Multi-particle-size detection method and system based on particle microscopic high-definition imaging
Through microscopic high-definition imaging technology, the problems of low particle detection robustness and poor particle size information quality during multi-particle size mixing are solved, and high-quality detection and accurate particle size analysis of particle images are achieved.
Patent Information
- Application Number
- CN202411005007.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-12-20
AI Technical Summary
The existing micro-nano particle detection methods have problems such as low robustness and poor particle size information quality, especially when multiple particle sizes are mixed, it is difficult to obtain high-definition images and accurately detect particle size.
Through a multi-particle size detection method based on high-definition imaging of particle microscopy, the microscopic image is configured to obtain the microscopic image, the image definition evaluation function is determined, the particle position is marked for fusion reconstruction, the contrast texture characteristics of the particle area are analyzed, and the differences are distinguished. Particle size particles, generating bimodal or multimodal particle size distribution information.
It improves the clarity of particle images and the accuracy of particle size detection, ensures reliable expression of particle characteristic information, and improves the robustness of detection and the quality of particle size information.
Smart Images
Figure CN119169069A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of micro-nano particle detection, and in particular to a multi-particle size detection method and system based on particle microscopic high-definition imaging. Background Art
[0002] At present, products related to micro-nano particles are constantly emerging in the industrial field and scientific research field. The microscopic characteristics of particles affect the performance of related materials in various industries. For example, the particle size distribution of cement has a great relationship with the setting time and strength of concrete. In order to obtain qualified particle samples in the particle production stage, there are strict requirements for the detection method of particle characteristic information. The imaging method is a method that is easier to realize automatic particle detection. Through the imaging method, the multivariate characteristic information of particles can be obtained, and the process of traditional manual microscopy and subjective data expression can be broken. Its detection accuracy is related to the imaging quality.
[0003] For image detection methods, whether it is traditional image processing technology or deep learning technology, the primary goal is to obtain high-quality particle images. The clarity of the image will directly affect the accuracy of the image method to detect the multivariate feature information of the particles. In the process of particle microscopic imaging, since particles often exist in dispersions, the agglomeration of particles and the different particle sizes will cause the centers of each particle to be not on the same focal plane, which will result in low clarity in the local image particle area, making it impossible to achieve clarity observation and accurate image feature information extraction. In addition, the particle size distribution curve is a commonly used method to characterize particle size information today, and the mixing of particles of different sizes will reduce the accuracy of the single-peak particle size distribution characterization information of the particles. Therefore, when multiple particle sizes are mixed in the same system, it is an urgent problem to obtain high-definition images of particles and realize accurate detection of particle size.
[0004] To solve the problem of image blur in some particle areas and multiple particle size characterization, image algorithms can be used to improve image clarity and particle size characterization accuracy. However, poor particle focusing will lead to the loss of some particle information, and this lost information cannot be perfectly reproduced by the image algorithm. At the same time, affected by the complex environment, the algorithm has low robustness. It can be seen that the existing detection methods have the problems of low robustness and poor quality of particle size information. Summary of the invention
[0005] The present invention provides a multi-particle size detection method and system based on particle microscopic high-definition imaging, so as to solve the problems of low robustness and poor quality of particle size information in the existing detection methods.
[0006] In order to achieve the above object, the present invention is implemented by the following technical solutions:
[0007] In a first aspect, the present invention provides a multi-particle size detection method based on particle microscopic high-definition imaging, comprising:
[0008] configuring a particle mixed solution, obtaining a particle microscopic image of the particle mixed solution according to grayscale information, and determining a particle image clarity evaluation function based on the particle microscopic image;
[0009] Acquire an optical image of the particle mixture solution, and mark the positions of the particles in the optical image, use the marked area as the foreground area, and use the area in the optical image other than the marked area as the background area;
[0010] The foreground area and the background area are fused and reconstructed to obtain a reconstructed image, and the non-agglomerated particle size information and agglomerated particle shape information are analyzed on the reconstructed image, and the particle area contrast texture characteristics are obtained according to the analysis results;
[0011] Particles of different particle sizes are distinguished according to the particle area contrast texture characteristics to obtain bimodal distribution of particles, multi-particle size distribution information and single particle size distribution information.
[0012] Optionally, obtaining a particle microscopic image of the particle mixture solution according to grayscale information includes:
[0013] According to subjective visual judgment and grayscale information, several particle microscopic images with different focusing effects are collected, and the grayscale information of several particle microscopic images is compared, and the particle microscopic image with the highest definition is selected according to the grayscale information.
[0014] Optionally, determining a particle image clarity evaluation function based on the particle microscopic image includes:
[0015] Performing image clarity evaluation on the particle microscopic image using a Variance function and a Laplace function that are independent of the pixel direction, and comparing the image clarity obtained by the two functions;
[0016] According to the comparison result, the function with the highest image clarity is selected as the particle image clarity evaluation function of the particle microscopic image.
[0017] Optionally, marking the positions of particles in the optical image includes:
[0018] The size of the particle region is determined in the optical image, and based on the size of the particle region, a rectangular frame of a corresponding size is selected to mark the position of the particle in the optical image.
[0019] Optionally, fusing and reconstructing the foreground area and the background area to obtain a reconstructed image includes:
[0020] Compare the clarity of all foreground areas and select the foreground area with the highest clarity, and compare all background areas and select the background area with the highest clarity;
[0021] The foreground area with the highest clarity is fused and reconstructed with the background area with the highest clarity to obtain a reconstructed image.
[0022] Optionally, the analyzing the non-agglomerated particle size information and the agglomerated particle shape information on the reconstructed image and obtaining the particle region contrast texture features according to the analysis results includes:
[0023] extracting contrast data of each particle in the reconstructed image and sorting the contrast data, and determining a grouping method according to the contrast data;
[0024] When there are two particles with greatly different particle sizes in the solution, the contrast data of their respective images differ greatly, and their overall mean can be used as a threshold to distinguish the two.
[0025] When there are three or more particles with greatly different particle sizes, the particles are grouped according to the gradient changes.
[0026] Optionally, the step of distinguishing particles of different sizes according to the particle area contrast texture feature to obtain bimodal distribution of particles and multi-size distribution information and single-size distribution information includes:
[0027] The particle size data of different groups are determined according to the particle grouping, thereby generating multiple particle size distribution curves of the particles.
[0028] In a second aspect, an embodiment of the present application provides a multi-particle size detection system based on a particle microscopic high-definition program, including a processor and a memory;
[0029] Memory, used to store computer programs;
[0030] The processor is used to implement any method step described in the first aspect when executing a program stored in the memory.
[0031] Beneficial effects:
[0032] The multi-particle size detection method based on the particle microscopy high-definition program provided by the present invention applies the focused stack technology to particle microscopy imaging, first determines the image clarity evaluation function, then marks the particle area to be observed or detected, obtains the focused stack image data set according to the motor-controlled microscope, and then reconstructs the particle image to obtain a high-definition microscopic image, and after the image clarity is improved, realizes multi-particle size detection based on the particle texture features, so as to accurately and reliably express the characteristics of the particles in the image, thereby achieving the purpose of high-quality detection of the particle image. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 One of the flow charts of multi-particle size detection based on particle microscopy high-definition program according to a preferred embodiment of the present invention;
[0034] Figure 2 This is the second flow chart of multi-particle size detection based on particle microscopy high-definition program according to the preferred embodiment of the present invention;
[0035] Figure 3 A schematic diagram of image clarity evaluation using Variance and Laplace functions according to a preferred embodiment of the present invention;
[0036] Figure 4 Schematic diagram of the focused stack dataset image and position annotation of the preferred embodiment of the present invention, wherein (a) is the focused stack dataset, the number in the lower left corner represents the particle position information based on the base point, such as 8 represents 4 microns above the base point position 0, -13 represents -6.5 microns below the base point position 0, and (b) is the particle position annotation;
[0037] Figure 5 A schematic diagram of particle microscopic image reconstruction according to a preferred embodiment of the present invention;
[0038] Figure 6 Schematic diagram of reconstructed image particle feature information analysis of a preferred embodiment of the present invention, wherein (a) is the particle size error of non-agglomerated particles, and (b) is the particle shape error of agglomerated particles;
[0039] Figure 7 Schematic diagram of bimodal and unimodal particle size distribution of polystyrene particles according to a preferred embodiment of the present invention, wherein (a) is a bimodal distribution and (b) is a unimodal distribution. DETAILED DESCRIPTION
[0040] The technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0041] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the usual meanings understood by persons with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "one" do not indicate quantity restrictions, but indicate the existence of at least one. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship also changes accordingly.
[0042] See also Figure 1-7 The present application embodiment provides a multi-particle size detection method based on particle microscopic high-definition imaging, comprising:
[0043] Preparing a particle mixed solution, obtaining a particle microscopic image of the particle mixed solution according to grayscale information, and determining a particle image clarity evaluation function based on the particle microscopic image;
[0044] Acquire an optical image of the particle mixture solution, and mark the positions of the particles in the optical image, use the marked area as the foreground area, and use the area in the optical image other than the marked area as the background area;
[0045] The foreground area and the background area are fused and reconstructed to obtain a reconstructed image, and the non-agglomerated particle size information and agglomerated particle shape information are analyzed on the reconstructed image, and the particle area contrast texture characteristics are obtained according to the analysis results;
[0046] Particles of different particle sizes are distinguished according to the particle area contrast texture characteristics to obtain bimodal distribution of particles, multi-particle size distribution information and single particle size distribution information.
[0047] For the above embodiment, the specific steps are as follows:
[0048] Step 1: Determination of the clarity evaluation function. By coarsely adjusting the stage to roughly adjust the focus position, 8 microscopic images of particles with different focusing effects were collected based on subjective visual judgment and grayscale information. The clarity of the fifth image was the highest. The Variance function and Laplace function, which are independent of the pixel direction, were used to evaluate the image clarity. It was determined which of the two functions was more consistent with the results corresponding to the grayscale information of the particle area, and finally the Variance clarity evaluation function was determined to be more applicable.
[0049] Step 2: Focus stack dataset image acquisition and particle position area annotation. First, prepare a mixed solution of 5-micron and 10-micron polystyrene particles, add an appropriate amount of anhydrous ethanol dispersant, put it under a microscope, and manually adjust the microscope to find an area with better dispersion effect. Secondly, fine-tune the image focus based on an optical microscope that can be controlled by a motor and can achieve automatic focus. First, automatically adjust it to a position with higher particle clarity, and then control the motor to move the stage up and down to fine-tune the focus position. The moving step is 0.5 microns. An image is collected for each step. The maximum moving distance up and down is the ratio of the standard particle size of the particle to the step length, so that a series of image data with different focal lengths in this field of view can be obtained. Finally, a rectangular frame is used to mark the position of the particle area. The area of multiple particles constitutes the foreground area of the image, and the rest can be considered as the background area.
[0050] Step 3: High-definition particle image fusion and reconstruction. There are two paths for image fusion and reconstruction: one is to fuse the particle area with the highest clarity and the background area in the data set to achieve clear observation of the entire image; the other is to fuse the particle area with the highest clarity and the background area with the lowest clarity in the data set, that is, to blur the background and clarify the foreground area to reduce the impact of background information on image processing and extraction of particle feature information. According to the highest clarity position of the foreground and background areas, the clarity of the marked particle positions in the data set is evaluated to obtain a series of particle area clarity data. The area position corresponding to the data with the highest clarity value is the best focus position at that position. In this way, the maximum clarity position of each particle area and background area in the data set is determined, and the background lowest clarity position is the image position corresponding to its lowest clarity data. Then, according to the particle detection requirements, the image fusion and reconstruction path is selected. Since the experimental background interference is low, path 1 is selected here to generate a high-definition image, and the clarity value of the reconstructed image is higher than the clarity of any image in the data set.
[0051] Step 4: Realize multi-size particle detection based on particle texture features. By analyzing the texture feature data of 5-micron and 10-micron polystyrene particles in the image, particles of different sizes can be distinguished by contrast data. The contrast data of each particle in the reconstructed image is extracted and sorted, and then the grouping method is determined according to the contrast data. When there are two particles with large differences in particle size in the solution, the contrast data of each image differs greatly, and their overall mean can be used as a threshold to distinguish the two; when there are three or more particles with large differences in particle size, the particles are grouped according to the gradient change, that is, in the data sequence, the contrast value of the last particle belonging to the same particle size range changes more than the contrast value of the next particle, and its gradient is larger. Among them, agglomerated particles can be removed by morphological thresholds such as area. Finally, the particle size data of different groups are determined according to the particle grouping, so as to generate multiple particle size distribution curves of the particles.
[0052] The clarity value of the reconstructed image of 5-micron and 10-micron polystyrene particles is 24881.94, which is higher than the clearest image in the focus stack dataset. At the same time, an image with good focus and an image with poor focus are selected from the dataset, and the particle size information of non-agglomerated particles and the particle shape information of agglomerated particles in the reconstructed image are analyzed. It can be seen that the particle size data of the reconstructed image has an error of 2.77%, 2.5%, 4.21%, 3.8%, 1.72%, 0.39%, 1.26%, 3.55%, 1.65%, 0.21%, 1.89% and 2.76% compared with the standard data of the particles, and the particle shape data error is 3.67%, 5.93%, 7.97%, 4.6%, 3.13% and 5.23% respectively, with the lowest error, and its characteristic data is closer to the true value. The standard data is obtained by manually measuring the particle area multiple times. Therefore, the clarity of the particle reconstructed image is improved, and the characteristic information of the particles is more accurate.
[0053] The bimodal distribution of the two polystyrene particles was obtained. The D10, D50, and D90 data of the 5-micron polystyrene particles were 5.3348, 5.5508, and 5.7668, respectively. The D10, D50, and D90 data of the 10-micron polystyrene particles were 9.9546, 10.0957, and 10.2368, respectively. The D10, D50, and D90 data of the unimodal distribution were 4.7751, 7.6912, and 10.6073, respectively. Therefore, the multi-particle size distribution can better characterize the characteristic information of the mixed particles.
[0054] The embodiment of the present application also provides a multi-particle size detection system based on particle microscopic high-definition imaging, including a processor and a memory;
[0055] Memory, used to store computer programs;
[0056] The processor is used to implement any method step in the multi-particle size detection method based on particle microscopic high-definition imaging when executing the program stored in the memory.
[0057] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.
Claims
1. A multi-particle size detection method based on high-definition particle microscopic imaging, characterized in that: include: Preparing a particle mixed solution, obtaining a particle microscopic image of the particle mixed solution according to grayscale information, and determining a particle image clarity evaluation function based on the particle microscopic image; Acquire an optical image of the particle mixture solution, and mark the positions of the particles in the optical image, use the marked area as the foreground area, and use the area in the optical image other than the marked area as the background area; The foreground area and the background area are fused and reconstructed to obtain a reconstructed image, and the non-agglomerated particle size information and agglomerated particle shape information are analyzed on the reconstructed image, and the particle area contrast texture characteristics are obtained according to the analysis results; Particles of different particle sizes are distinguished according to the particle area contrast texture characteristics to obtain bimodal distribution of particles, multi-particle size distribution information and single particle size distribution information.
2. The multi-particle size detection method based on particle microscopic high-definition imaging according to claim 1 is characterized in that: The step of obtaining a microscopic image of particles of the particle mixture solution according to grayscale information comprises: According to subjective visual judgment and grayscale information, several particle microscopic images with different focusing effects are collected, and the grayscale information of several particle microscopic images is compared, and the particle microscopic image with the highest definition is selected according to the grayscale information.
3. The multi-particle size detection method based on particle microscopic high-definition imaging according to claim 1 is characterized in that: The step of determining a particle image clarity evaluation function based on the particle microscopic image comprises: Performing image clarity evaluation on the particle microscopic image using a Variance function and a Laplace function that are independent of the pixel direction, and comparing the image clarity obtained by the two functions; According to the comparison result, the function with the highest image clarity is selected as the particle image clarity evaluation function of the particle microscopic image.
4. The multi-particle size detection method based on particle microscopic high-definition imaging according to claim 1 is characterized in that: The marking of the positions of the particles in the optical image comprises: The size of the particle region is determined in the optical image, and based on the size of the particle region, a rectangular frame of a corresponding size is selected to mark the position of the particle in the optical image.
5. The multi-particle size detection method based on particle microscopic high-definition imaging according to claim 1 is characterized in that: The fusing and reconstructing the foreground area and the background area to obtain a reconstructed image includes: Compare the clarity of all foreground areas and select the foreground area with the highest clarity, and compare all background areas and select the background area with the highest clarity; The foreground area with the highest clarity is fused and reconstructed with the background area with the highest clarity to obtain a reconstructed image.
6. The multi-particle size detection method based on particle microscopy high-definition program according to claim 1 is characterized in that: The step of analyzing the non-agglomerated particle size information and the agglomerated particle shape information on the reconstructed image and obtaining the particle area contrast texture features according to the analysis results includes: extracting contrast data of each particle in the reconstructed image and sorting the contrast data, and determining a grouping method according to the contrast data; When there are two particles with greatly different particle sizes in the solution, the contrast data of their respective images differ greatly, and their overall mean can be used as a threshold to distinguish the two. When there are three or more particles with greatly different particle sizes, the particles are grouped according to the gradient changes.
7. The multi-particle size detection method based on particle microscopy high-definition program according to claim 1 is characterized in that: The method of distinguishing particles of different particle sizes according to the particle area contrast texture characteristics to obtain bimodal distribution of particles, multi-particle size distribution information and single particle size distribution information includes: The particle size data of different groups are determined according to the particle grouping, thereby generating multiple particle size distribution curves of the particles.
8. A multi-particle size detection system based on a particle microscopic high-definition program, characterized in that: Including processor and memory; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 7 when executing a program stored in a memory.