line is 1, the weight of the quadratic Bezier curve is 2, and the weight of the cubic Bezier curve is 3. The weight of the number of regions is 1, the weight of the number of colors is 3, and the weight of the colors is higher than that of the regions because different regions can have the same color, and the greater the number of colors in the vector diagram, the more the value of the simplicity is, the greater the number of colors.
And S140, carrying out normalization processing on the three measurement values, and then carrying out comprehensive weighting on the indexes based on visual perception driving.
fperceive(I)＝αw(i)+βw(j)+γw(k)i,j,k∈N，
fperceive(I) A comprehensive weighted value representing the vectorization effect evaluation of the raster image I; i represents the sequencing of the accuracy of the raster image I vectorization in the material library; j represents the sequence of the continuity of the raster image I vectorization in the material library; k represents the sequence of the simplicity of the raster image I vectorization in the material library; w (×) represents an assignment function that penalizes indicators that are either too high or too low in rank. Alpha, beta and gamma are the weights of three dimensions, so that a user can conveniently sort only a certain interested dimension. Ordering the comprehensive weighted values of all raster images, and evaluating the vectorization effect by the comprehensive weighted values
The method of the invention is based on the following three heuristic principles: the visual effect is poor when the three quantitative indexes are too high or too low; the vector diagram has good visual effect, and the simplicity and the continuity are always positively correlated; the vector diagram with good visual effect, simplicity and accuracy are negative correlation. And after the iteration is finished, obtaining the metric values of three dimensions of the vector diagram and the bit sequence in the material library.
In addition to the above embodiments, the present invention may have other embodiments. All technical solutions formed by adopting equivalent substitutions or equivalent transformations fall within the protection scope of the claims of the present invention.

Claims (9)

1. A vectorization effect evaluation method based on perception driving comprises the following steps:
step 1, collecting raster images needing vectorization processing in a material library, and converting the raster images into vector diagrams, wherein each vector diagram has a raster image as an original image corresponding to the raster image;
step 2, analyzing the vector diagram, wherein the vector diagram is represented as a tree T (V, E), a node V of the tree is an n-element array, each element is a Key-Value Key Value pair, and attributes of the nodes and values of the attributes are represented; the edge E of the tree represents the hierarchical relationship in the vector diagram; traversing the tree T to obtain a closed region of the vector diagram, and filled color and geometric information, wherein the geometric information comprises a plurality of path paths, and the path paths comprise a plurality of curves;
step 3, calculating the following three indexes for each vector diagram:
3.1, accuracy, calculated by the following formula:
Figure FDA0003061729900000011
Figure FDA0003061729900000012
faccuracy(I) representing the accuracy of vectorization of the grating pattern I, | I | represents the number of pixels of the grating pattern I, Δ (x, y) is an intermediate quantity, and I (x, y) represents the corresponding grating patternI a component of RGB colour space at co-ordinate (x, y), I ' (x, y) representing a component of RGB colour space at co-ordinate (x, y) of a bitmap I ' obtained by rasterization after vectorisation of the corresponding raster pattern I, | I ' (x, y) -I (x, y) |2Representing the Euclidean distance between I (x, y) and I' (x, y), wherein tolerance represents the visual tolerance and is a preset threshold;
3.2, continuity, calculated by the following formula:
Figure FDA0003061729900000013
fcontinuiy(T) denotes the continuity of the vector diagram corresponding to tree T, Cx, Cy are two curves joined end-to-end in path, angleCx,CyRepresenting the angle at which the curve Cx intersects the curve Cy; llangIs a preset continuity parameter;
3.3, simplicity, calculated by the following formula: :
fsimplicty(I)＝(∑e∈E 1+deg(e))+count(regions,I)+3count(color,I)，
wherein f issimplicty(I) Representing the simplicity of vectorization of the raster pattern I, E represents that the raster pattern I corresponds to a curve unit in the vector diagram, E represents that the raster pattern I corresponds to all curves in the vector diagram, deg (E) represents the highest item times of the curve unit E, and count (regions) represents the number of closed areas of the tree T corresponding to the raster pattern I; count (color, I) represents the number of colors of the closed area of the tree T corresponding to the raster pattern I;
3.4, comprehensively weighting the accuracy, the continuity and the simplicity:
fperceive(I)＝αw(i)+βw(j)+γw(k)i,j,k∈N，
fperceive(I) a comprehensive weighted value representing the vectorization effect evaluation of the raster image I; i represents the sequencing of the accuracy of the raster image I vectorization in the material library; j represents the sequence of the continuity of the raster image I vectorization in the material library; k represents the sequence of the simplicity of the raster image I vectorization in the material library; w (×) represents the distribution function, α, β, γ are the weights of the three dimensions;
and 3.5, sequencing the comprehensive weighted values of all the raster patterns, and evaluating the vectorization effect through the comprehensive weighted values.
2. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in the step 1, the vector image material can be obtained by any existing image vectorization algorithm or manually traced by a designer.
3. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in the step 1, the material is divided into the raster original image and the vector diagram, and the raster image and the vector diagram are in one-to-one relationship.
4. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in the step 2, the tree T corresponding to the vector diagram is traversed to obtain relevant information, and then the accuracy, the continuity and the simplicity of the vector diagram are calculated according to the perception-driven vectorization effect evaluation method.
5. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in step 3.4, the metric values are normalized.
6. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in step 3.1, the tolerance value range of the color image is [10 ]2,302]The tolerance of the binary image is taken to be [50 ]2,1002]。
7. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in step 3.2, the parameter llang＝135°。
8. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: in step 3.4, the user can set the weights according to his preferences.
9. The vectorization effect evaluation method based on perceptual driving according to claim 1, wherein: the designer manually traces the resulting vector diagram with a precision measure weighted 0.
CN202110515383.6A 2021-05-12 2021-05-12 Vectorization effect evaluation method based on perception drive Active CN113112570B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110515383.6A CN113112570B (en) 2021-05-12 2021-05-12 Vectorization effect evaluation method based on perception drive

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110515383.6A CN113112570B (en) 2021-05-12 2021-05-12 Vectorization effect evaluation method based on perception drive

Publications (2)

Publication Number Publication Date
CN113112570A true CN113112570A (en) 2021-07-13
CN113112570B CN113112570B (en) 2022-05-20

Family

ID=76721969

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110515383.6A Active CN113112570B (en) 2021-05-12 2021-05-12 Vectorization effect evaluation method based on perception drive

Country Status (1)

Country Link
CN (1) CN113112570B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102136151A (en) * 2011-03-11 2011-07-27 山东大学 Method for vectorizing raster image
WO2012114386A1 (en) * 2011-02-25 2012-08-30 三菱電機株式会社 Image vectorization device, image vectorization method, and image vectorization program
CN105741280A (en) * 2016-01-26 2016-07-06 广州市城市规划勘测设计研究院 Fuzzy mathematic vector regional evaluation method and apparatus
US20180232863A1 (en) * 2016-06-22 2018-08-16 Adobe Systems Incorporated Enhanced Vectorization of Raster Images
CN111462136A (en) * 2020-03-31 2020-07-28 上海大学 Image vectorization system and method based on boundary segmentation algorithm

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012114386A1 (en) * 2011-02-25 2012-08-30 三菱電機株式会社 Image vectorization device, image vectorization method, and image vectorization program
CN102136151A (en) * 2011-03-11 2011-07-27 山东大学 Method for vectorizing raster image
CN105741280A (en) * 2016-01-26 2016-07-06 广州市城市规划勘测设计研究院 Fuzzy mathematic vector regional evaluation method and apparatus
US20180232863A1 (en) * 2016-06-22 2018-08-16 Adobe Systems Incorporated Enhanced Vectorization of Raster Images
CN111462136A (en) * 2020-03-31 2020-07-28 上海大学 Image vectorization system and method based on boundary segmentation algorithm

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
周武杰等: "基于视觉感知和零水印的部分参考立体图像质量评价模型", 《电子与信息学报》 *

Also Published As

Publication number Publication date
CN113112570B (en) 2022-05-20

Similar Documents

Publication Publication Date Title
US10896524B2 (en) Method and system for color representation generation
RU2680765C1 (en) Automated determination and cutting of non-singular contour of a picture on an image
CN101079952B (en) Image processing method and image processing apparatus
CN107610110B (en) Global and local feature combined cross-scale image quality evaluation method
KR20030048435A (en) Method and apparatus for image analysis and processing by identification of characteristic lines and corresponding parameters
Berretti et al. 3d mesh decomposition using reeb graphs
CA2622729A1 (en) Framed art visualization software
US20040050400A1 (en) Mascara selecting method, mascara selecting system, and mascara counseling tool
CN111666900B (en) Land cover classification map obtaining method and device based on multi-source remote sensing image
US20130093768A1 (en) Systems and Methods for Creating Texture Exemplars
CN112990183B (en) Method, system and device for extracting homonymous strokes of offline handwritten Chinese characters
US20050135667A1 (en) Method and apparatus for labeling images and creating training material
Gong et al. User-assisted image shadow removal
CN113396424A (en) System and method for automated material extraction
CN112634125A (en) Automatic face replacement method based on off-line face database
CN113112570B (en) Vectorization effect evaluation method based on perception drive
Narwaria et al. An objective method for high dynamic range source content selection
Adamek et al. Using dempster-shafer theory to fuse multiple information sources in region-based segmentation
CN114548250A (en) Mobile phone appearance detection method and device based on data analysis
dos Passos et al. Sample-based synthesis of illustrative patterns
Cai et al. Unsupervised segmentation evaluation: an edge-based method
Zhang et al. Computer simulation of archaeological drawings
Munoz-Pandiella et al. A 3D feature-based approach for mapping scaling effects on stone monuments
CN111325732B (en) Face residue detection method and related equipment
CN117176983B (en) Video generation evaluation system based on panoramic image synthesis

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