Search

CN-121213549-B - Particle size detection method and system based on image processing

CN121213549BCN 121213549 BCN121213549 BCN 121213549BCN-121213549-B

Abstract

The application relates to the technical field of image processing, in particular to a particle size detection method and system based on image processing; the method comprises the steps of obtaining a particle image, preprocessing the particle image to obtain an image to be detected, constructing a density index and boundary adhesion degree for any pixel point in the image to be detected, taking a density index normalization result and a boundary adhesion degree normalization result corresponding to the pixel point as segmentation feature vectors of the pixel point, taking each pixel point in the image as a seed point, segmenting the image by using a region growing method based on the distance of each segmentation feature vector to obtain a segmented image, extracting a particle region in the segmented image, and analyzing particle quality. The application has the effect of improving the accuracy of particle quality detection.

Inventors

  • LIU SHAOHUI
  • LI YUANCHUN
  • LIU SHAOXING

Assignees

  • 广东聚诚智能科技有限公司

Dates

Publication Date
20260508
Application Date
20251125

Claims (8)

  1. 1. The particle size detection method based on image processing is characterized by obtaining a particle image and preprocessing the particle image to obtain an image to be detected; For any pixel point in an image to be detected, constructing an observation window, and constructing a density index reflecting the density of particles in the current area according to the gray value and the change of the gradient value in the observation window, wherein the gradient mean value and the gradient standard deviation of the gradient amplitude of the pixel point in the observation window are obtained; the method comprises the steps of obtaining a second derivative value of a pixel point in an observation window by using a Laplace operator, analyzing the gray level difference of the pixel point and the whole gray level in the observation window, building boundary adhesion based on the second derivative value and the gray level difference, obtaining a change trend index based on the second derivative value, building a local gray level difference based on the difference between a gray level mean value in the observation window and the gray level value of the pixel point, taking the ratio of the local gray level difference to the standard deviation of the gray level value of the pixel point in the observation window as a gray level valley index, taking the product of the change trend index and the gray level valley index as boundary adhesion, and taking the density index normalization result corresponding to the pixel point and the boundary adhesion normalization result as segmentation feature vectors of the pixel point; The method comprises the steps of taking each pixel point in an image as a seed point, dividing the image by using a region growing method based on the distance of each divided feature vector to obtain a divided image, extracting a particle region from the divided image, and analyzing the particle quality.
  2. 2. The method for detecting particle size based on image processing of claim 1, wherein the directions corresponding to the gray level co-occurrence matrix of the observation window at least comprise 0 degree direction, 45 degree direction, 90 degree direction and 135 degree direction.
  3. 3. The method according to claim 1, wherein the step of obtaining the trend index based on the second derivative value includes taking the second derivative value as the trend index in response to the second derivative value being greater than 0 and taking 0 as the trend index in response to the second derivative value being less than 0.
  4. 4. The method for detecting particle size based on image processing of claim 1, wherein the step of constructing the local gray scale difference based on a difference between the gray scale average value and the pixel gray scale value in the observation window comprises setting the difference as the local gray scale difference in response to the difference between the gray scale average value and the pixel gray scale value in the observation window being greater than 0, and setting 0 as the local gray scale difference in response to the difference between the gray scale average value and the pixel gray scale value in the observation window being less than 0.
  5. 5. The method for detecting particle size based on image processing of claim 1, wherein the step of constructing the observation window comprises taking a square window with the pixel as a center and within a preset side length as an observation window of the pixel for any pixel.
  6. 6. The method for detecting grain size based on image processing of claim 1, wherein the step of extracting the grain region from the segmented image comprises calculating a mean value of boundary adhesion in the segmented region as a boundary index of the segmented region for any one of the segmented regions in the segmented image, and taking the segmented region as the grain region in response to the boundary index of the segmented region being smaller than a preset boundary threshold.
  7. 7. The method for detecting particle size based on image processing as defined in claim 1, wherein the step of analyzing the particle quality includes taking a mean value of position coordinates of each pixel point in the particle area as a center point, taking a mean value of Euclidean distances between each edge pixel point and the center point in the particle area as a center distance of the particle area, taking an inverse of a standard deviation of the center distance as a particle quality index, setting a quality threshold, and determining that the particle quality of the area is disqualified in response to the particle quality index being smaller than the quality threshold.
  8. 8. An image processing based particle size detection system comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement an image processing based particle size detection method according to any of the preceding claims 1-7.

Description

Particle size detection method and system based on image processing Technical Field The application relates to the technical field of image processing, in particular to a particle size detection method and system based on image processing. Background Underwater pelletizers are used to produce a wide variety of polymers and thermoplastics, and primarily spherical particles. Compared with the traditional granulating process, the underwater granulating technology can provide more uniform and round particle raw materials for industrial production. Particle size detection is a key quality control link in the granulation industry, and the performance of the product in the aspects of fluidity, solubility and the like can be effectively evaluated through measurement of particle size distribution. The traditional granularity detection method comprises screening analysis, microscopic image measurement and the like, has the problems of low detection efficiency, high manual participation, difficulty in realizing real-time online detection and the like, and cannot meet the requirements of modern industrial continuous production and fine quality control. Therefore, in recent years, particle size detection technology based on image processing is gradually developed, the method relies on an image segmentation technology to extract the boundary of single particles, and then automatic measurement of the particle size is realized through statistical analysis, so that the method has the advantages of non-contact and high efficiency. The region growing method is a common image segmentation method, and mainly based on the gray level difference of pixel points as a growing condition, if the gray levels of the pixels are similar, the same region is included, and the segmentation of the image is completed. In the related art, for the situation that materials in an image are sparse, the traditional region growing method can be effectively divided due to the fact that the background and the materials have large gray scale difference. However, the granular raw materials formed after finishing the granulation of the underwater granulator are usually in a stacked state in the conveying process, and for the situation, the gradient transition is smooth due to the dense granules, so that the boundary gray level difference is not obvious. And then make the in-process that regional growth is difficult to obtain effective granule region, appear a plurality of granule and discern as the condition in a region easily, and then lead to the condition that the accuracy degree was relatively poor to the granule detection. Disclosure of Invention In order to solve the problem of low detection accuracy in the traditional particle detection technology, the application provides a particle size detection method and system based on image processing. In a first aspect, the present application provides a method for detecting particle size based on image processing, which adopts the following technical scheme: the particle size detection method based on image processing comprises the steps of acquiring a particle image and preprocessing to obtain an image to be detected, constructing an observation window for any pixel point in the image to be detected, and constructing a density index reflecting the particle density of the current area according to the change of a gray value and a gradient value in the observation window; The method comprises the steps of obtaining a second derivative value of a pixel point in an observation window by using a Laplacian operator, analyzing gray level difference of the pixel point and integral gray level in the observation window, constructing boundary adhesion degree based on the second derivative value and the gray level difference, taking a density index normalization result corresponding to the pixel point and the boundary adhesion degree normalization result as segmentation feature vectors of the pixel point, segmenting an image by using a region growing method based on the distance of each segmentation feature vector by taking each pixel point in the image as a seed point to obtain a segmented image, extracting a particle region in the segmented image, and analyzing particle quality. Firstly, an observation window is constructed for each pixel point, and a density index representing the distribution density degree of particles is extracted from gray scale and gradient dimensions. The relation among the gradient mean value, the gray level co-occurrence matrix entropy value and the gradient standard deviation is introduced, so that the structural complexity and the edge characteristic change of the area around the pixel are comprehensively reflected, and the system can sense the accumulation degree of local particles. On the basis, a second derivative value of the pixel point is calculated through the Laplacian operator, and boundary adhesion is constructed by combining the difference between window gray level average val