A particle size analysis method of a laser particle size analyzer

Abstract

The application provides a particle size analysis method of a laser particle size analyzer. The method first collects continuous frame images, then performs smoothing processing on each frame image, adjusts the weight coefficient of the RGB channel, inputs each gray-scale image into a UNet neural network to obtain a mask image, performs bilateral filtering processing and binarization processing in sequence, traverses each contour pixel in the binarization result image, calculates the size of each particulate matter in at least three directions based on the boundary coordinate index, finally, detects each filter image by using a canny edge detection operator, and normalizes the particulate matter calibrated by adjacent frame images by using a radius threshold filtering method to obtain a particle size detection result. The application no longer takes a static image as an analysis object, but corrects adjacent continuous frames in a video image, and improves the detection precision through optimization of an algorithm. The application has lower requirements for sample dispersity, higher detection precision, and outstanding technical advantages.

Description

Technical Field

[0001] This invention relates to the field of laser particle size analyzer technology, and more specifically to a particle size analysis method using a laser particle size analyzer. Background Technology

[0002] A laser particle size analyzer, also known as a laser particle size distribution analyzer, is an instrument used to measure and analyze the abundance of physical particles. Based on the dispersion system, it is classified into wet testing instruments, dry testing instruments, and combined wet and dry testing instruments. Its principle is that during light propagation, the wavefront is confined by pores or particles of a scale equivalent to the wavelength. The emission from each elementary wave at the confined wavefront interferes in space, producing diffraction and scattering. The spatial distribution of diffracted and scattered light energy is related to the wavelength of the light wave and the size of the pores or particles. Using a laser as the light source, and with the light being monochromatic with a fixed wavelength, the spatial distribution of diffracted and scattered light energy is only related to the particle size. For the diffraction of a particle group, the quantity of each particle size determines the amount of light energy obtained at each specific angle. The proportion of light energy at each specific angle in the total light energy should reflect the abundance of each particle size. Following this approach, a mathematical and physical model characterizing the abundance of particle sizes and the light energy obtained at each specific angle can be established. This allows for the development of instruments to measure light energy, and the comparison between the light energy measured at a specific angle and the total light energy can deduce the abundance proportion of the corresponding particle size size in the particle group. In existing technologies, conventional algorithms have high requirements for sample dispersion. When the uniformity of dispersion is not ideal, it is often difficult to obtain objective detection results. In addition, the detection accuracy of current conventional analysis methods also needs to be improved. Summary of the Invention

[0003] This invention aims to address the technical deficiencies of existing technologies by providing a particle size analysis method using a laser particle size analyzer, thereby solving the technical problem that current conventional algorithms have high requirements for sample dispersion.

[0004] Another technical problem that this invention aims to solve is how to further improve detection accuracy.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0006] A particle size analysis method using a laser particle size analyzer, comprising:

[0007] 1) Collect video information of the sample during injection and extract continuous frame images from the video;

[0008] 2) Smooth each frame of the image using median filtering, and then adjust the weight coefficients of the RGB channels of each frame to obtain a grayscale image;

[0009] 3) Input each grayscale image into the UNet neural network to obtain the mask image;

[0010] 4) Perform bilateral filtering on each mask image, receive annotation instructions, and form a filtered image with edge contours;

[0011] 5) Binarize each of the filtered images, traverse the pixels within each contour of the binarized result image, and then calculate the size of each particle in at least three directions based on the boundary coordinate index;

[0012] 6) The Canny edge detection operator is used to detect each of the filtered images, and then the radius threshold filtering method is used to normalize the particles marked in the adjacent frame images to obtain the particle size detection results.

[0013] Preferably, in step 1), while acquiring video information, the excitation signal generated by the sample impacting the acquisition plate is obtained, and the excitation signal is subjected to feature analysis to obtain a signal feature spectrum.

[0014] Preferably, in step 2), after smoothing, morphological processing for noise reduction, erosion, and dilation are performed sequentially.

[0015] Preferably, in step 3), when inputting each grayscale image into the UNet neural network, the vector field inside the particles is used as the target for fitting.

[0016] Preferably, in step 4), the image edges and image shadows are preserved during the bilateral filtering process, while the image texture is removed.

[0017] Preferably, in step 4), after receiving the annotation instruction, the target body is first subjected to connected component detection to obtain a filtered image with edge contours.

[0018] Preferably, step 5) involves traversing each pixel within the contour of the binarized result image, which includes: traversing each pixel in the binarized image using an algorithm and dividing the image region of each non-connected individual particle.

[0019] Preferably, in step 6), the Canny edge detection operator removes noise using Gaussian filtering.

[0020] This invention provides a particle size analysis method for a laser particle size analyzer. The method first acquires consecutive frame images, then smooths each frame and adjusts the weighting coefficients of the RGB channels. Next, each grayscale image is input into a UNet neural network to obtain a mask image. Then, bilateral filtering and binarization are performed sequentially. The pixels within each contour of the binarized image are traversed, and the size of each particle in at least three directions is calculated based on boundary coordinate indices. Finally, the Canny edge detection operator is used to detect each filtered image, and a radius threshold filtering method is used to normalize the particles identified in adjacent frames to obtain the particle size detection result. This invention no longer uses static images as the analysis object, but instead uses adjacent consecutive frames in video images for correction, and improves detection accuracy through algorithm optimization. This invention has lower requirements for sample dispersion, higher detection accuracy, and outstanding technical advantages. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in detail below. To avoid excessive and unnecessary detail, well-known structures or functions will not be described in detail in the following embodiments. The approximate language used in the following embodiments is for quantitative purposes, indicating that a certain degree of variation in quantity is permissible without changing the basic function. Unless otherwise defined, the technical and scientific terms used in the following embodiments have the same meaning as commonly understood by those skilled in the art to which this invention pertains.

[0023] Example 1

[0024] A particle size analysis method using a laser particle size analyzer, comprising:

[0025] 1) Collect video information of the sample during injection and extract continuous frame images from the video;

[0026] 2) Smooth each frame of the image using median filtering, and then adjust the weight coefficients of the RGB channels of each frame to obtain a grayscale image;

[0027] 3) Input each grayscale image into the UNet neural network to obtain the mask image;

[0028] 4) Perform bilateral filtering on each mask image, receive annotation instructions, and form a filtered image with edge contours;

[0029] 5) Binarize each of the filtered images, traverse the pixels within each contour of the binarized result image, and then calculate the size of each particle in at least three directions based on the boundary coordinate index;

[0030] 6) The Canny edge detection operator is used to detect each of the filtered images, and then the radius threshold filtering method is used to normalize the particles marked in the adjacent frame images to obtain the particle size detection results.

[0031] Example 2

[0032] A particle size analysis method using a laser particle size analyzer, comprising:

[0033] 1) Collect video information of the sample during injection and extract continuous frame images from the video;

[0034] 2) Smooth each frame of the image using median filtering, and then adjust the weight coefficients of the RGB channels of each frame to obtain a grayscale image;

[0035] 3) Input each grayscale image into the UNet neural network to obtain the mask image;

[0036] 4) Perform bilateral filtering on each mask image, receive annotation instructions, and form a filtered image with edge contours;

[0037] 5) Binarize each of the filtered images, traverse the pixels within each contour of the binarized result image, and then calculate the size of each particle in at least three directions based on the boundary coordinate index;

[0038] 6) The Canny edge detection operator is used to detect each of the filtered images, and then the radius threshold filtering method is used to normalize the particles marked in the adjacent frame images to obtain the particle size detection results.

[0039] In step 1), while acquiring video information, the excitation signal generated by the sample impacting the acquisition plate is obtained, and feature analysis is performed on the excitation signal to obtain a signal feature map. In step 2), after smoothing, morphological processing for noise reduction, erosion, and dilation are performed sequentially. In step 3), when inputting each grayscale image into the UNet neural network, the vector field inside the particle is used as the target for fitting. In step 4), during the bilateral filtering process, image edges and image shadows are preserved, while image texture is removed. In step 4), after receiving the annotation instruction, connected component detection is first performed on the target body to obtain a filtered image with edge contours. In step 5), traversing each pixel within the contour of the binarized result image includes: traversing each pixel in the binarized image using an algorithm and dividing the image region of each non-connected individual particle. In step 6), the Canny edge detection operator removes noise using Gaussian filtering.

[0040] The embodiments of the present invention have been described in detail above, but the content described is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the scope of the present invention should be included within the protection scope of the present invention.

Claims

1. A particle size analysis method using a laser particle size analyzer, characterized in that, include: 1) Collect video information of the sample during injection and extract continuous frame images from the video; 2) Smooth each frame of the image using median filtering, and then adjust the weights of the RGB channels of each frame. The coefficients are used to obtain a grayscale image; 3) Input each grayscale image into the UNet neural network to obtain the mask image; 4) Perform bilateral filtering on each mask image, receive annotation instructions, and form a filtered image with edge contours; 5) Binarize each of the filtered images, traverse the pixels within each contour of the binarized result image, and then calculate the size of each particle in at least three directions based on the boundary coordinate index to obtain particles with size assignments. 6) The Canny edge detection operator is used to detect each of the filtered images, and then the radius threshold filtering method is used to normalize the particles with size assignments marked in the adjacent frame images to obtain the particle size detection results.

2. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 1), while acquiring video information, the excitation signal generated by the sample impacting the acquisition plate is obtained, and the excitation signal is subjected to feature analysis to obtain a signal feature spectrum.

3. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 2), after smoothing, morphological processing, noise reduction, erosion, and dilation are performed in sequence.

4. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 3), when each grayscale image is input into the UNet neural network, the vector field inside the particles is used as the target for fitting.

5. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 4), the image edges and image shadows are preserved during the bilateral filtering process, while the image texture is removed.

6. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 4), after receiving the annotation instruction, the target body is first subjected to connected component detection to obtain a filtered image with edge contours.

7. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, Step 5) describes traversing each pixel within the contour of the binarized result image, which includes: traversing each pixel in the binarized image using an algorithm and dividing the image region of each non-connected individual particle.

8. The particle size analysis method of a laser particle size analyzer according to claim 1, characterized in that, In step 6), the Canny edge detection operator removes noise using Gaussian filtering.

Images