CN109325934B - Method and system for automatically analyzing and evaluating fabric glossiness - Google Patents
Method and system for automatically analyzing and evaluating fabric glossiness Download PDFInfo
- Publication number
- CN109325934B CN109325934B CN201810777296.6A CN201810777296A CN109325934B CN 109325934 B CN109325934 B CN 109325934B CN 201810777296 A CN201810777296 A CN 201810777296A CN 109325934 B CN109325934 B CN 109325934B
- Authority
- CN
- China
- Prior art keywords
- fabric
- glossiness
- image
- grade
- gray
- 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
Links
Images
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
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Quality & Reliability (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Treatment Of Fiber Materials (AREA)
Abstract
The invention discloses a method and a system for automatically analyzing and evaluating fabric glossiness, wherein the method comprises the following steps: a glossiness grade dividing step, namely quantifying the glossiness of the fabric into N grades; establishing a corresponding relation between the glossiness levels and the characteristic parameters, namely selecting M characteristic parameters of the image, wherein each characteristic parameter is provided with one or more search intervals, and each search interval corresponds to one glossiness level or a plurality of glossiness levels; and a fabric glossiness evaluation step, namely acquiring a fabric image of the fabric to be evaluated and M characteristic parameters of the fabric image, and finding out the glossiness grade of the fabric to be evaluated according to the M characteristic parameters. The method of the invention obtains a stable fabric image in a light stable environment by using the principle of optical imaging and computer vision, establishes the relation between the characteristic parameters of the image and the glossiness grade, obtains the characteristic parameters of the fabric image and the fabric glossiness grade, and realizes the objective evaluation conforming to the fabric glossiness effect perceived by human eyes.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method and a system for automatically analyzing and evaluating fabric glossiness.
Background
The fabric gloss is the main component of the fabric visual style-appearance and is an important content for evaluating the fabric appearance quality. The test evaluation of the fabric gloss not only has important significance on the improvement of the fabric visual effect and the improvement of the finished product style, but also is the basis of the fabric surface gloss design, the process improvement and the quality control. Therefore, the research on the fabric gloss test and evaluation technology is particularly important.
Currently, there are two main methods for evaluating fabric gloss: sensory evaluation and instrumental testing. The sensory evaluation method is a subjective evaluation of the relative quality of the fabric gloss by the visual sense of a human. Generally, a certain number of inspectors with abundant practical experience and strong judgment capability are concentrated, and the gloss is ranked according to the grade by visual judgment. The method for evaluating the fabric gloss is simple, convenient, rapid and visual. However, this method is susceptible to the environment of inspection and human factors, and the evaluation results are related to experience and psychological state, and it can only relatively compare the gloss of the fabric and cannot obtain quantitative values. Instrumental methods are objective evaluations of fabric gloss and are quantified by various physical quantities related to reflected light, i.e. the amount of gloss is measured with a relevant instrument. The starting point of the instrument test is to replace human eye sense and form a quantitative mechanism of gloss description, but the physical quantity of the fabric gloss given by the existing instrument test method is indirect representation for reflecting the fabric gloss visual attribute, and although the physical quantity has a certain correlation with the appearance style of the fabric gloss, the physical quantity cannot replace the evaluation connotation of human eye sense. In general, based on different fabric gloss theories, the indexes of various special textile gloss instruments developed at home and abroad are different. To summarize, it can be divided into: (a) the surface gloss intensity is reflected by: regular reflection light intensity, diffuse reflection light intensity, reflectivity, contrast glossiness and two-dimensional contrast glossiness; (b) the gloss distribution (directional difference) is reflected by: angle-variable gloss distribution, three-dimensional contrast gloss, Jeffries contrast gloss; (c) the following are reflected in the composition of the surface reflection light and the internal reflection light (difference between the inside and the outside): the gloss of polarized light; and so on. Although the test indexes are many, the method is from different theories and is suitable for different test methods. The merits of each method are controversial in the industry, the regulations of the application range of some instruments are reversed with the judging sequence, and the reliability of the test result is questioned.
As early as 2001, Yao Muscou et al suggested that a correct study method for evaluating fabric gloss should be a method that objectively combines subjectivity (simulating subjectivity to some extent) to achieve subjective and objective agreement. That is, the correct evaluation method should be an objective evaluation technique that conforms to the human eye's visual perception, since the fabric's glossy feeling is realized by human feeling and perception, which is the evaluation of human subjective perception information of gloss. Therefore, theoretical and testing approaches closer to the characterization of human subjective perception of gloss information must be sought. Computers have been popular more than two decades ago, and the research on fabric gloss testing still extends to the improvement of gloss meters based on traditional physical principles, and the computers are only introduced into the research on some fabric gloss meters as data processing means. The theory and practice of computer vision and image recognition analysis technology are mature, the technology has the advantage of quantitatively describing the performance of a visual surface, and the technology is a research approach combining objectivity and subjectivity.
In view of the fact that no ideal objective evaluation technology of fabric gloss according with human eye vision feeling exists at home and abroad at present, a need exists for the invention of obtaining a stable fabric true color image in a light stable environment by using computer vision, distinguishing gloss components from a specific fabric surface image by using a computer, describing gloss states, evaluating gloss, converting the connotation of visual styles (such as: dull, matt, astigmatic, shiny) formed by the fabric surface gloss observable by human eyes into image data, depicting the image data in a form of digitized characteristic parameters, and describing the relationship between quantitative characteristics of the image data and the fabric gloss (the result is shown in table 1), so as to replace the uncertainty evaluated by human eyes and the defects measured by the existing instruments. The objective evaluation technology based on computer vision and according with human visual perception can realize the simulation evaluation of the fabric luster.
Disclosure of Invention
The invention aims to solve the problems of uncertainty of human eye evaluation of the existing textile fabric gloss evaluation and the defects of measurement of the existing instrument. The method and the system for automatically analyzing and evaluating the fabric gloss appearance are provided, and the problems can be solved. The method comprises the steps of configuring a spherical seat according to an optical imaging principle to coat a fabric sample, obtaining a stable fabric hemispherical image in a light stable environment by using computer vision, compiling analysis software, and obtaining the fabric gloss grade and the performance description language from the fabric image gloss sense so as to realize objective evaluation of the fabric gloss according with human eye vision sense.
In order to solve the technical problems, the invention adopts the following technical scheme:
an automatic analysis and evaluation method for fabric glossiness comprises the following steps:
a glossiness grade dividing step, namely quantifying the glossiness of the fabric into N grades, wherein N is an integer greater than 1;
establishing a corresponding relation between the glossiness levels and the characteristic parameters, namely selecting M characteristic parameters of the image, wherein each characteristic parameter is provided with one or more search intervals, each search interval corresponds to one glossiness level or a plurality of glossiness levels, and M is an integer greater than 1;
and a fabric glossiness evaluation step, namely acquiring a fabric image of the fabric to be evaluated, processing the fabric image, acquiring M characteristic parameters of the fabric image, and searching the glossiness grade of the fabric to be evaluated from the search interval according to the M characteristic parameters.
Further, in the step of evaluating the glossiness of the fabric, the step of processing the fabric image comprises the following steps:
(11) converting the colorful fabric image into a gray image;
(12) and converting the gray level image into a binary image.
Further, in step (11), after the color fabric image is converted into a weighted average value gray image and R, G, B three-component gray images, the gray contrast is enhanced by enhancing the darkness of the four parallel gray images, and finally the gray image with the highest brightness and darkness contrast is selected as the operation object in step (12).
Further, the M characteristic parameters include a combination of any two or more of a nominal filling degree, a gray scale contrast, a gloss filling degree range, a gray scale co-occurrence matrix contrast, and a saturation contrast.
Further, the nominal filling degree K is calculated by: area S occupied by white pixels in binary imageWWith the total area S of the imaged area of the fabric in the fabric imageZThe ratio of the areas of (a):
the calculation method of the gray contrast C comprises the following steps:
carrying out gray difference calculation on the gray image to obtain a gray contrast C;
the method for calculating the extremely poor gloss filling degree phi comprises the following steps:
area S occupied by white pixels in binary imageWWith the total area S of the rectangular window in the fabric imageRThe ratio of the areas of (a):
the method for calculating the contrast CONN of the gray level co-occurrence matrix comprises the following steps:
carrying out image texture analysis on the gray level image to obtain a gray level co-occurrence matrix contrast CONN;
saturation contrast SCThe calculation method comprises the following steps:
obtaining an S component image in a fabric image of an HSI color space, and obtaining a gray contrast of the S component image as a saturation contrast SC。
Further, in the step of evaluating the fabric glossiness, the method for finding out the glossiness grade of the fabric according to the M characteristic parameters of the fabric comprises the following steps:
(21) after M characteristic parameters of the fabric image are obtained, comparing each characteristic parameter with the corresponding search interval respectively;
(22) if a certain characteristic parameter falls into one of the search intervals corresponding to the characteristic parameter and the search interval only corresponds to a glossiness grade, the glossiness grade of the fabric is judged to be consistent with the glossiness grade corresponding to the search interval.
Further, after the step (22), if a certain characteristic parameter falls into one of the search intervals corresponding to the characteristic parameter and the search interval corresponds to two or more gloss levels, the gloss levels corresponding to the search intervals in which other characteristic parameters fall are continuously judged, and one gloss level corresponding to different search intervals is compared and used as the gloss level of the fabric.
Furthermore, the automatic analysis and evaluation method for the fabric glossiness adopts a discrete Hopfield neural network for evaluation, firstly, sample images corresponding to N grades are used as balance points of the discrete Hopfield neural network, then the discrete Hopfield neural network is learned by an orthogonalization method, so that evaluation indexes of each grade tend to the balance points of the network, and after learning is completed, the balance points are grade division points corresponding to each fabric glossiness grade;
and inputting the characteristic parameters of the fabric to be evaluated as a new initial state into the discrete Hopfield neural network, wherein the discrete Hopfield neural network performs calculation, and when the network state is not changed any more, the grade corresponding to the new balance point is the glossiness grade of the fabric to be evaluated.
The invention also provides an automatic analysis and evaluation system for fabric glossiness, which comprises an image acquisition device and a processing device, wherein the image acquisition device comprises a CCD camera, an LED light source, a spherical seat and a shading box body, the CCD camera, the LED light source and the spherical seat are arranged in the shading box body, the fabric to be evaluated is covered on the spherical seat, the CCD camera acquires an image of the fabric to be evaluated and sends the acquired image of the fabric to the processing device, and the processing device evaluates the grade of the fabric to be evaluated according to the automatic analysis and evaluation method for fabric glossiness recorded in the foregoing.
Compared with the prior art, the invention has the advantages and positive effects that: the method for automatically analyzing and evaluating the fabric glossiness obtains a stable fabric image in a light stable environment by using a computer vision based on an optical imaging principle, establishes a relation between a characteristic parameter of the image and a glossiness grade, obtains the characteristic parameter of the fabric image according to the relation and obtains the fabric glossiness grade so as to realize objective evaluation of the fabric glossiness according with the visual sense of human eyes.
Other features and advantages of the present invention will become more apparent from the detailed description of the embodiments of the present invention when taken in conjunction with the accompanying drawings.
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 flow chart of an embodiment of the method for automatically analyzing and evaluating the gloss of a fabric according to the present invention;
FIG. 2 is a corresponding image in step 11 of one embodiment of the proposed evaluation method of the present invention;
FIG. 3 is a corresponding image of step 12 of one embodiment of the proposed evaluation method of the present invention;
FIG. 4 is a K value analysis diagram in one embodiment of the evaluation method of the present invention;
FIG. 5 is a graph of gray scale contrast C analysis in an embodiment of the proposed evaluation method of the present invention;
FIG. 6 is a very poor R gloss fill level in one example of the evaluation method proposed by the present inventionΦAnalyzing the graph;
FIG. 7 is a graph illustrating a contrast CONN analysis of a gray level co-occurrence matrix in an embodiment of the evaluation method of the present invention;
FIG. 8 is a graph of the grayscale contrast SC analysis of an S component image in an embodiment of the evaluation method proposed by the present invention;
fig. 9 is a schematic structural diagram of an embodiment of an automatic analysis and evaluation system for fabric gloss according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the 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.
Example one
The embodiment provides an automatic analysis and evaluation method for fabric glossiness, as shown in fig. 1, including the following steps:
a glossiness grade dividing step, namely quantifying the glossiness of the fabric into N grades, wherein N is an integer greater than 1;
establishing a corresponding relation between the glossiness levels and the characteristic parameters, namely selecting M characteristic parameters of the image, wherein each characteristic parameter is provided with one or more search intervals, each search interval corresponds to one glossiness level or a plurality of glossiness levels, and M is an integer greater than 1;
and a fabric glossiness evaluation step, namely acquiring a fabric image of the fabric to be evaluated, processing the fabric image, acquiring M characteristic parameters of the fabric image, and searching the glossiness grade of the fabric to be evaluated from the search interval according to the M characteristic parameters.
In the present embodiment, "fabric" refers to a woven article, and the psychovisual amount of fabric gloss refers to a glossy style language. On the basis of the existing language for describing fabric gloss, the SD method is combined, and the following psychophysical semantic quantifier is selected: the "dull", "lusterless", "diffuse", "shiny", and "shiny" 5 semantic quantifiers were rated at a grade of 5, and then the physical content grasping criteria in constructing the subjective scoring table was introduced to give an explanation of the physical properties of each semantic quantifier: dull-fabric surface dull and dull, no reflected diffuse light; lack of gloss-the fabric surface has no reflected light, less diffuse light, and less gloss; diffused, namely weak reflected light exists on the surface of the fabric, and diffused light which is strong and uniformly distributed is obtained, and the luster is not dull or soft; bright-the fabric surface reflected light is stronger, but the distribution is concentrated, the gloss is softer; the fabric surface is shiny, has extremely strong reflected light, bright visual sensation and even and very soft reflected light. Finally, the description of the fabric luster feeling is similar to the luster feeling of other objects, so that the semantic quantifier is more specific and easier to understand, and the recognition feeling of people is enhanced, namely: dull fabric gloss gives a feeling as if it were mud with little gloss; the lack of gloss is better than milky gloss, and the overall gloss is very poor; the diffused reflected light is poor, and the diffused light is uniformly distributed, similar to semi-metallic luster; the light has strong reflection intensity, and has twinkling light sensation, and pearl-like luster; the glittering reflected light is strong, the visual sensation is bright, and the glittering luster is like diamond. The representation of the fabric gloss visual physical quantity is a data form which represents a quantitative value of the fabric gloss intensity and gives a corresponding grade evaluation, and the description of the visual psychological quantity is a language expression form which describes the visual style of the fabric gloss sense, and the two are classified in a special way and both refer to the same 'target'. Since the machine cannot recognize the human language meaning, the machine can realize automatic recognition and evaluation by grading the glossiness, quantizing the glossiness of the fabric into a plurality of grades and giving corresponding judgment conditions of the machine. Of course, the machine outputs the gloss level through judgment and analysis, and the user can determine the semantic description corresponding to the level according to the corresponding level, for example, in order to better distinguish the difference between the integer levels and make the rating result more accurate, in this embodiment, according to the convention of appearance rating, the gloss level is set as 9 levels, and 0.5 level is added between every two levels on the basis of 5 levels in 1.2. The specific semantic content of each level is shown in table 1.
TABLE 1
The method for automatically analyzing and evaluating the fabric glossiness of the embodiment acquires a stable fabric image in a light stable environment by using a computer vision based on an optical imaging principle, establishes a relation between a characteristic parameter of the image and a glossiness grade, acquires the characteristic parameter of the fabric image and obtains the fabric glossiness grade according to the relation, and accordingly realizes objective evaluation of the fabric glossiness according with the visual sense of human eyes.
In the step of evaluating the glossiness of the fabric, the step of processing the fabric image comprises the following steps:
s11, converting the color fabric image into a gray image, wherein the RGB image contains a large amount of color information consisting of three components of red, green and blue, and the image processing requires analyzing the gray image, so the RGB image needs to be converted into a weighted average gray image and a R, G, B gray image, as shown in fig. 2;
s12, converting the gray level image into a binary image, observing the fabric gloss image to find that the forms of the gloss-formed reflective areas are different, the bright pixels of the strip-shaped reflective areas are obviously concentrated in one direction and are directional, and the bright pixels of the circular reflective areas and the scattered point-shaped reflective areas are similar and have no directionality. To find the number of bright pixels of the fabric gloss image, a binarization process for the gray scale image is required, because the result of the binarization of the fabric hemispherical gray scale image will make the reflective areas white pixels and the non-reflective areas black pixels, as shown in fig. 3.
In step S11, after the color fabric image is converted into a weighted average value grayscale image and R, G, B three-component grayscale image, the grayscale contrast is enhanced by enhancing the darkness of the four parallel grayscale images, and the grayscale image with the maximum brightness and darkness contrast is selected as the operation object in step S12, where the grayscale image with the maximum brightness and darkness contrast refers to the image with the maximum grayscale cv (coefficient of variation).
The technical scheme is characterized in that the information of the digital representation of the fabric hemispherical image, which can replace human eyes to perceive the gloss visual style, is extracted and is expressed by a computer. Therefore, through experimental research on a certain number of fabric samples, the sizes, the brightness intensities, the brightness concentration or the brightness dispersion distributions of the light reflection areas of the gloss images of the fabrics with different gloss are different, and a certain rule exists, so that the fabric can be used as an analysis object of the gloss description characteristics of the fabric hemispherical image. Five feature parameters including nominal filling degree, gray scale contrast, gloss filling degree range, gray scale co-occurrence matrix contrast, saturation contrast, or any combination of two or more of the above five feature parameters are extracted from the four aspects of gray scale feature, direction feature, texture feature and color feature.
In order to examine the nature and the quantity of white pixels in a binary image, a nominal filling degree K is defined, and the calculation method of the nominal filling degree K is as follows: area S occupied by white pixels in binary imageWWith the total area S of the imaged area of the fabric in the fabric imageZThe ratio of the areas of (a):
by analyzing the K value, as shown in fig. 4, it can be found that: the glossy fabric, namely the fabric with the gloss grade of 2 or above, has a nominal filling value K of less than 75 percent, but has no obvious clustering effect on the same grade, so that the K has no separability on the fabric with the grade of 2-5; the nominal fill value K of the matt fabric is not less than 75%, wherein the fabric K with a gloss rating of 1 is not less than 90%, and the fabric K with a gloss rating of 1.5 ranges from 75% to 90%. A correlation analysis of the nominal fill K and gloss D ratings was made for fabric samples having gloss ratings of 1 and 1.5, with a 1-degree fit between the two. The characteristic parameters can be used as the basis for objectively evaluating the grade 1 and the grade 1.5.
The image reflection area contains perceptual brightness information, namely, the brightness of colorimetry or gray scale characteristics in digital image processing. The existence, strength and form distribution state of the reflecting area in the image can be distinguished by taking the gray scale and the contrast as the description characteristics of the image reflecting property. The calculation method of the gray contrast C comprises the following steps:
carrying out gray difference calculation on the gray image to obtain a gray contrast C;
and extracting and analyzing the distribution of the gray contrast value C of the fabric sample image. As can be seen from FIG. 5, the C of the fabric sample class 2 is ≦ 10, the C of the fabric samples class 2.5, class 3 and class 3.5 is between 10 and 20, the C of the fabric samples class 4 and class 4.5 is between 20 and 40, and the C of the fabric sample class 5 is > 40. The correlation analysis of the gray contrast C and the fabric gloss level D is carried out, the correlation coefficient of the gray contrast C and the fabric gloss level D is 89.59%, because the gray contrast is the calculation of gray difference, the sensitivity of human eyes to gray mutation boundary effect is far lower than the sensitivity of data, and therefore the correlation coefficient is reduced, but the result is not poor, because the characterization effect of the data can make up the deficiency of the visual perception of human eyes, and therefore C can be used as one of the indexes for characterizing the 2-level and 5-level fabric gloss.
In order to distinguish whether the fabric gloss image has directionality or not, the gloss filling degree phi is defined, and the calculation method of the gloss filling degree phi is as follows:
area S occupied by white pixels in binary imageWWith the total area S of the rectangular window in the fabric imageRThe ratio of the areas of (a):
for glossy fabrics, the larger the Φ, the more glossy pixels of the image, i.e., the larger the retroreflective area, the more glossy the image. Experimental research shows that the gloss filling degree has extremely poor RΦThe expression rule (difference between the maximum and minimum gloss fill levels) is: the directional gloss image has a large difference in gloss filling level, and the nondirectional gloss image has a small difference in gloss filling level. When R is shown in FIG. 6ΦAt ≦ 20%, diffuse light samples and circular reflectance samples are classified as one, while in combination with gray scale contrast C pair: if C is less than or equal to 20, the sample is judged to be diffused light, and the gloss grade of the diffused light can be 2.5 grade, 3 grade and 3.5 grade; if 20, a<C is less than or equal to 40, the sample is judged to be circular and reflective, and the gloss grade D is 4. The ranking result is completely consistent with the known ranking. When R isΦ>At 20%, the fabric gloss morphology resolves to bar reflectance, when C satisfies 20<When C is less than or equal to 40, the gloss grade D is 4.5 grade.
In summary, RΦA broad classification of gray contrast C versus 2.5-4.5 sample gloss can be verified and fabric samples of 4 and 4.5 levels can be distinguished.
The method for calculating the contrast CONN of the gray level co-occurrence matrix comprises the following steps:
carrying out image texture analysis on the gray level image to obtain a gray level co-occurrence matrix contrast CONN;
the image experiment researches show that the gloss forms of scattered point-like bright areas exist in certain fabric hemispherical images, the gloss characteristics of the fabric hemispherical images can be obtained by adopting image texture analysis, the contrast CONN of a gray level co-occurrence matrix most commonly used for describing textures is selected for analysis, the larger the value of the contrast CONN is, the more obvious the image texture characteristics are, the image texture characteristics belong to strong textures, and the images are represented as scattered bright points. From the analysis of the CONN values in fig. 7, it can be seen that: when CONN <0.5, then D ═ 2.5 or D ═ 3.5; when CONN is more than or equal to 0.5, D is 3.
Saturation contrast SCThe calculation method comprises the following steps:
obtaining an S component image in a fabric image of an HSI color space, and obtaining a gray contrast of the S component image as a saturation contrast SC。
The visual perception of fabric gloss effect on the color attributes of fabric surfaces also requires the study of their chroma characteristics. Chroma, i.e. the purity of a color, is referred to in colorology as saturation, and gives a measure of the degree to which a pure color is diluted by white light. The higher the chroma, the purer and more brilliant the color; the lower the chroma, the more turbid and the darker the color. S component in HSI color model is saturation, selecting gray contrast SC of S component image as color characteristic parameter, SCThe larger the difference in saturation, the better the gloss of the fabric. The remaining fabric samples of grade 2.5 and 3.5 were analyzed for saturation contrast values, as shown in fig. 8: when S isC<When 10, D is 2.5; sCWhen the value is more than or equal to 10, D is 3.5.
In the step of evaluating the fabric glossiness, the method for finding out the glossiness grade of the fabric according to the M characteristic parameters of the fabric comprises the following steps:
s21, after M characteristic parameters of the fabric image are obtained, comparing each characteristic parameter with the corresponding search interval respectively;
s22, if a certain characteristic parameter falls into one of the search intervals corresponding to the characteristic parameter and the search interval only corresponds to a glossiness grade, the glossiness grade of the fabric is judged to be consistent with the glossiness grade corresponding to the search interval.
After step S22, if a certain characteristic parameter falls into one of the search intervals corresponding to the characteristic parameter and the search interval corresponds to two or more gloss levels, the gloss levels corresponding to the search intervals in which other characteristic parameters fall are continuously determined, and a gloss level corresponding to different search intervals is compared as the gloss level of the fabric.
The automatic analysis and evaluation method of the fabric glossiness adopts a discrete Hopfield neural network for evaluation, firstly, sample images corresponding to N grades are used as balance points of the discrete Hopfield neural network, then the discrete Hopfield neural network is learned by an orthogonalization method, so that evaluation indexes of each grade tend to the balance points of the network, and after learning is completed, the balance points are grade division points corresponding to each fabric glossiness grade;
and inputting the characteristic parameters of the fabric to be evaluated as a new initial state into the discrete Hopfield neural network, wherein the discrete Hopfield neural network performs calculation, and when the network state is not changed any more, the grade corresponding to the new balance point is the glossiness grade of the fabric to be evaluated.
And dividing the fabric gloss grade according to the five indexes of the nominal filling degree, the gray scale contrast, the gloss filling degree range difference, the gray scale co-occurrence matrix contrast and the saturation contrast. Firstly, five characteristic parameter values of 70 fabric sample gloss images are extracted, the average value of each evaluation index corresponding to each grade sample is divided into four grades to be used as an ideal evaluation index of each grade, namely as a balance point of a Hopfield neural network. Corresponding to the encoding mode in the Hopfield neural network, that is, each index is represented by 4 neurons, since the states of discrete Hopfield neural network neurons are only '1' and '-1', when the evaluation index is mapped to the state of the neuron, the evaluation index needs to be encoded. The encoding rule is as follows: when the index value is larger than or equal to a certain level, the corresponding neuron state is set to be 1, otherwise, the neuron state is set to be 1. The ideal 9-level evaluation index codes, wherein ● indicates that the neuron state is "1", i.e., greater than or equal to the ideal evaluation index value of the corresponding level, and vice versa, is indicated by o. And taking the 70 fabric sample data as learning samples, and creating a discrete Hopfield neural network by utilizing a neural network toolbox function carried by MATLAB, wherein the network learning process of the 70 samples is a process that the evaluation index gradually approaches to a balance point of the Hopfield neural network.
Selecting another 50 fabric gloss grade evaluation index codes to be classified as input of the Hopfield neural network, performing experiments and learning for 20 times to obtain a stable output value of the system, namely obtaining a grade evaluation simulation result, and comparing the simulation result with the known real grade, wherein the result is listed in Table 2.
| Sample numbering | Simulation result | Real result | Sample numbering | Simulation result | Real result |
| 71 | 5 | 5 | 96 | 1.5 | 1.5 |
| 72 | 4.5 | 4.5 | 97 | 4.5 | 4.5 |
| 73 | 4 | 4 | 98 | 5 | 5 |
| 74 | 3.5 | 3.5 | 99 | 3 | 3 |
| 75 | 3 | 3 | 100 | 1 | 1 |
| 76 | 2.5 | 2.5 | 101 | 1 | 1 |
| 77 | 2 | 2 | 102 | 3.5 | 3.5 |
| 78 | 1.5 | 1.5 | 103 | 4 | 4 |
| 79 | 1 | 1 | 104 | 4 | 4 |
| 80 | 4 | 4 | 105 | 2 | 2 |
| 81 | 1 | 1 | 106 | 1.5 | 1.5 |
| 82 | 3.5 | 3.5 | 107 | 4.5 | 4.5 |
| 83 | 5 | 4.5 | 108 | 3 | 2.5 |
| 84 | 2.5 | 2.5 | 109 | 5 | 5 |
| 85 | 2.5 | 2.5 | 110 | 5 | 5 |
| 86 | 3 | 3 | 111 | 2 | 2 |
| 87 | 1 | 1 | 112 | 4 | 4 |
| 88 | 2 | 2 | 113 | 2.5 | 2.5 |
| 89 | 4 | 4 | 114 | 3.5 | 3 |
| 90 | 4 | 4 | 115 | 1.5 | 1.5 |
| 91 | 1.5 | 1.5 | 116 | 4.5 | 4.5 |
| 92 | 1 | 1 | 117 | 1 | 1 |
| 93 | 3.5 | 3.5 | 118 | 4.5 | 4.5 |
| 94 | 2 | 2 | 119 | 4 | 4 |
| 95 | 5 | 5 | 120 | 3.5 | 3.5 |
TABLE 2
As can be seen from the observation of Table 2, the simulation results of only three samples, 83, 108 and 114, of the 50 fabric samples are different from the real results, and the accuracy is 94%, which indicates that the simulation results of the computer are better consistent with the real results. Moreover, the simulation results of the three samples only have a difference of 0.5 grade from the real results, because the difference between the adjacent grades is too small, after subjective re-evaluation, the grades of the three samples are considered to be between the two grades, and the subjective evaluation of any one grade can be accepted when the grade is divided, which shows that the resolution and the accuracy of the established fabric gloss grade classification model are superior to the results of human eye subjective judgment, thereby proving that the classification capability and the prediction capability of the model are reliable, and the method is particularly suitable for the single-color fabrics.
In a second embodiment, the present embodiment provides an automatic analysis and evaluation system for fabric glossiness, as shown in fig. 9, the system includes an image acquisition device and a processing device, the image acquisition device includes a CCD camera 11, an LED light source 12, a spherical seat 13 and a shading box 14, the CCD camera 11, the LED light source 12 and the spherical seat 13 are disposed in the shading box 14, the fabric to be evaluated is covered on the spherical seat 13, the CCD camera 11 acquires an image of the fabric to be evaluated, and sends the acquired image of the fabric to the processing device 15, and the processing device evaluates the grade of the fabric to be evaluated according to the automatic analysis and evaluation method for fabric glossiness described in the first embodiment, which may specifically refer to the first embodiment, and will not be described herein again.
It is to be understood that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make modifications, alterations, additions or substitutions within the spirit and scope of the present invention.
Claims (6)
1. An automatic analysis and evaluation method for fabric glossiness is characterized by comprising the following steps:
a glossiness grade dividing step, namely quantifying the glossiness of the fabric into N grades, wherein N is an integer greater than 1;
establishing a corresponding relation between the glossiness levels and the characteristic parameters, namely selecting M characteristic parameters of the image, wherein each characteristic parameter is provided with one or more search intervals, each search interval corresponds to one glossiness level or a plurality of glossiness levels, and M is an integer greater than 1;
the method comprises the following steps of evaluating the glossiness of the fabric, namely acquiring a fabric image of the fabric to be evaluated, processing the fabric image, acquiring M characteristic parameters of the fabric image, and searching the glossiness grade of the fabric to be evaluated from the search interval according to the M characteristic parameters;
in the step of evaluating the glossiness of the fabric, the step of processing the fabric image comprises the following steps:
(11) converting the colorful fabric image into a gray image;
(12) converting the gray level image into a binary image;
the M characteristic parameters comprise any two or more combinations of nominal filling degree, gray scale contrast, gloss filling degree range, gray scale co-occurrence matrix contrast and saturation contrast;
the nominal filling degree K is calculated by the following method: area S occupied by white pixels in binary imageWWith the total area S of the imaged area of the fabric in the fabric imageZThe ratio of the areas of (a):
the calculation method of the gray contrast C comprises the following steps:
carrying out gray difference calculation on the gray image to obtain a gray contrast C;
the method for calculating the extremely poor gloss filling degree phi comprises the following steps:
area S occupied by white pixels in binary imageWWith the total area S of the rectangular window in the fabric imageRThe ratio of the areas of (a):
the method for calculating the contrast CONN of the gray level co-occurrence matrix comprises the following steps:
carrying out image texture analysis on the gray level image to obtain a gray level co-occurrence matrix contrast CONN;
saturation contrast SCThe calculation method comprises the following steps:
obtaining an S component image in a fabric image of an HSI color space, and obtaining a gray contrast of the S component image as a saturation contrast SC。
2. The method for automatically analyzing and evaluating the gloss of a fabric according to claim 1, wherein in the step (11), after the color fabric image is converted into the weighted average value gray image and the R, G, B three-component gray image, the gray contrast is enhanced by enhancing the darkness of the four parallel gray images, and finally the gray image with the highest brightness and darkness contrast is selected as the operation object in the step (12).
3. The method for automatically analyzing and evaluating the glossiness level of the fabric according to claim 1 or 2, wherein in the fabric glossiness evaluation step, the method for finding out the glossiness level of the fabric according to the M characteristic parameters of the fabric comprises the following steps:
(21) after M characteristic parameters of the fabric image are obtained, comparing each characteristic parameter with the corresponding search interval respectively;
(22) if a certain characteristic parameter falls into one of the search intervals corresponding to the characteristic parameter and the search interval only corresponds to a glossiness grade, the glossiness grade of the fabric is judged to be consistent with the glossiness grade corresponding to the search interval.
4. The method for automatically analyzing and evaluating the glossiness of the fabric according to claim 3, wherein after the step (22), if a certain characteristic parameter falls into one of the search sections corresponding to the characteristic parameter and the search section corresponds to two or more glossiness levels, the glossiness levels corresponding to the search sections in which other characteristic parameters fall are continuously determined, and one glossiness level corresponding to different search sections is compared and used as the glossiness level of the fabric.
5. The automatic analyzing and evaluating method for fabric glossiness according to any one of claims 1 or 2, wherein the automatic analyzing and evaluating method for fabric glossiness adopts a discrete Hopfield neural network for evaluation, firstly, sample images corresponding to N grades are used as balance points of the discrete Hopfield neural network, then the discrete Hopfield neural network is learned by an orthogonalization method, so that evaluation indexes of each grade tend to the balance points of the network, and after learning is completed, the balance points are grade division points corresponding to each fabric glossiness grade;
and inputting the characteristic parameters of the fabric to be evaluated as a new initial state into the discrete Hopfield neural network, wherein the discrete Hopfield neural network performs calculation, and when the network state is not changed any more, the grade corresponding to the new balance point is the glossiness grade of the fabric to be evaluated.
6. An automatic analysis and evaluation system for fabric glossiness is characterized by comprising an image acquisition device and a processing device, wherein the image acquisition device comprises a CCD camera, an LED light source, a spherical seat and a shading box body, the CCD camera, the LED light source and the spherical seat are arranged in the shading box body, a fabric to be evaluated is fixed on the spherical seat, the CCD camera acquires an image of the fabric to be evaluated, the acquired image of the fabric is sent to the processing device, and the processing device evaluates the grade of the fabric to be evaluated according to the automatic analysis and evaluation method for fabric glossiness of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810777296.6A CN109325934B (en) | 2018-07-16 | 2018-07-16 | Method and system for automatically analyzing and evaluating fabric glossiness |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810777296.6A CN109325934B (en) | 2018-07-16 | 2018-07-16 | Method and system for automatically analyzing and evaluating fabric glossiness |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109325934A CN109325934A (en) | 2019-02-12 |
| CN109325934B true CN109325934B (en) | 2021-05-14 |
Family
ID=65263785
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810777296.6A Active CN109325934B (en) | 2018-07-16 | 2018-07-16 | Method and system for automatically analyzing and evaluating fabric glossiness |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109325934B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111209936B (en) * | 2019-12-26 | 2023-04-25 | 新绎健康科技有限公司 | Method and system for determining face gloss based on k-means clustering |
| CN114386295B (en) * | 2022-03-23 | 2022-07-05 | 武汉纺织大学 | Textile computer simulation method based on color separation and color change of colored spun yarns |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103149210A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | System and method for detecting fabric cashmere content based on scale graphic features |
| CN105787562A (en) * | 2016-03-24 | 2016-07-20 | 上海工程技术大学 | Yarn dyed fabric texture type recognition method |
| CN107153969A (en) * | 2017-04-20 | 2017-09-12 | 温州市鹿城区中津先进科技研究院 | The big data processing method that fabric scheduling is instructed is carried out based on positioning label |
-
2018
- 2018-07-16 CN CN201810777296.6A patent/CN109325934B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103149210A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | System and method for detecting fabric cashmere content based on scale graphic features |
| CN105787562A (en) * | 2016-03-24 | 2016-07-20 | 上海工程技术大学 | Yarn dyed fabric texture type recognition method |
| CN107153969A (en) * | 2017-04-20 | 2017-09-12 | 温州市鹿城区中津先进科技研究院 | The big data processing method that fabric scheduling is instructed is carried out based on positioning label |
Non-Patent Citations (2)
| Title |
|---|
| Study on Automatic Recognition of Fabric Color and Matching to Standard Color Chip by Computer Vision and Image Analysis Technology;Liqing Xie等;《Journal of Fiber Bioengineering and Informatics》;20161231;第29-37页 * |
| 织物光泽测试技术研究现状与展望;申悦等;《纺织科技进展》;20170331;第2017年卷(第3期);第5-9页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109325934A (en) | 2019-02-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chadwick et al. | The perception of gloss: A review | |
| CN106529429B (en) | A kind of skin of face analysis system based on image recognition | |
| Obein et al. | Difference scaling of gloss: Nonlinearity, binocularity, and constancy | |
| CN109325934B (en) | Method and system for automatically analyzing and evaluating fabric glossiness | |
| CN105678734B (en) | A kind of heterologous test image scaling method of image matching system | |
| CN112396573A (en) | Facial skin analysis method and system based on image recognition | |
| Guo et al. | The foundation of a color-chip evaluation system of jadeite-jade green with color difference control of medical device | |
| CN110378885A (en) | A kind of focal area WSI automatic marking method and system based on machine learning | |
| Hadjianfar et al. | A new method for measuring luster index based on image processing | |
| CN101539517A (en) | Method for evaluating high-gloss coating surfaces by fuzzy logic model | |
| CN108460380A (en) | A kind of bamboo cane method for sorting colors and system based on domain color | |
| Toyota et al. | Optical measurement of interference color of pearls and its relation to subjective quality | |
| De Vita et al. | Statistical techniques applied to the automatic diagnosis of dermoscopic images | |
| Saifullah et al. | K-means segmentation based-on lab color space for embryo detection in incubated egg | |
| Chen et al. | Identification of tea varieties using computer vision | |
| CN115545383A (en) | Binocular stereoscopic vision detection method for crisp appearance quality of strawberries | |
| Furferi et al. | Towards automated and objective assessment of fabric pilling | |
| Hamdi et al. | A multidimensional scaling analysis of surface perceptual parameters on scratch and mar visibility resistance in polymers | |
| Winston Wang et al. | Looking into special surface effects: diffuse coarseness and glint impression | |
| Hadjianfar et al. | Fuzzy classification of luster of textile fabrics | |
| CN116077030A (en) | Skin evaluation method based on skin component volume content | |
| Haeri et al. | Novel digital image analysis using fractal dimension for assessment of skin radiance | |
| Filip et al. | Perceptual attributes analysis of real-world materials | |
| Griffin | Similarity of psychological and physical colour space shown by symmetry analysis | |
| Wang et al. | Determining vitreousness of durum wheat using transmitted and reflected images |
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 |