CN109087364B - Method for predicting color harmony degree according to mode - Google Patents

Method for predicting color harmony degree according to mode Download PDF

Info

Publication number
CN109087364B
CN109087364B CN201810894794.9A CN201810894794A CN109087364B CN 109087364 B CN109087364 B CN 109087364B CN 201810894794 A CN201810894794 A CN 201810894794A CN 109087364 B CN109087364 B CN 109087364B
Authority
CN
China
Prior art keywords
color
pairs
harmony
score
theme
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
CN201810894794.9A
Other languages
Chinese (zh)
Other versions
CN109087364A (en
Inventor
杨柏林
魏天祥
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.)
Zhejiang Gongshang University
Original Assignee
Zhejiang Gongshang 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 Zhejiang Gongshang University filed Critical Zhejiang Gongshang University
Priority to CN201810894794.9A priority Critical patent/CN109087364B/en
Publication of CN109087364A publication Critical patent/CN109087364A/en
Application granted granted Critical
Publication of CN109087364B publication Critical patent/CN109087364B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • 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
    • G06T2207/10024—Color image

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Spectrometry And Color Measurement (AREA)
  • Color Image Communication Systems (AREA)

Abstract

The invention discloses a method for predicting color harmony degree according to a mode. The method is based on a large amount of data sets, according to the input color subjects of five different colors, the color pair score obtained after the color subjects are divided by the five color subjects is obtained through a mode method, then the color pairs are replaced to the five color subjects, and the predicted harmony score is obtained through the mode method, so that the concept of harmony of colors is quantized into the score. The method can meet the requirement that the formed color table is diversified, the obtained harmony score has high accuracy, and meanwhile, the method can be used in scenes with a plurality of color collocation in real life.

Description

Method for predicting color harmony degree according to mode
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a method for predicting color harmony degree according to a mode.
Background
Color matching plays an indispensable role in many aspects of life. With the development of multimedia technology and software design technology, designers can rapidly design color theme schemes with different styles aiming at different object groups, and can not distinguish and manually extract collocate colors through professional knowledge and naked eyes like the prior art, so that the efficiency is greatly improved. However, how to grasp the harmony degree of the colors needing to be matched and the harmony degree of the colors are always a problem to be solved. Since the harmony degree of the colors directly affects the color matching result, it is important to find a method capable of quantifying and predicting the concept of the harmony degree of the colors.
The traditional color harmony prediction method is based on a color harmony theory, analyzes some simple color characteristics such as hue, brightness, chroma and the like in a small amount of sample colors by a mathematical modeling method, and then calculates the harmony value by a formula. Such a method obtains relatively single results because of fewer color features to consider. In recent years, researchers train a data set with a large number of color features to achieve a prediction effect through a LASSO regression method based on big data in machine learning. Although the method gets rid of the defect of single result, the experimental result becomes more objective and more robust through the training of big data. However, the harmony of the predictions of this method is not accurate enough. Therefore, it is very important to find a prediction method capable of representing color harmony more accurately.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method for predicting the color harmony degree according to the mode.
The method specifically comprises the following steps:
the method comprises the following steps: color themes obtained by crowdsourcing under Amazon Mechanical Turk are adopted, and each color theme is formed by combining five different color blocks. In the color theme of different five colors, the color relationship between each other is obtained based on two adjacent color pairs. Therefore, all the color themes in the data set are divided into pairwise adjacent color pairs, that is, a five-color theme contains four divided pairwise adjacent color pairs.
Step two: and assigning the color and the harmony score corresponding to each color theme in the data to the corresponding four color pairs, namely assigning the scores of the original five-color theme to the four different segmented color pairs respectively, wherein the scores of the four color pairs are the same.
Step three: all color pairs are arranged according to the positive sequence of the color components in the RGB color space, then the color pairs in which the color components are repeated are removed and the harmony scores of the color pairs are kept, so that the remaining color pairs after the deduplication contain all the same scores as the color components of the color pairs before the deduplication.
Step four: and carrying out mode operation on all the color pairs, namely extracting the value with the highest occurrence number of the harmony score in each color pair. r isi pairAs a new score for the ith color pair, n represents the number of repeated color pairs and the harmony score:
ri pair＝mode(r1,r2,r3,…,rn)
step five: all color pairs and their harmonious values ri pairSubstituting the positions of the original five-color themes into the color themes to enable each five-color theme to contain four same color pairs and corresponding harmony scores of four different color pairs.
Step six: taking out the harmony score with the most occurrence times in the five-color theme as the final score r 'of the jth color theme by a mode value taking method'jThereby, the prediction effect is achieved:
Figure GDA0003163884170000021
the invention has the beneficial effects that: compared with the prior art, the method adopts a totally different mode-based method in statistics to solve the problems of color harmony prediction, greatly improves the accuracy of the result and reduces the error.
Detailed Description
The present invention will be further illustrated by the following examples
The technical scheme adopted by the invention for solving the technical problems is as follows:
the method comprises the following steps: the invention adopts color themes obtained by crowdsourcing under Amazon Mechanical Turk, and each color theme is formed by combining five different color blocks. Research shows that in the color theme of different five colors, the mutual color relationship is obtained based on two adjacent color pairs. Therefore, all the color themes in the data set are divided into pairwise adjacent color pairs, that is, a five-color theme contains four divided pairwise adjacent color pairs.
Step two: and assigning the color and the harmony score corresponding to each color theme in the data to the corresponding four color pairs, namely assigning the scores of the original five-color theme to the four different segmented color pairs respectively, wherein the scores of the four color pairs are the same.
Step three: all color pairs are arranged according to the positive sequence of the color components in the RGB color space, then the color pairs in which the color components are repeated are removed and the harmony scores of the color pairs are kept, so that the remaining color pairs after the deduplication contain all the same scores as the color components of the color pairs before the deduplication.
Step four: and carrying out mode operation on all the color pairs, namely extracting the value with the highest occurrence number of the harmony score in each color pair. r isi pairAs a new score for the ith color pair, n represents the number of repeated color pairs and the harmony score:
ri pair＝mode(r1,r2,r3,…,rn)
step five: all colors are paired and summedValue r of harmonic degreei pairSubstituting the positions of the original five-color themes into the color themes to enable each five-color theme to contain four same color pairs and corresponding harmony scores of four different color pairs.
Step six: taking out the harmony score with the most occurrence times in the five-color theme as the final score r 'of the jth color theme by a mode value taking method'jThereby, the prediction effect is achieved:
Figure GDA0003163884170000031
to illustrate the effectiveness of the present invention, the Mean Square Error (MSE) and Mean Absolute Error (MAE), as well as the pearson correlation coefficient R (correlation), were used to measure the experimental results and compared to the predicted results generated by LASSO regression. The final results are shown in the following table:
TABLE 1 comparison of color harmony prediction results
MAE MSE R
Mode prediction result 0.0227 0.2435 78.69％
LASSO prediction results 0.0518 0.4456 73.61％
In conclusion, the invention finally obtains an interval score to describe the harmony degree by inputting a group of multi-color-block color themes with different colors, thereby achieving the purpose of quantifying the fuzzy concept of color harmony and enabling a user to intuitively distinguish the harmony degree of the colors through the score.

Claims (1)

1.根据众数对颜色和谐程度进行预测的方法，其特征在于该方法包含如下步骤：1. the method for predicting the degree of color harmony according to the mode is characterized in that the method comprises the steps: 步骤一：采用在Amazon Mechanical Turk下众包得到的颜色主题，每种颜色主题都是由五种不同的色块组合而成的；在不同五种颜色的颜色主题中，相互之间的颜色关系都是基于两两相邻的颜色对来得到的；所以将数据集中所有的颜色主题分割成两两相邻的颜色对，即一个五色的颜色主题中含有四个被分割后两两相邻的颜色对；Step 1: Use the color themes crowdsourced under Amazon Mechanical Turk, each color theme is composed of five different color blocks; in the color themes of different five colors, the color relationship between each other All are obtained based on pairwise adjacent color pairs; therefore, all the color themes in the dataset are divided into pairwise adjacent color pairs, that is, a five-color color theme contains four pairs of adjacent color themes after being divided. color pair; 步骤二：将数据中每个颜色主题对应的颜色和谐度分数赋给对应的四个颜色对，即原来五色颜色主题的分数分别赋给分割后的四个不同的颜色对，且四个颜色对分数相同；Step 2: Assign the color harmony score corresponding to each color theme in the data to the corresponding four color pairs, that is, assign the scores of the original five-color color theme to the divided four different color pairs, and the four color pairs same score; 步骤三：将所有的颜色对按照RGB颜色空间中的颜色分量正序排列，然后去除其中颜色分量重复的颜色对，并且保留其和谐度分数；Step 3: Arrange all the color pairs in the positive order of the color components in the RGB color space, then remove the color pairs in which the color components are repeated, and retain their harmony scores; 步骤四：对所有的颜色对进行取众数操作，即提取每个颜色对分数中和谐度分数出现次数最多的值；ri pair作为第i个颜色对的新分数,n代表重复的颜色对和谐度分数的数量：Step 4: Perform the mode operation on all color pairs, that is, extract the value with the most occurrences of the harmony score in each color pair score; r i pair is used as the new score of the ith color pair, and n represents the repeated color pair. Number of Harmony Scores: ri pair＝mode(r1,r2,r3,...,rn)r i pair = mode(r 1 ,r 2 , r 3 ,...,rn ) 步骤五：将所有的颜色对及其和谐度的值ri pair按照原有五色颜色主题的位置代入到颜色主题中，使每一个五色颜色主题都含有四种与之前相同的颜色对以及对应的四个不同的颜色对的和谐度分数；Step 5: Substitute all color pairs and their harmony value r i pair into the color theme according to the position of the original five-color color theme, so that each five-color color theme contains four same color pairs as before and the corresponding Harmony scores for four different color pairs; 步骤六：通过众数取值的方法取出五色颜色主题中出现次数最多的和谐度分数作为第j个颜色主题的最终分数rj′，从而达到预测效果：Step 6: Take out the harmony score with the most occurrences in the five-color color theme by the method of mode value as the final score r j ′ of the jth color theme, so as to achieve the prediction effect:
Figure FDA0003204138430000011
Figure FDA0003204138430000011
CN201810894794.9A 2018-08-08 2018-08-08 Method for predicting color harmony degree according to mode Active CN109087364B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810894794.9A CN109087364B (en) 2018-08-08 2018-08-08 Method for predicting color harmony degree according to mode

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810894794.9A CN109087364B (en) 2018-08-08 2018-08-08 Method for predicting color harmony degree according to mode

Publications (2)

Publication Number Publication Date
CN109087364A CN109087364A (en) 2018-12-25
CN109087364B true CN109087364B (en) 2021-11-12

Family

ID=64833837

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810894794.9A Active CN109087364B (en) 2018-08-08 2018-08-08 Method for predicting color harmony degree according to mode

Country Status (1)

Country Link
CN (1) CN109087364B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109903349B (en) * 2019-01-28 2021-07-02 浙江工商大学 A Color Harmony Prediction Method Based on Maximum Likelihood Estimation

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8085276B2 (en) * 2006-11-30 2011-12-27 Adobe Systems Incorporated Combined color harmony generation and artwork recoloring mechanism
US10026192B2 (en) * 2013-10-18 2018-07-17 Ford Global Technologies, Llc Color harmony verification system
CN106775747B (en) * 2016-12-28 2020-03-24 Tcl集团股份有限公司 Color configuration method and device
CN107390976B (en) * 2017-07-26 2020-09-08 上海展扬通信技术有限公司 Icon display processing method and electronic equipment

Also Published As

Publication number Publication date
CN109087364A (en) 2018-12-25

Similar Documents

Publication Publication Date Title
Zhong et al. Research on deep learning in apple leaf disease recognition
Yu et al. A unified pruning framework for vision transformers
CN110188225B (en) Image retrieval method based on sequencing learning and multivariate loss
CN104517122A (en) Image target recognition method based on optimized convolution architecture
CN103793702A (en) Pedestrian re-identifying method based on coordination scale learning
CN111028203B (en) CNN blind image quality evaluation method based on significance
CN109087364B (en) Method for predicting color harmony degree according to mode
Gu et al. Leaf species recognition based on VGG16 networks and transfer learning
CN108510483B (en) A computationally generated color image tampering detection method using VLAD coding and SVM
CN115713634B (en) A Color Matching Evaluation Method Combining Similarity Measurement and Visual Perception
CN108961346A (en) The method that color harmony degree is predicted based on BP neural network
CN111881756A (en) Waste mobile phone model identification method based on convolutional neural network
CN110163855A (en) A kind of color image quality evaluation method based on multichannel depth convolutional neural networks
CN114821293A (en) Prototype spectral set generation method based on super-pixels
CN108428234A (en) Interactive Segmentation performance optimization method based on image segmentation result evaluation
CN109903349B (en) A Color Harmony Prediction Method Based on Maximum Likelihood Estimation
CN103646393B (en) Oil painting base colors of pigment Forecast of Spectra method
Utami et al. The EfficientNet performance for facial expressions recognition
AU2021101210A4 (en) An Endmember Extraction Method Of Hyperspectral Image
CN105335980A (en) Method suitable for image SIFT feature matching in converting color image to brightness image
CN111738298B (en) A classification method for MNIST handwritten digit data based on depth-width variable multi-kernel learning
TW201545119A (en) Data clustering method applicable to color images
Mi Face image recognition via collaborative representation on selected training samples
Kuo et al. Chroma Component Generation of Gray Images Using Multi-Scale Convolutional Neural Network
CN110750672A (en) Image retrieval method based on deep metric learning and structural distribution learning loss

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