GB/T 38879-2020. Particle- Size analysis- Color image analysis methods.
1 Scope
GB/T 38879 specifies the methods for the collection and processing of particle color images and particle size analysis.
GB/T 38879 is applicable to the determination of particle size by color image acquisition devices such as image-based particle size and shape analyzers, microscopes, scanners or cameras.
2 Normative references
The following documents are indispensable for the application of this document. For all dated references, only the dated version applies to this document. For all undated references, the latest version (including all amendments) applies to this document.
GB/T 15445.1 Presentation of particle size analysis results Part 1: Graphical representation
GB/T 21649.1-2008 Particle size analysis - Image analysis method Part 1: Static image analysis method.
3 Terms and definitions
. The following terms and definitions apply to this document.
3.1
Color space
Three-dimensional space representing colors.
Note 1: Each color is represented as a point in a three-dimensional coordinate system. Color space is also called color model. Commonly used color spaces include RGB, CIE LAB, HSV and HSI. ||
tt||Note 2: Rewrite GB/T 5698-2001, definition 4.57.
3.2
Hue
Represents the color characteristics of red, yellow, green, blue and purple.
One of the three attributes of color.
Note: An important attribute of color, which determines the essence of color, is the feeling people have about a certain color presented on the surface of an object.
[GB/T 5698-2001, definition 5.7]
3.3
Lightness
The degree of visual perception characteristics of the surface of an object under the same lighting conditions, with a white board as the reference.
One of the three attributes of color. ||
tt||Note: Rewrite GB/T 5698-2001, definition 5.8.
3.4
?? Saturation
The visual attribute used to evaluate the components of pure color in the whole visual field.
[GB/T 5698-2001, definition 5.10]
3.5
Contrast
The estimation of the difference in appearance when two parts appear simultaneously or successively in the field of vision. Contrast includes brightness contrast, lightness contrast and color contrast.
Note: In particle size analysis, it is the brightness or color difference between the foreground pixels that make up the particle object and the background pixels in the adjacent area of ??the particle, or between the pixels of different particle objects.
[GB/T 5698-2001, definition 5.33]
3.6
Color histogram
A graph that describes the proportion of different color components in an image.
This standard specifies the methods for collecting and processing particle color images and analyzing particle size. This standard is applicable to the determination of particle size by color image acquisition devices such as image-based particle size and shape analyzers, microscopes, scanners or cameras.

ICS19.120
National Standard of the People's Republic of China
GB/T38879—2020
Particle
Particle size analysis
Color image analysis
Particle—Size analysis-—Color image analysis methods2020-06-02Published
State Administration for Market Regulation
National Administration of Standardization
Published
2020-08-01Implementation
Foreword
Introduction
Scope
Normative references
Terms and definitions
Sample preparation
Image acquisition
General
Acquisition
Image processing and analysis
General
Image preprocessing
Particle image segmentation
Particle size measurement and calculation
Classification of particles
Calibration and traceability
Test report
Appendix A (Informative Appendix)
Appendix B (Informative Appendix)
Appendix C (Informative Appendix)
Appendix D (Informative Appendix)
References
Typical steps for particle size analysis based on color imagesFlowchart Commonly used color spaces for image processing and the conversion relationship between themGB/T38879—2020
Basic principles and steps of color image segmentation algorithm based on fuzzy C-means clustering…Particle color image segmentation process and result example 12 in CIELAB color space
Foreword
This standard was drafted in accordance with the rules given in GB/T1.1-2009 GB/T38879—2020
This standard was proposed and coordinated by the National Technical Committee for Particle Characterization and Sorting and Screen Standardization (SAC/TC168). The drafting units of this standard are: Shanghai University of Technology, PetroChina Daqing Oilfield Co., Ltd. Exploration and Development Research Institute, Dandong Baite Instrument Co., Ltd., Beijing Coast Hongmeng Standard Material Technology Co., Ltd., Shenzhen Defang Nanotechnology Co., Ltd., Institute of Process Engineering, Chinese Academy of Sciences, China Machinery Productivity Promotion Center, Beijing Physical and Chemical Analysis and Testing Center, and Yisiqi (Beijing) Technology Development Co., Ltd. The main drafters of this standard are: Zhou Lu, Deng Wenjing, Xu Xiqing, Cai Xiaoshu, Dong Qingyun, Li Zhaojun, Su Mingxu, Li Li, Wang Yuanhang, Gao Yuan, Yu Fang, and Yang Zhenghong.
GB/T38879—2020
Image method has become one of the main methods for particle analysis. Static and dynamic image methods are generally based on the grayscale image of the particle for threshold segmentation, and the particle size and other information are extracted from the binary particle image. Compared with the grayscale image of the particle, the color image contains richer information such as color and texture, which is helpful for the segmentation and identification of the particle. Therefore, this standard conducts particle size analysis on the color image of the particle, aiming to provide guidance on the acquisition, processing and application scope of the color image of the particle. N
1 Scope
Particles
Particle size analysis
Color image analysis method
This standard specifies the methods for the collection and processing of particle color images and particle size analysis GB/T38879-2020
This standard applies to the determination of particle size by color image acquisition equipment such as dry image method particle size and shape analyzer, microscope, scanner or camera.
Normative referenced documents
The following documents are indispensable for the application of this document. For all dated referenced documents, only the dated version applies to this document. For all undated referenced documents, the latest version (including all amendments) applies to this document. GB/T15445.1 Presentation of particle size analysis results Part 1: Graphic representation Part 1: Static image analysis method
Image analysis method
GB/T21649.1-2008 Particle size analysis
Terms and definitions
The following terms and definitions apply to this document. 3.1
Color space
colorspace
Three-dimensional space representing color.
Note 1: That is, each color is represented as a point in a three-dimensional coordinate system. Color space is also called color model. Commonly used color spaces include RGB, CIELAB, HSV and HSI.
Note 2: Rewrite GB/T5698-2001, definition 4.57. 3.2
Hue
Represents color characteristics such as red, yellow, green, blue and purple. One of the three attributes of color.
Note: An important attribute of color, which determines the essence of color, is people's feeling of a certain color presented on the surface of an object. [GB/T5698—2001, definition 5.7]
Brightness
lightness
The degree of visual perception characteristics of the surface of an object under the same lighting conditions, with a white board as the reference. One of the three attributes of color.
Note: Rewrite GB/T5698—2001, definition 5.8. 3.4
Saturation
saturation
The visual attribute used to evaluate the components of pure color in the whole vision. [GB/T5698—2001, definition 5.10]
GB/T38879—2020
Contrast
contrast
The estimation of the difference in appearance when two parts appear simultaneously or successively in the field of vision. Contrast includes brightness contrast, lightness contrast and color contrast.
Note: In particle size analysis, it is the brightness or color difference between the foreground pixels of the particle object and the background pixels in the adjacent area of ??the particle, or between the pixels of different particle objects.
［GB/T5698
3—2001, definition 5.33
colorhistogram
Color histogram
A graph describing the proportion of different color components in an image. 3.7
Color moment
colormoment
The color feature representation of the basic moment of each component graph of a color image Note: Color moment can be used in any color space, including first-order moment, second-order moment, third-order moment, etc. 3.8
Spatial relation
spatialrelation
The mutual spatial position or relative direction relationship between pixels in an image or between multiple objects segmented out. 3.9
Texture
texture
Feature information obtained by statistical calculation based on a region containing multiple pixels. 3.10
Image segmentation
image segmentation
Divide the regions with a certain specific meaning in the image into non-intersecting regions, and each region meets the similarity criteria such as grayscale, texture, and color.
3.11
Threshold
The grayscale level set to distinguish the object from the background [GB/T21649.1—2008, definition 3.1.13] 3.12
Clustering
cluster
The process of dividing a collection of physical or abstract objects into multiple classes consisting of similar objects. Note: In the particle size analysis of color images, the basic color features are generally clustered. 3.13
edge detection
The process of determining the contour based on the grayscale step change of the edge pixels of the target object. 3.14
test frame
measurement frame
An area within the field of view, in which the particles are statistically analyzed and the image is analyzed. Note: A series of test frames constitute the total test area [GB/T21649.1—2008, definition 3.1.2] 3.15
similarity
similarity
Measures the similarity of the pixels in the image in terms of certain feature information. 2
3.16
3.17
3.18
3.19
3.20
Distortion
The distortion of the image of an object relative to the object itself. Noise
noise
Unnecessary or redundant interference information in the image data. Robustness
robustness
The ability of a system to maintain its performance under the disturbance of uncertainty. Feretdiameter
Feretdiameter
The distance between parallel lines tangent to the two sides of the particle image contour. [GB/T21649.1—2008, definition 3.1.6] Legendre ellipse of inertia
GB/T38879—2020
An ellipse with its center at the particle's centroid and the same first-order and second-order geometric moments of inertia as the original particle area. Note 1: An ellipse can be characterized by its major and minor axes, and the position and direction of its center of gravity. Note 2: Rewrite GB/T15445.6—2014, definition 8.1.2. 3.21
Ellipticity
ellipseratio
The ratio of the minor axis length to the major axis length of the Legendre ellipse of inertia. Note: Rewrite GB/T15445.6—2014, definition 8.1.3. 3.22bzxz.net
Aspect ratio
aspectratio
The ratio of the shortest Feret diameter to the longest Feret diameter. [GB/T15445.6—2014, definition 8.1.3 Sample preparation
Different colored particle samples have different preparation methods or industry requirements. Sample preparation should be carried out in accordance with relevant standards to ensure that the test or analysis samples selected from the original samples are representative. Sample preparation should fully disperse the particles, try to avoid overlapping contours, and select appropriate light source strength to make the boundaries of the particles clear enough.
Image acquisition
5.1 General
When using color image acquisition equipment to obtain color images of particles and measure particle size, the color or grayscale of the particles and the background should be greatly different under the premise of following the relevant operating specifications of the instrument and equipment. 5.2
2 Acquisition
The requirements for image acquisition are as follows:
a) According to the particle size range of the particles to be measured and the required measurement accuracy, select the appropriate optical magnification 3
GB/T38879—2020
b) After the acquisition equipment is installed, convert the measurement scale and pixels under different optical magnifications, obtain the correlation coefficient and save it. In the formal measurement, complete the corresponding settings according to the actual measurement magnification c) When acquiring images, adjust the aperture to avoid overexposure or underexposure, select the appropriate filter and polarization angle, obtain satisfactory contrast and saturation, and then obtain a clear image. d) Select identifiable particles in the image and extract all particles in the test frame as much as possible. e) Select a considerable number of non-repetitive test frames in each image. During the test, the operating conditions cannot be changed. 6 Image processing and analysis
6.1 General
The prerequisite for particle size analysis based on color image analysis is to fully understand and utilize the information of the image. The reliability of each step of the image processing process should be considered. Particle color image processing and analysis mainly includes particle color image preprocessing, image segmentation, and particle image particle size measurement. See Appendix A for a typical step flow chart of particle size analysis based on color images. 6.2 Image preprocessing
6.2.1 Color image preprocessing includes geometric correction, filtering and denoising, contrast enhancement and other processes. As long as the measurement results meet the requirements, specific image preprocessing operations can be selected as needed. 6.2.2 When the original image has local distortion, image geometric correction is required to make the geometric position, shape and size of the particles in the original image consistent with their corresponding sample reference. If the instrument has been geometrically calibrated before leaving the factory, it is not necessary to perform geometric calibration again if it is consistent with the factory settings during use. If the manufacturer provides the corresponding calibration method, use the calibration method provided by the instrument manufacturer. In other cases, it is recommended to use the "Zhang Zhengyou Calibration Method" for geometric calibration. Note: "Zhang Zhengyou Calibration Method" For related literature, please refer to reference [5]. 6.2.3 During the atlas acquisition process, different countermeasures should be taken to suppress the intensity of the noise according to the source and nature of the noise, and at the same time, a suitable filtering algorithm should be selected to extract more realistic contour edge information to avoid distortion caused by filtering as much as possible. 6.2.4 In order to separate the particle object from its background area, sufficient contrast is crucial for particle target recognition and size measurement. Image enhancement algorithms are allowed to enhance image contrast before segmentation. 6.3 Particle Image Segmentation
6.3.1 General
In image segmentation, the pixels of the original particle image are divided into background pixels and target pixel sets belonging to the color particles according to the corresponding features. The segmentation of the particle color image should take into account but It is not limited to color features. Other features such as spatial relationship features, texture features and shape features can also be used as the basis for segmentation. Filling holes in particle color images and removing small areas can be regarded as additional processes of image segmentation. Color image segmentation and analysis methods often rely on one or more parameters, and the robustness of the particle color image analysis method needs to be verified in advance. If these parameters are set by the user, multiple verification tests should be carried out based on the particle color image to be analyzed to check the accuracy and stability of the parameter settings to ensure that slight changes in the parameters will not affect the final particle size analysis results. 6.3.2 Image feature extraction
Color features are the most widely used visual features in color image processing, and are also the primary image features considered for particle color images. Color features The features are less dependent on the size, direction and viewing angle of the image itself and have high robustness. When extracting the color features of an image, an appropriate color space should be selected to facilitate the segmentation of granular color images. Please refer to Appendix B for the commonly used color spaces in image processing and the conversion relationships between them.
If necessary, on the basis of color features, the spatial relationship features of pixels or the texture features of multi-pixel areas can be combined, and 4
corresponding quantization methods can be used to express various feature information in the form of vectors as the basis for segmentation. 6.3.3 Color image segmentation methods
6.3.3.1 Color image segmentation methods are mainly divided into the following categories: ———Threshold-based segmentation methods;
——Clustering-based segmentation methods;||tt ||一 Segmentation method based on edge detection.
GB/T38879—2020
The appropriate segmentation theory model and processing algorithm should be selected according to the actual particle image characteristics, or the advantages of multiple segmentation ideas should be combined to achieve accurate segmentation and recognition of color particles.
Note: For relevant documents on typical color image segmentation ideas, see references [6] and [7]. 6.3.3.2 In the case where the target particles and the background occupy different grayscale ranges in the particle color image and have a strong grayscale contrast, the color image can be grayed and then threshold segmentation can be performed based on the grayscale image. If the particles and the background cannot be accurately segmented based on the grayscale image, the threshold segmentation should be based on a single component image in the color space, or the threshold segmentation of multiple component images should be combined. When the threshold segmentation cannot meet the requirements, it is advisable to consider the image segmentation method based on color space clustering. At this time, attention should be paid to selecting the optimal number of clusters and the initial cluster center. In the cluster similarity judgment criterion, the image color features, spatial relationship features and texture features can be comprehensively considered. The basic principles and steps of the color image segmentation algorithm based on fuzzy C-means clustering are shown in Appendix C. See Appendix D for an example of particle image segmentation in CIELAB color space. The segmentation results can be divided into different layers according to color information for further particle size measurement and calculation.
The particle color image can also be segmented using edge detection operators. The edge detection results on each component image can be combined, or each pixel in the image can be treated as a vector in the color space, and then the characteristics of the vector space are used for edge detection. 6.3.4 Segmentation of Adhering Particles
The particle color image analysis method can use color and spatial information to perform a certain degree of segmentation and identification of the color image of adhered particles. If there are still closely overlapping agglomerated particles or adhered particle images with similar colors, automatic and manual methods can be used for segmentation to ensure accurate and reliable segmentation of adhered particles.
6.3.5 Particle Hole Filling and Small Area Removal There may be holes or small areas composed of noise pixels in the image after particle color image segmentation. Particles with a certain degree of transparency will form bright spots during the imaging process, which will cause holes after image processing. Particle holes should be filled.
The small area caused by the noise pixel is generally significantly smaller than the particle size being measured. The small area should be removed. 6.4 Particle size measurement and calculation
For the segmented particle images, the particle size measurement and calculation should be carried out by counting pixels one by one. The direct measurement data obtained should include:
a) the projected area A of each particle;
b) the longest Feldspar diameter aFmax.i of each particle; c) the shortest Feldspar diameter aFmin. of each particle; d) the major axis length mxi of the Legendre inertia ellipse of each particle; e) the minor axis length imin.i of the Legendre inertia ellipse of each particle. The parameters of particle size analysis should include the area equivalent diameter Ai, aspect ratio AR and ellipticity ER of the particle. Before giving the final report of the quantitative particle size analysis results, it is advisable to convert the pixels into actual sizes for the convenience of quantitative analysis. The area equivalent diameter I A. of the particle is calculated according to formula (1): 5
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.