WO2010015196A1

WO2010015196A1 - Device and method for testing fabric color

Info

Publication number: WO2010015196A1
Application number: PCT/CN2009/073087
Authority: WO
Inventors: 克汗杜尔·阿思曼达; 白求智; 胡金莲; 辛斌杰
Original assignee: 香港纺织及成衣研发中心
Priority date: 2008-08-04
Filing date: 2009-08-04
Publication date: 2010-02-11
Also published as: CN101788341B; CN101788341A

Abstract

A device for testing the fabric color comprises a holder, a scanner, a computer and a color analysis system, wherein the color analysis system used to analyze the color feature of a scanned object is stored in the computer. Besides, a method for testing the fabric color comprises scanning the image information of a scanned object fixed on a holder by a scanner, inputting the scanned image information into a computer, converting the inputted scanned object image information from RGB value into CIE*l*a*b value by the computer, setting the related color tolerance information by an user at the user-friendly interface of the computer, conducting color classifying, color searching and auto-matching on the CIE*l*a*b value of the scanned object and a fabric standard color sample database module in the computer database, outputting the standard fabric information conformed with the scanned object information.

Description

Apparatus and method for detecting fabric color

Technical field

According to the non-linear color model, the present invention can accurately obtain device independent color coordinates from the scanned fabric image, thereby performing color analysis. To characterize a commercial scanner, only a standard fabric-based model is required, eliminating the need for traditional expensive equipment such as colorimeters, spectrophotometers, multi-source cameras, and more. The methods and systems can be used as quality control tools in the field of surface color industry, such as testing the quality of fabrics, paints, paper, leather, plastics, and prints.

Background technique

The color of the RGB format acquired by the scanner is related to the type of illumination and the characteristics of the optical filter that can divide the visible wavelength into multiple color components. Conventional desktop scanners have poor color rendering. All common imaging systems, including digital cameras, scanners, and CRTs, use a similar 3-band color coding scheme for color reproduction because their purpose is to input information into the human visual system. The triple sensor is a many-to-one reflection spectroscopy instrument. We have not been able to reconstruct a perfect original spectrum from a simplified image so far, nor can it restore the precise surface reflection characteristics of a surface. The best approach is to use a three-parameter description to illustrate the reflection coefficient and illumination properties.

CRT (Cathode Ray Tube) displays, scanners, color printers, and prepress systems have a high degree of equipment dependencies: If you don't consider the characteristics of different devices, qualitative standards, colors, and surface contours, put RGB values from one When the device is transferred to another device, it is likely that the result will get a different color. This situation is very prominent in textiles, so a system with high precision is required. The device color reproduction features are appropriately processed, such as surface contours, exposed portions of image distribution noise, etc., in order to correctly evaluate pixel-level RGB values. The color feature can be defined as the relationship between the device "color space" and the CIE color measurement system such as the color conversion space, or it can be defined as a mathematical model consisting of a set of equations. The human vision system, which uses three-band color coding, exhibits extraordinary adjustments for sudden illumination changes. CIE uses another set of well-designed original stimuli with specific properties, defined as X, Y, and Z, including the corresponding tristimulus values X, Y, and Z, and the color matching function.

b _ b _

X = ^ K.E( ).X ( ).r( ).d Υ = ^ Κ.Ε(λ).Ϋ(λ) .Γ(λ).Μ b

/ = κ κ.Ε{λ)Ι{λ).ν{λ).άλ

. (1)

(in) = energy distribution function, X (person) = color matching function of the standard observer, Ι (λ) = reflection coefficient of the measured object.

Berns and Shyu proposed a color mixing method based on Beer-Bouger and Kubelka-Munk theory and scanning signals from 1994 to 1995. They use polynomial regression. In 1996, Hardeberg et al. proposed an analytical method based on third-order polynomial regression. They use the CIE color space value and the 8.7 scan values of the IT8.7/2 calibration color card. It turns out that the polynomial regression is better than the results obtained by other methods.

Finlayson and Drew mentioned that color values measured by color devices such as scanners, color copiers, and color cameras must be converted to colorimetric "tri-stimulus" values to characterize them in the form of device-independent colors. They also propose a constrained regression method that finds the minimum point of the sum of the squared distances of the RGB data and the corresponding XYZ tristimulus values, even if the calibration color is incomplete. Do it.

Kang pointed out in 1997 that many scientists have successfully applied regression methods to convert the scanned RGB values into colorimetric values using the IT8.7/2 calibration color card and the Sharp JX 450 scanner. He compared the results obtained with different order polynomials and different standard light sources.

In 2001, Noriega et al. used positive density and negative density measurements to measure the RGB values of the scan and the CIE XYZ color space values to determine the properties of the scanner. They therefore conclude that the deviation of the colorimetric values scanned by the scanner depends on the properties of the device and the associated color management system. In a model based on polynomial regression, the more color space contains samples, the less error there is at the gamut boundary. The number of samples at the domain boundaries in the commonly used test indicators like ISO 12640 and 12641 is very limited. However, in 2000, Green proposed a new test metric that defines the media gamut boundary, using a second-order polynomial regression. Although recent studies have mentioned that the colorimetric characterization methods of cameras include polynomial regression, neural networks, and lookup tables, polynomial regression is actually the most appropriate method in scanner characterization. The main limiting factor in colorimetric characterization is the need to combine illumination and observation functions. A recent study found that colorimetric characterizations are more accurate than spectroscopy at the same source.

U.S. Patent No. 2,005,018,191 A1 proposes an 11-variable polynomial model RGBXYZ with camera reflectance evaluation. However, the 23-variable model is ideal for reducing errors, which is also best for RGB-to-CIE L*a*b and material-based features. Therefore, typical fabric standards based on standard color cards are proposed to characterize the scanner color of all ATSM illumination and observer functions under a particular variable model. The best choice of color characterization maps is how many (or which) samples will affect characterization performance. Cheung and Westland passed the experimental verification in 2006 and proposed the standard Gretag Macbeth color. The card, the Gretag Macbeth color card DC (DC) color uses only a subset of the 1269 Munsell surface color optimized color samples. In U.S. Patent 7,230,707 B2, it is required that a spectrophotometer should be used in conjunction with a camera that accurately characterizes the color of the material, and that such material has typical surface reflection properties.

Summary of the invention

The present invention relates to an apparatus and method for analyzing the color of a fabric using a scanner. Most of the prior art color analysis uses expensive equipment such as a colorimeter, a spectrophotometer, and a multi-source camera. The problem to be solved by the present invention is to provide an apparatus and method for low cost color inspection. The technical solution adopted by the invention is: adopting a color characteristic analysis system module, according to the nonlinear color model, the device independent color coordinates can be accurately obtained from the scanned fabric image, thereby performing color analysis. To characterize a commercial scanner, you only need a standard fabric-based model. The methods and systems can be used as quality control tools in the field of surface color industry, such as testing the quality of fabrics, paints, paper, leather, plastics, and prints. The apparatus and method for detecting the color of a fabric according to the present invention comprises: a fixture: mounted on a scanner for fixing a detection object; a scanner: acquiring image information data of the detection object; and a computer: connected to the scanner for processing detection Object image information data; color analysis system: stored in a computer for analyzing the color characteristics of the scanned object. In the apparatus and method for detecting the color of a fabric according to the present invention, the color analysis system includes a variable nonlinear model module for converting scanned image information from RGB values to CIE*l*a*b values. In the apparatus and method for detecting the color of a fabric according to the present invention, the multivariable nonlinear model is optimal when 23 variables are used.

J[(L * -Lp*) + (a * -ap*) + (b * -bp*)]

ERROR Error = In the apparatus and method for detecting the color of a fabric according to the present invention, the 23 variables are attribute information values of the fabric. In the apparatus and method for detecting the color of a fabric of the present invention, the color analysis system includes a fabric standard color sample database module for color sorting, color search, and automatic matching with the scanned image information. In the apparatus and method for detecting the color of a fabric according to the present invention, the color analysis system includes a user-friendly interface module. The user sets the color tolerance condition in the user-friendly interface, and the multi-variable nonlinear model module converts the scanned object CIE*l*a*b value and the fabric standard color sample information in the fabric standard color sample database module according to the set color capacity condition. Perform color classification, color search, and automatic matching; the user-friendly interface obtains and displays device independent CIE color coordinates according to the RGB values of the detected objects. In the apparatus and method for detecting the color of a fabric according to the present invention, the fabric standard color sample database module is 153 representative fabrics selected according to typical surface contour features of common fabrics and robustness analysis results of polynomial regression models. According to the standard color samples, those fabric materials that are very transparent, smooth and complex in surface are excluded. After the standard color samples are selected, the selected standard color samples are collected by a spectrophotometer, and the collected data is the fabric. Standard color sample database module. In the apparatus and method for detecting the color of a fabric according to the present invention, the jig for fixing the detection object functions as a positioning object to be detected and is mounted on the scanner. In the apparatus and method for detecting the color of a fabric according to the present invention, the method comprises the following steps: S1: scanning, by a scanner, image information of a scanned object fixed on the fixture;

S2: input image information of the scanned object to be scanned into a computer on the scanner;

S3: The computer converts the input scanned object image information from the RGB value to the CIE*l*a*b value;

S4: The user sets relevant color information in the user-friendly interface of the computer, such as: DE, DL*, Da*, Db*, Dc* &Dh*;

S5: Perform color classification, color search and automatic matching on the scanned object CIE*l*a* b value obtained by S3 and the fabric standard color sample database module in the computer database, and output and detect the object according to the user-defined color capacity condition in S4. Standard fabric information that matches the information. In the apparatus and method for detecting the color of a fabric according to the present invention, the color analysis system predicts to be based on XYZ/CIE L*a*b by filling a basic database of XYZ/CIE L*a*b* values of the substrates. * Matching dye formulations.

The following benefits can be obtained by using the apparatus and method for detecting the color of the fabric provided by the present invention - device independent CIE color coordinates under multiple light sources and observers (ASTM color measurement standards) can be scanned from the measured area (ROI) to RGB The value is obtained without using an expensive colorimeter, spectrophotometer, etc., and the system for fabric color analysis is suitable for different materials such as wool, silk or other natural or synthetic fibers, after user-defined color tolerance, The database can color and automatically match these different materials. The method can be applied to many commercial desktop scanners in the industrial field of surface color applications, including textiles, drawings, paper, leather, plastics, printed matter and the like.

DRAWINGS

The present invention will be further described below in conjunction with the accompanying drawings and embodiments, in which:

Figure 1 shows a comparison table of high-order variable results for scanning the Xrite color card verification model with FABRIC EYE D2000;

Figure 2 shows a comparison table obtained by using FABRIC EYE D2000 to scan high-order variables based on fabric characterization models of 153 fabrics;

Figure 3 shows a table of results obtained by scanning a 52 random fabrics based on a 153 fabric-based characterization model with a FABRIC EYE D2000 scan;

4-6 are schematic diagrams showing the system configuration of the fabric color analysis by the color characterization of the scanner of the present invention; FIG. 7 is a schematic diagram showing the working principle of the system; FIG. 8 is a user-friendly ROI color analysis of the scanned fabric. Interface diagram;

Replacement page (Article 26) 9 and 10 are schematic diagrams showing the color characterization results of two different scanners D-2000 and Canon 8887 for the X-rite SGA color card 140;

Figure 11 shows a comparison of the results obtained by scanning the Xrite color card with the FABRIC EYE D2000 using the RGB L*a*b* model and the RGBXYZ 23 Coffs.

Figure 12 shows a comparison of the results of the higher-order variables obtained by scanning the Xrite color card with the FABRIC EYE D2000;

Figures 13 and 14 show the distribution of the X-rite color card and the standard L*, a*, b* (D65 64) of the 153 fabric;

Figure 15, Figure 16, and Figure 17 show the predictive experimental values of CIE L*, a*, b* (D65-64) obtained by fabric color characterization of the D-2000 scanner.

detailed description

4-6 are schematic diagrams showing the composition of a system for characterizing fabric color by a scanner according to the present invention. The system includes: a personal computer shown in Fig. 4 and a scanner connected to a personal computer, a fabric template with a color card sample shown in Fig. 5, and 153 standard fabric color samples shown in Fig. 6. A fabric template with a color card sample is used to calibrate the scanner. The polynomial regression characterization technique uses a standard X-write color card to evaluate the color of the fabric. The following is a detailed description of the results of the device independent color CIE parameters when connecting three typical scanners to the spectrophotometer under different lighting conditions. . However, it is not appropriate to use the resulting characterization model to verify the CIE color coordinates of a solid color fabric. Therefore, it is recommended to use typical fabric standards based on the color card domain (shown in Figures 13 and 14) to characterize scanners that analyze conventional solid color fabrics. Taking into account the typical surface characteristics of the fabric and the robustness of the polynomial regression model, after excluding typical samples (very transparent, very smooth and complexly contoured), 153 solid-colored fabrics (shown in Figure 6) were selected. It is suitable to use RGB based on 23 variables.

Replacement of 100 (Article 26) Model of L*a*b* fabric.

Figure 7 shows a schematic diagram of the working principle of the system. Key parameter characteristics have an effect on the color display. First, the bandpass error of the spectrophotometer should be checked. The tristimulus values and L*, a*, b* must be adjusted to 400- by reference to ASTM E308-06 (Standard for Calculating Object Colors with the CIE System). 700nm. After the first step is completed, a nonlinear model based on 23 variables (Equation 2) can be obtained by optimizing each source and observing the target error function (Equation 3) of the condition ό r.o r 6 Γ. Γg rl b.

1/2

J [(L * -Lp*) + (a * -ap*) + (6 * -bp*)]

Figure 8 is a schematic illustration of a user friendly interface for color analysis of the ROI of the fabric being scanned. After characterizing and optimizing the nonlinear model, the user can select each source and observer in a user-friendly interface and analyze the color parameters by selecting the region to be tested (ROI). This gives the device independent color parameters that can be used in all subsequent steps (color communication, presentation, rendering, color categorization, and matching). Replacement page (Article 26) Figures 9 and 10 show the results of color characterization of the X-rite SGA color card 140 using two different scanners D-2000 and Canon 8887.

Figure 11 shows a comparison of the results of the RGB L*a*b* and RGBXYZ 23 Coffs. models for the FABRIC EYE D2000 scan Xrite color card. For each scanner, RGB L*a*b* imaging is better than RGBXYZ imaging. In all cases, the RGBL*a*b* model predicted color differences (DE average, maximum, minimum, and standard deviation) were smaller than the RGBXYZ model.

Figure 12 shows a comparison of the results of high-order variables for scanning Xrite color cards with the FABRIC EYE D2000. Errors such as DE average, maximum, minimum, and standard deviation decrease as polynomial variables increase. When measuring fabric samples, the error is no longer significantly improved with more than 23 variables - so this is considered to be the best solution.

Figures 1, 2, and 3 verify the results of the facts. Figure 1 shows the comparison of the higher-order variables of the model with the Xrite color card scan using the FABRIC EYE D2000. Figure 2 shows a comparison of higher-order variables using a FABRIC EYE D2000 scan using a 153 fabric-based fabric characterization model. Figure 3 shows the results of a verification of the high-order variables of 52 random fabrics based on 153 fabric models scanned with the FABRIC EYE D2000.

Figures 13 and 14 show the L*, a*, b* (D65 64) distribution of the X-rite color card and the 153 standard fabrics. These fabric characterization techniques, typical fabric standards based on the color card domain, are considered to be an effective method for analyzing commonly used fabrics.

Figure 15, Figure 16, and Figure 17 show the fabric color characterization D-2000 scanner using the technology pre-replacement page (Rule 26) The measured distribution results of L*, a*, b* (D6564).

Replacement page (Article 26)

Claims

Rights request

A device and method for detecting a color of a fabric, the device comprising: a jig: mounted on a scanner for fixing a test object;

Scanner: acquiring image information data of the detection object;

Computer: connected to the scanner, configured to process image data of the detection object;

Color Analysis System: Stored in a computer to analyze the color characteristics of scanned objects.

2. Apparatus and method for detecting the color of a fabric according to claim 1, wherein said color analysis system comprises variable nonlinearity for converting scanned image information from RGB values to CIE*l*a*b values. Model module.

3. Apparatus and method for detecting the color of a fabric according to claim 2, wherein said multivariable nonlinear model is optimal when 23 variables are used.

J[(L * -Lp*) + (a * -ap*) + (b * -bp*)]

ERROR error =

4. Apparatus and method for detecting the color of a fabric according to claim 3, wherein the 23 variables are attribute information values of the fabric.

5. Apparatus and method for detecting the color of a fabric according to claim 1 wherein said color analysis system includes a fabric standard color sample database module for color sorting, color searching, and automatic matching with scanned image information.

6. The apparatus and method for detecting the color of a fabric according to claim 1, wherein the color is divided into The analysis system includes a user-friendly interface module, the user sets the color tolerance condition in the user-friendly interface, and the multi-variable nonlinear model module converts the scanned object CIE*l*a*b value and the fabric standard color in the fabric standard color sample database module. The sample information is color-classified, color-searched, and automatically matched according to the set color-capacity conditions;

The user friendly interface derives and displays device independent CIE color coordinates based on the RGB values of the detected objects.

7. The apparatus and method for detecting the color of a fabric according to claim 1, wherein the fabric standard color sample database module is selected based on a typical surface contour feature of a common fabric and a robustness analysis result of a polynomial regression model. 153 kinds of representative fabric standard color samples, those very transparent, smooth and complex surface materials are excluded; after selecting the standard color samples, the selected standard color samples are collected by spectrophotometer. The collected data is the fabric standard color sample database module.

8. The apparatus and method for detecting the color of a fabric according to claim 1, wherein the fixture for fixing the detection object functions as a positioning object to be detected and is mounted on the scanner.

9. Apparatus and method for detecting the color of a fabric, the method comprising the steps of: S1: scanning, by a scanner, image information of a scanned object fixed on the fixture;

S5: The CIE*l*a* b value of the scanned object obtained by S3 and the fabric standard color in the computer database The sample database module performs color classification, color search, and automatic matching, and outputs standard fabric information that matches the detected object information according to the user-defined color capacity condition in S4.

10. Apparatus and method for detecting the color of a fabric according to claim 1, wherein said color analysis system predicts to be based on XYZ/ by filling a basic database of XYZ/CIE L*a*b* values of the substrates. CIE L*a*b* matched dye formulation.