System and method for measuring particle size and particle shape of aerosol

Abstract

The invention discloses an aerosol particle size and particle shape measurement system and method. The system comprises an image acquisition module, an aerosol microscopic imaging device, a system parameter calibration module, a particulate matter acquisition module, a particle size calculation module and a particle shape calculation module. The aerosol microscopic imaging device obtains contour information of aerosol based on a transmission-type microscopic imaging method; the image acquisition module acquires an aerosol particle image through the aerosol microscopic imaging device, and draws a corresponding gray value on each pixel point of the image to obtain a complete aerosol particle image; the system parameter calibration module is used for calibrating system parameters; the particulate matter acquisition module acquires particulate matters of the aerosol particle image; the particle size calculation module calculates the aerosol particle size; the particle shape calculation module calculates the aerosol particle shape. According to the invention, the problems of small depth of field, repeated particle calculation, low calculation efficiency and the like in aerosol optical microscopic imaging are solved, and high-precision and high-efficiency measurement of the particle size and the particle shape of the aerosol is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of optical microscopy imaging, and in particular to a system and method for measuring aerosol particle size and particle shape. Background Art

[0002] Aerosol is a mixed system of gas and particles with a certain stability and low settling velocity in the gravitational field. It has an important impact on atmospheric environmental changes and human health. Aerosols have a wide range of sources and various causes of formation. The particle size distribution is between 0.001-100μm (including large-particle pollen aerosol particles), and its particle shapes include spherical, flake-shaped, needle-shaped and various irregular shapes. Aerosol particle size describes the size of aerosol particles. Aerosol particles of different sizes have different movement and transmission in the atmosphere. Large-particle aerosols will settle on the surface faster, while small-particle aerosols will usually stay in the atmosphere for a week or even longer. This has important research significance for atmospheric radiation transmission and pollutant diffusion. In addition, the size of aerosol particles also determines their ability to penetrate into human organs, which is very important for human health.

[0003] Regarding aerosol particle shape, many current atmospheric detection instruments and atmospheric radiation transfer models still treat aerosol particles as standard spherical particles. In reality, standard spherical aerosol particles only account for a small fraction of the atmosphere, leading to significant errors in instrument detection and theoretical calculations.

[0004] In recent decades, researchers at home and abroad have devoted considerable effort and financial resources to analyzing aerosol shape parameters, combining particle size, shape, and scale spectra. These efforts have focused on measuring aerosol light scattering coefficients, scattering phase functions, and absorption coefficients, all of which are used to analyze aerosol physical and optical properties. While there has been more research on aerosol particle size measurement, there has been less research on aerosol shape measurement, particularly given the current scarcity of instruments capable of simultaneously measuring aerosol size and shape. Existing aerosol detection techniques include transmission electron microscopy, dynamic light scattering, specific surface area analysis, centrifugal sedimentation, and laser particle size analysis. Related instruments include aerosol electrometers, condensation nucleus particle counters, scanning mobility spectrometers, and dust particle counters.

[0005] Existing aerosol particle size and shape detection methods include two categories: based on light scattering principle and based on imaging principle. Scattering measurement methods are divided into single-particle scattering and multi-particle scattering. The disadvantage of this method is that it is difficult to complete the measurement of a specific particle among many aerosol particles. This is because it is impossible to subdivide the scattered light output by a single particle from the scattered light of the particle group on the photodetector, which means that the method has relatively poor single-target and multi-target collaborative measurement capabilities. To complete single-particle measurement, it is necessary to design complex flow stabilization equipment and rationally plan the pipeline and nozzle size to ensure that only one aerosol particle enters the optical measurement area at a time. This is also a difficulty in mechanical manufacturing, increasing system complexity and cost. In addition, light scattering methods usually measure particle size and particle shape through inversion measurement using scattered intensity or polarization intensity at different angles. This measurement method is relatively indirect and has lower accuracy than direct imaging. The parameters output by scattering methods are relatively insufficient. In particular, in particle shape measurement, they can only distinguish whether a particle is spherical, non-spherical, or rod-shaped, but cannot output the specific concavities and convexities of the particle, making it difficult to provide a detailed description of the particle characteristics and having certain limitations. At the same time, scattering methods often cannot simultaneously consider particle size information when measuring particle shape.

[0006] The particle detection method based on optical imaging belongs to computer vision technology and can quickly and non-contactly measure aerosol particles. The limitations of this method are mainly slow calculation speed, small depth of field, low resolution, and it is restricted by the limits of optical imaging. Due to the limitations of imaging depth of field caused by numerical aperture and magnification, the current method mainly prepares the sample in advance and places it in front of the microscope. The depth of field position is found through the optical adjustment frame to complete the static measurement of the clear image. This method reduces detection efficiency and increases operational complexity. In addition, this method is limited by the field of view, and the number of single detections is small, resulting in few related studies on aerosol optical imaging. Summary of the Invention

[0007] In order to overcome the defects and shortcomings of the existing technology, the present invention provides an aerosol particle size and particle shape measurement system and method. The present invention obtains aerosol contour information based on the transmission microscopy method, and solves the problems of small depth of field, calculation of repeated particles and low computational efficiency in aerosol optical microscopy imaging through computer vision, thereby achieving high-precision and high-efficiency measurement of aerosol particle size and particle shape; by judging whether the particulate matter is within the depth of field range, the blurred aerosol particles that are not within the depth of field are eliminated. In addition, the system detects and removes repeated particles in multiple frames of images, and each calculation includes all aerosol particles in dozens of frames of images. To address the problem of low computational efficiency caused by complex algorithms in image processing, the aerosol particle size and particle shape are calculated in parallel.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] The present invention provides an aerosol particle size and particle shape measurement system, comprising: an image acquisition module, an aerosol microscopic imaging device, a system parameter calibration module, a particle acquisition module, a particle size calculation module, and a particle shape calculation module;

[0010] The aerosol microscopic imaging device is used to obtain aerosol profile information based on a transmission microscopic imaging method;

[0011] The image acquisition module collects aerosol particle images through an aerosol microscopy imaging device, analyzes each frame of collected image data to generate image grayscale values, and plots the corresponding grayscale values ​​on each pixel of the image to obtain a complete aerosol particle image;

[0012] The system parameter calibration module calibrates the system parameters, including background field calibration, pixel size calibration, depth of field characteristic grayscale value calibration and depth of field size calibration;

[0013] The particle acquisition module is used to acquire particles in aerosol particle images, specifically including: image denoising, aerosol marking, defocus detection, and duplicate particle removal;

[0014] The particle size calculation module is used to calculate the aerosol particle size, and the output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, and Fret minimum diameter;

[0015] The particle shape calculation module is used to calculate the aerosol particle shape, and the output aerosol particle shape parameters include: particle convexity, particle circularity, particle roundness, and particle sphericity.

[0016] As a preferred technical solution, the aerosol microscopy imaging device includes: an aerosol generator, an aerosol dryer, an aerosol neutralizer, an imaging cavity, an LED light source, a microscope objective lens, a tube lens, an extension tube, and a CMOS camera;

[0017] The imaging cavity is provided with an air inlet and an air outlet in the vertical direction, and flat glass is provided on both sides of the imaging cavity in the horizontal direction;

[0018] The aerosol generator generates aerosol particles, which pass through an aerosol dryer and an aerosol neutralizer in sequence, and the output port of the aerosol neutralizer is connected to the air inlet of the imaging cavity;

[0019] The LED light source is located on one side of the imaging cavity. The light beam is incident from the flat glass on one side and emitted from the flat glass on the opposite side. The light beam carrying aerosol particles is magnified by the microscope objective lens and converged onto the photosensitive surface of the CMOS camera through the tube lens and extension tube.

[0020] As a preferred technical solution, the background field calibration specifically includes:

[0021] The background field is removed from each frame of the image, leaving the aerosol particles in the foreground. The background value of each pixel is obtained in multiple background images. During calibration, multiple frames of empty background images are collected while removing the target object. The sub-minimum grayscale value at each pixel in all background images is obtained as the background value at that point. The background value of each pixel is calculated cyclically and then the background image is output;

[0022] The pixel size calibration specifically includes:

[0023] The pixel size is measured using an optical graticule with a set line spacing, the position of the optical graticule is adjusted, and the optical graticule image is acquired;

[0024] Perform binarization and morphological optimization denoising on the optical reticle image;

[0025] Marking the line pattern of the optical reticle using a line detection algorithm;

[0026] The corner points of the optical graticule image are detected through the coordinates of the straight lines;

[0027] The pixel size is calculated based on the distance between the two farthest corner points in the pixel coordinate system and the world coordinate system, and the pixel size is obtained after multiple calibrations;

[0028] The depth of field characteristic grayscale value calibration specifically includes:

[0029] Using the optical graticule as the target object, the average grayscale value of the foreground area on the graticule image is calculated according to the weight on the image. The calculation formula is expressed as:

[0030]

[0031] Among them, I(x i ,y i ) represents the gray value of the i-th pixel on the straight line of the graticule in the foreground area of ​​the image. This formula assigns a smaller weight to the pixel with a smaller gray value;

[0032] Move the optical graticule to find the maximum average grayscale value as the characteristic grayscale value of the depth of field, and take the average value after completing multiple calibrations;

[0033] The depth of field calibration specifically includes:

[0034] Move the optical reticle and calculate the line width and line length of the reticle in real time on the image. Based on the clarity evaluation algorithm, record the farthest and nearest optical reticle positions within the set measurement error. The distance between the two is the depth of field. After completing multiple calibrations, take the average value.

[0035] As a preferred technical solution, the particle acquisition module is used to acquire particles in aerosol images, specifically including: image denoising, aerosol marking, defocus detection, and repeated particle removal;

[0036] The image denoising specifically includes:

[0037] Obtain an aerosol particle foreground image after removing the background field image, perform bilateral filtering on the image, binarize the filtered image, process the binarized image using morphological closing operations, perform median filtering to remove remaining noise, and obtain an aerosol particle image;

[0038] The aerosol marker specifically includes:

[0039] Perform edge detection on the denoised image to obtain the position of aerosol particles and the coordinates of the contour points of aerosol particles;

[0040] The defocus detection specifically includes:

[0041] The vertical and horizontal distances of the aerosol particles are calculated based on the contour coordinates of the aerosol particles, and a local image of the aerosol particles is constructed. The width of the local image is twice the horizontal distance, the height is twice the vertical distance, and the center point coordinate is the average of the horizontal and vertical coordinates of the particle contour point.

[0042] Obtain the maximum grayscale value in the local image area. If this value is less than the calibrated depth of field characteristic grayscale value, the aerosol particle is determined to be a defocused particle.

[0043] The defocus degree of aerosol particles is calculated based on the defocus degree algorithm. The image variance value of the local image of the aerosol particles detected by the defocus degree is calculated, and the defocused particles with variance values ​​less than the set range are eliminated;

[0044] The repeated particle removal specifically includes:

[0045] Arrange and combine the aerosol particles, inputting two aerosol particles that have not been calculated each time;

[0046] The frame difference range for duplicate particles is set based on the gas flow rate and camera frame rate. It is then determined whether the number of frames in which two aerosol particles appear is less than the maximum frame difference. If this condition is met, a minimum bounding rectangle is constructed for each aerosol particle. The size of the minimum bounding rectangle is used to determine whether the two aerosol particles are similar in size. If the aerosol particles are similar in size, the local images of the two aerosol particles are input into the similarity evaluation algorithm to obtain a similarity index. This similarity index is then used to determine the likelihood of duplicate particles. All permutations and combinations are then recalculated. For each group of duplicate particles, only the first particle is retained, and the remaining particles are discarded.

[0047] As a preferred technical solution, the defocus algorithm evaluates the defocus of aerosol particles by combining the grayscale value difference in the local image of aerosol particles and the spatial distance between pixels. The calculation formula is as follows:

[0048]

[0049] Among them, i refers to the index of a pixel point in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), (x i ,y i ) represents the pixel coordinate value of pixel point i, (x max ,y max ) represents the pixel coordinates of the pixel with the maximum grayscale value in the local image, H and W are the height and width of the local image respectively, and n is the total number of pixels in the local image;

[0050] The image variance value of the local image is calculated using the following formula:

[0051]

[0052] Among them, i refers to the index of a pixel point in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), (x i ,y i ) represents the pixel coordinate value of pixel i, μ is the average grayscale value of the local image, and n is the total number of pixels in the local image.

[0053] As a preferred technical solution, the similarity evaluation algorithm is specifically expressed as follows:

[0054]

[0055]

[0056] Where p1 and p2 represent two aerosol particles, a(p1, p2) represents the angle between the line connecting the two aerosol particles and the X-axis, and (x1, y1) and (x2, y2) represent the coordinates of the center points of the two aerosol particles, respectively.

[0057] As a preferred technical solution, the particle size calculation module is used to calculate the aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, and Fret minimum diameter. The specific calculation method includes:

[0058] The vertical distance and horizontal distance represent the maximum distance between particle contour points in the vertical and horizontal directions respectively;

[0059] The equivalent circle diameter refers to the diameter of a circle with the same area as the particle area;

[0060] The Fret diameter is determined by the distance between two parallel tangent lines tangent to the particle. All tangent lines of the current angle are obtained at multiple angles. The largest tangent line spacing is taken as the Fret diameter at that angle. The maximum Fret diameter and the minimum Fret diameter are selected.

[0061] As a preferred technical solution, the particle shape calculation module is used to calculate the aerosol particle shape. The output aerosol particle shape parameters include: particle convexity, particle roundness, particle roundness, and particle sphericity. The specific calculation method includes:

[0062] By detecting whether the line between the aerosol contour points is contained inside the aerosol particle, its convex hull region is constructed and the convex hull area is calculated. The particle convexity is obtained by dividing the particle area by the convex hull area of ​​the particle.

[0063] The particle circularity was calculated from the area and perimeter of the particle;

[0064] The particle roundness is obtained by obtaining points with curvature radii within a set range on the particle contour and dividing the average curvature radius by the minimum circumscribed circle radius of the particle.

[0065] The sphericity of a particle is expressed as the ratio of the particle's equivalent circular circumference to its actual circumference.

[0066] As a preferred technical solution, a parallel computing module is also provided. The parallel computing module is based on the CPU and GPU to coordinate and parallelly calculate the particle size and particle shape, specifically including:

[0067] Parallel calculation of parameters on the CPU side, including minimum enclosing rectangle, minimum enclosing circle, convex hull and particle area;

[0068] Parallel logic is built on the GPU side. When calculating horizontal distance, vertical distance, particle perimeter, curvature radius, tangent slope, and roundness parameters, a separate thread is created for each contour point of all aerosol particles. The calculation of each parameter occupies an independent CUDA stream and kernel function, and a synchronization mechanism is used to transmit data.

[0069] When calculating the Fret diameter, a thread block is constructed for each particle, and a thread is constructed for each angle in each thread block to achieve parallel calculation, thereby synchronously outputting the Fret diameter of all particles at each angle.

[0070] The present invention also provides a method for measuring aerosol particle size and particle shape, comprising the following steps:

[0071] Obtain aerosol profile information based on transmission microscopy;

[0072] Collect aerosol particle images, analyze each frame of image data to generate image grayscale values, and plot the corresponding grayscale values ​​on each pixel of the image to obtain a complete aerosol particle image;

[0073] Calibrate system parameters, including background field calibration, pixel size calibration, depth of field feature grayscale value calibration, and depth of field size calibration;

[0074] Acquire particulate matter from aerosol particle images, including image denoising, aerosol labeling, defocus detection, and duplicate particle removal;

[0075] Calculate aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, maximum Fret diameter, and minimum Fret diameter.

[0076] Calculate aerosol particle shape. The output aerosol particle shape parameters include: particle convexity, particle circularity, particle roundness, and particle sphericity.

[0077] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0078] (1) The present invention obtains aerosol profile information based on transmission microscopy and uses computer vision to solve the problems of small depth of field, repeated particle calculation and low computational efficiency in aerosol optical microscopy, thereby achieving high-precision and high-efficiency measurement of aerosol particle size and shape;

[0079] By determining whether the particles are within the depth of field, blurred aerosol particles that are not within the depth of field are eliminated. In addition, the detection is performed in multiple frames of images, and all aerosol particles in dozens of frames are calculated each time, and duplicate particles are removed.

[0080] To address the problem of low computational efficiency caused by complex algorithms in image processing, parallel calculations of aerosol particle size and shape are performed.

[0081] (2) The aerosol microscopy imaging device of the present invention is simple to operate and highly reliable. Compared with existing light scattering measurement equipment, it is cheaper and more stable. In addition, the device has the characteristics of small footprint, light weight, and high degree of automation.

[0082] It can also directly detect aerosols. Compared with most light scattering measurement equipment, the present invention measures aerosol particle size and shape through image processing, without the need for parameter inversion, and can complete aerosol detection more directly and accurately.

[0083] (3) Compared with other microscopic imaging technologies, the technology proposed in the present invention combines image denoising, edge detection with defocus detection, and repeated particle removal, thereby effectively removing areas of no interest and retaining targets that need to be detected. The targets that can be extracted from the image are smaller, more accurate, and faster, and aerosol particles of 10nm-100μm can be obtained.

[0084] (4) Whether in the macroscopic or microscopic fields, the use of image analysis technology to measure the particle size and shape of target objects usually has the problem of low computational efficiency. The present invention innovatively realizes comprehensive and highly parallel operation of many particle size and particle shape parameters. The calculation speed is related to the aerosol particle concentration. In general, the synchronous output of parameters and images of the present invention can almost reach the frame rate of the camera itself, achieving high-speed operation while outputting more parameters with higher accuracy, and has a very low parallel cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 Schematic diagram of the implementation architecture of aerosol microscopy imaging in the aerosol particle size and shape measurement system of the present invention;

[0086] Figure 2 Schematic diagram of the structure of the imaging cavity of the present invention;

[0087] Figure 3 A schematic diagram of the process of calibrating system parameters of the present invention;

[0088] Figure 4 Schematic diagram of the process of obtaining aerosol particle image data of the present invention;

[0089] Figure 5 Schematic diagram of the process of parallel calculation of aerosol particle size and particle shape of the present invention.

[0090] Among them, 1-air inlet, 2-flat glass, 3-outer frame, 4-fixing hole, 5-air outlet. DETAILED DESCRIPTION

[0091] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0092] Example 1

[0093] This embodiment provides an aerosol particle size and particle shape measurement system, including: an image acquisition module, an aerosol microscopic imaging device, a system parameter calibration module, a particle acquisition module, a particle size calculation module, and a particle shape calculation module;

[0094] Aerosol microscopy imaging device: aerosol generator, aerosol dryer, aerosol neutralizer, imaging cavity, LED light source, microscope objective lens, tube lens, extension tube, CMOS camera;

[0095] The imaging cavity is provided with an air inlet and an air outlet in the vertical direction, and flat glass is provided on both sides of the imaging cavity in the horizontal direction;

[0096] The aerosol generator generates aerosol particles, which pass through the aerosol dryer and the aerosol neutralizer in sequence. The output port of the aerosol neutralizer is connected to the air inlet of the imaging cavity.

[0097] The LED light source is installed on one side of the imaging cavity. The light beam is incident from the flat glass on one side and emitted from the flat glass on the opposite side. The light beam carrying aerosol particles is magnified by the microscope objective lens and converged onto the photosensitive surface of the CMOS camera through the tube lens and extension tube.

[0098] This embodiment comprehensively considers factors such as imaging depth of field, imaging field of view, magnification, and frame rate, and adopts an infinity-corrected microscope optical system with an aerosol generator as the main hardware structure. The light source of this embodiment adopts an LED (Light-Emitting Diode) white light source. A laser light source is not used because the monochromatic nature of the laser will easily produce interference fringes on the image plane, thereby affecting the imaging effect. The microscope objective lens adopts an infinity-corrected optical objective lens, which outputs parallel light. An additional optical lens can be equipped with an additional optical lens at the rear end of the objective lens. The function of the tube lens and extension tube is to converge the parallel light beams output by the objective lens onto the photosensitive surface of the camera. The tube lens is mainly used for focusing. The photoelectric sensor adopts a CMOS camera with a global shutter. Both the CMOS camera and the LED light source have external triggering functions. The CMOS camera also has a software control interface. The driving power supply of the LED light source has a trigger mode and a continuous light mode.

[0099] like Figure 1 As shown, the aerosol microscopy imaging of this embodiment is based on transmission microscopy, which uses the principle of light resistance to obtain aerosol contour information on the image. The aerosol particles are generated by an atomizing aerosol generator or natural aerosol. To prevent the aerosol particles from containing moisture and thus overestimating their particle size, they are passed through an aerosol dryer to remove moisture before entering the imaging chamber. To further prevent entanglement between aerosol particles due to static electricity and to prevent adsorption of particles by the tube wall, the aerosol particles are further input into an aerosol neutralizer to achieve electrostatic equilibrium of the particles. Finally, the aerosol particles output by the aerosol neutralizer enter the imaging chamber. The aerosol particles in the optical measurement area are recorded as image information and input into a computer for identification.

[0100] like Figure 2As shown, gas is input from the air inlet 1 of the imaging chamber and output from the top to the air outlet 5. The LED light source emits a nearly parallel white light beam, which is incident on the flat glass 2 at one end of the imaging chamber and emitted from the flat glass on the opposite side. It passes through aerosol particles on the way. The light beam carrying its contour information is then magnified by the microscope objective lens and converged onto the photosensitive surface of the CMOS camera through the tube lens and extension tube. This embodiment also provides an outer frame 3 for fixing the flat glass and a fixing hole 4 for the imaging chamber.

[0101] In this embodiment, the steps of performing image acquisition by the image acquisition module include:

[0102] (1) Connect the CMOS camera to the PC via a cable, and collect aerosol particle images using the CMOS camera according to the aerosol flow rate;

[0103] (2) The driving power supply of the LED light source is adjusted to the trigger mode. After the aerosol particles are introduced into the imaging chamber, the image stream is started on the software, and image data is continuously captured from the image stream according to the frame rate of the camera. Each time the image data is captured, the software triggers the CMOS camera to output a TTL level from the strobe interface to the driving power supply of the LED light source, thereby triggering the LED light source to generate a strobe. The duration of each strobe is related to the set value of the CMOS camera exposure time.

[0104] (3) Aerosol particles are photographed in a loop, and each frame of image data captured is input into the program for data analysis to generate image grayscale values, and the corresponding grayscale values ​​are plotted on each pixel point of the image to obtain a complete photographic image.

[0105] In this embodiment, the image acquisition module adopts a flash frequency, high speed, intelligent and fully automatic control image acquisition method, which helps to obtain aerosol particles with complex movement directions and fast flow rates, and can effectively avoid the image ghosting phenomenon that is prone to occur in dynamic image analysis methods.

[0106] Since this embodiment uses computer vision technology in microscopic space, the size base of the target object is usually at the nanometer or micrometer level. Therefore, the accuracy of calibration is particularly important to reduce measurement errors. Figure 3 As shown in the figure, the system parameter calibration module calibrates the system parameters, including background field calibration, pixel size calibration, depth of field characteristic grayscale value calibration, and depth of field size calibration. After fixing the position of the optical element and ensuring that the light intensity output by the LED light source remains unchanged, the calibration can be started. The specific steps include the following:

[0107] (1) Background field calibration: In order to avoid the influence of impurities on the image plane and enhance the contrast between aerosol particles and background images during the image acquisition process, this embodiment removes the background field from each frame of the acquired image, leaving only the aerosol particles in the foreground. The background field calibration is to obtain the background value of each pixel in multiple background images. During the calibration, dozens of frames of empty field background images are collected while removing the target object. Subsequently, the sub-minimum grayscale value at each pixel point in all background images is obtained as the background value at that point. The background value of each pixel point is calculated cyclically and the background image is output.

[0108] (2) Pixel size calibration: The pixel size indicates the correspondence between the length or width of a pixel in the image and the horizontal and vertical distances in real space. The accuracy of the pixel size directly determines the accuracy of aerosol particle size measurement.

[0109] This embodiment uses a standard optical graticule with a fixed line spacing to measure the system pixel size. The implementation scheme is as follows: First, the graticule is fixed to the optical adjustment frame in front of the microscope objective lens using a clamp, and the position of the graticule is adjusted so that its center is approximately at the center of the microscope objective lens;

[0110] Turn on the system to capture the reticle image. Fine-tune the front and back position of the reticle using the optical adjustment mount and observe on the software until a clear reticle image is obtained.

[0111] Perform binarization and morphological optimization denoising on the image;

[0112] Mark the line pattern of the reticle using a line detection algorithm;

[0113] Detect the corner points of the image through the coordinates of the straight line;

[0114] Calculate the pixel size based on the distance between the two farthest corner points in the pixel coordinate system and the world coordinate system;

[0115] The system pixel size is obtained after multiple calibrations.

[0116] (3) Depth of field characteristic grayscale value calibration: The characteristic grayscale value that can determine whether a particle is located in the depth of field is an important parameter in this embodiment to describe whether a particle can be used for detection. Observe the image after removing the background. The aerosol particles appear highlighted within the depth of field, that is, the grayscale value of the local area of ​​the particle is larger. Based on this, in order to be able to collect aerosol particles within the depth of field area, it is necessary to obtain the characteristic grayscale value of the depth of field, so that the aerosol particles whose maximum grayscale value cannot reach the characteristic grayscale value are considered to be in a defocused state. For the characteristic grayscale value calibration, the graticule is also used as the target object. The average grayscale value of the foreground area on the graticule image is calculated according to the weight on the image. The calculation formula is as follows:

[0117]

[0118] Among them, I(x i ,y i ) represents the grayscale value of the i-th pixel on the straight line of the graticule in the foreground area of ​​the image. This formula assigns smaller weights to pixels with smaller grayscale values, thereby avoiding errors caused by black spots and holes in the image;

[0119] By moving the graticule through the optical adjustment frame, the maximum average grayscale value is found as the characteristic grayscale value of the depth of field. After completing multiple calibrations, the average value is taken to reduce the error.

[0120] (4) Depth of field calibration: This embodiment only collects clear aerosol particles that fall within the depth of field. Therefore, the depth of field must be calibrated in advance for volume calculation of the collection area to facilitate output of aerosol volume concentration. During calibration, the graticule is similarly moved using the optical adjustment mount, and the line width and length of the graticule are calculated in real time on the image. A clarity evaluation algorithm is constructed to record the farthest and closest graticule positions within a small range of measurement error. The distance between the two is the depth of field. After completing multiple calibrations, the average value is taken to reduce the error.

[0121] like Figure 4 As shown in the figure, after collecting aerosol images, the particle acquisition module is used to obtain particles in the aerosol image. The particle acquisition is divided into four parts: image denoising, aerosol labeling, defocus detection, and duplicate particle removal. The input image of this part is the aerosol particle foreground image after removing the background field image from the original image. At this time, the background is black and the foreground is white, with a large contrast. Dozens of frames of aerosol images are acquired each time and input. The specific steps are as follows:

[0122] (1) Image denoising: Image processing techniques are used to perform denoising, thereby highlighting aerosol particles and helping to increase the accuracy of subsequent particle extraction.

[0123] First, bilateral filtering is performed on the entire image. Bilateral filtering has good edge preservation and can effectively filter out areas of no interest. The filtered image is then binarized and processed using a morphological closing operation, which effectively fills particle holes and removes noise to a certain extent. Finally, median filtering is performed to remove the remaining noise. After denoising, only the aerosol particles with high contrast and clarity remain on the image, facilitating subsequent edge detection.

[0124] (2) Aerosol marking: perform edge detection on the denoised image to obtain the particle position and the coordinates of the particle contour points;

[0125] (3) Defocus detection: The aerosol particles obtained in the above steps are further judged for their degree of defocus, so as to retain only the clear particles within the depth of field and eliminate the blurred particles. This embodiment constructs three conditions for defocus detection: First, the vertical distance and horizontal distance of the particle are calculated based on the contour coordinates of the particle to construct a local image of the particle. The width of the local image is twice the horizontal distance; the height is twice the vertical distance; the center point coordinates are the average of the horizontal and vertical coordinates of the particle contour point. The maximum grayscale value in the local image area is obtained. If the value is less than the characteristic grayscale value of the depth of field calibrated above, it indicates that no pixel point in the particle area can meet the highlight feature of the depth of field, and the particle is directly determined to be a defocused particle. If the characteristic grayscale value is detected, the local image is input into the defocus algorithm. The defocus algorithm proposed in this embodiment evaluates the defocus of the particle by combining the grayscale value difference in the local image of the particle and the spatial distance between the pixels, and considers that the particle with a small defocus degree is a blurred particle. The calculation formula is as follows:

[0126]

[0127] Among them, i refers to the index of a pixel point in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), (x i ,y i ) represents the pixel coordinate value of pixel i, (x max ,y max ) represents the pixel coordinates of the pixel with the maximum grayscale value in the local image, H and W are the height and width of the local image respectively, and n is the total number of pixels in the local image;

[0128] As can be seen from the formula, the distance factor is introduced into the defocus test, thus increasing the influence of farther pixels on the results. This is because the grayscale value of farther pixels can better express the blur of the particles. If the particles are clear, the image is more concentrated, and the farther pixels have a smaller grayscale. The image variance value of the local image is calculated based on the particles detected by defocus. The calculation formula is as follows:

[0129]

[0130] Among them, i refers to the index of a pixel point in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), (x i ,y i ) represents the pixel coordinate value of pixel i, μ is the average grayscale value of the local image, and n is the total number of pixels in the local image.

[0131] The size of the variance value is related to the degree of image discreteness. The more blurred the particle, the more discrete its local image is, and the smaller the variance value. This step eliminates out-of-focus particles with small variance values, and finally retains the clear particles that pass all tests for the next calculation;

[0132] (4) Duplicate particle removal: Due to the high camera frame rate and slow gas flow rate, the same particle is likely to appear in multiple adjacent frames. Repeatedly counting the same particle not only wastes computing resources but also leads to errors in the calculation of aerosol volume concentration. Therefore, this embodiment constructs a duplicate particle detection method so that duplicate particles are counted only once.

[0133] First, the aerosol particles input in the above steps are arranged and combined, and two uncalculated particles are input each time; the frame difference range of the repeated particles is given according to the gas flow rate and the camera frame rate, so as to determine whether the number of frames in which the two particles appear is less than the maximum frame difference; if the conditions are met, a minimum bounding rectangle is constructed for each of the two particles, and the size of the minimum bounding rectangle is used to determine whether the two particles are similar in size; if the particle sizes are similar, the local images of the two particles are input into the similarity evaluation algorithm, and the two particles are compared in three dimensions: brightness, contrast, and structure; since the airflow direction designed in this embodiment is from top to bottom, and the repeated particles in different frames almost all exist in the airflow direction, it means that the direction of movement of the particles is nearly perpendicular to the X-axis of the image. Based on this principle, the angle between the line connecting the positions of the two particles and the X-axis is calculated, and the angle is combined with SSIM to calculate the similarity index. The calculation formula is as follows:

[0134]

[0135]

[0136] Here, p1 and p2 represent two aerosol particles, a(p1, p2) represents the angle between the line connecting them and the x-axis (normalized to 0-1), and (x1, y1) and (x2, y2) represent the coordinates of the center points of the two particles, respectively. A higher similarity index indicates a higher likelihood that the two particles are duplicates. All permutations and combinations are recalculated, and for each set of duplicate particles, only the first particle is retained, discarding the remaining particles to avoid redundant calculations.

[0137] In the above calculations, all images are foreground images after removing the background field. The final aerosol particle contour data obtained will be stored in the computer memory as the input parameters for particle shape and size calculation.

[0138] In this embodiment, the particle size calculation module is used to calculate the aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, maximum Fret diameter, and minimum Fret diameter, specifically including:

[0139] Aerosol particle size calculation: The aerosol particle size output by this embodiment includes five parameters: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, and Fret minimum diameter. Among them, the vertical distance and horizontal distance represent the maximum distance between the particle contour points in the vertical and horizontal directions respectively, which are calculated by the maximum and minimum coordinates of the particle contour points; the equivalent circle diameter refers to the diameter of a circle with the same area as the particle area. The particle area is a mathematical calculation based on Green's theorem for the area enclosed by a closed curve in a plane; the Fret diameter is determined by a set of two parallel tangent distances tangent to the particle, so the Fret diameter is a set of diameters. The present invention obtains all tangents of the current angle at multiple angles, and the largest tangent spacing is taken as the Fret diameter at that angle by arrangement and combination, thereby outputting a set of Fret diameters. For the expression of particle size, only the maximum and minimum Fret diameters are used;

[0140] In this embodiment, the particle shape calculation module is used to calculate the aerosol particle shape. The output aerosol particle shape parameters include: particle convexity, particle circularity, particle roundness, and particle sphericity, specifically including:

[0141] Aerosol particle shape calculation: The aerosol particle shape output by this embodiment includes four parameters: particle convexity, particle circularity, particle roundness, and particle sphericity. Among them, the particle convexity is calculated by dividing the area of ​​the particle by the convex hull area of ​​the particle. The convex hull area is constructed and the area is calculated by detecting whether the line between the aerosol contour points is contained in the interior of the aerosol particle; the particle circularity indicates the degree to which the particle is close to a circle. It is calculated by the area and circumference of the particle. The closer the value is to 1, the rounder the particle; the particle roundness indicates whether the surface shape of the particle is smooth and the uniformity of its internal structure. When calculating, the particle roundness is obtained by obtaining points with larger curvature radius on the particle contour and dividing their average curvature radius by the minimum circumscribed circle radius of the particle. The closer the value is to 1, the smoother the particle; the sphericity of the particle is defined as the ratio of the particle equivalent circle circumference to the actual circumference. The actual circumference of the particle is obtained by accumulating the Euler distance between the contour points. Sphericity is a value between 0 and 1, where smaller values ​​result in more irregular particle shapes, as irregular shapes increase the perimeter.

[0142] Since the aerosol particle size and particle shape require a large amount of calculation during measurement, but the calculation repeatability is very high, this embodiment constructs a parallel calculation method for particle size and particle shape based on the collaboration of CPU and GPU, such as Figure 5As shown, GPU is better able to handle simple and highly repetitive tasks than CPU, while CPU has better results in calculating parameters with complex logic. Based on this situation, the present invention uses a computer parallel framework to open up threads on the CPU side, and completes parallel operations on the CPU side for parameters with complex calculation logic, including: minimum circumscribed rectangle, minimum circumscribed circle, convex hull and particle area. On the GPU side, parallel logic is designed based on NVIDIA's CUDA platform. When calculating the six parameters of horizontal distance, vertical distance, particle circumference, curvature radius, tangent slope and roundness, a separate thread is constructed for each contour point of all aerosol particles to achieve high parallelism. The calculation of each parameter occupies an independent CUDA stream and kernel function, and a synchronization mechanism is used to transmit data. When calculating the Fret diameter, a thread block is constructed for each particle, and a thread is constructed for each angle in each thread block to achieve parallel calculation, thereby synchronously outputting the Fret diameter of all particles at each angle.

[0143] In the design and development of optical systems, existing microscopic imaging technologies are mostly reflective imaging, with relatively few transmissive imaging technologies. Furthermore, there are very few studies using moving aerosol particles as detection targets. The present invention uses an infinity-corrected microscope objective lens. Compared to most common microscope objective lenses, its output light beam is parallel, which facilitates the subsequent addition of other optical elements to the rear end of the objective lens. The imaging cavity also takes into account depth of field limitations and gas sealing issues, adopting a four-window design to facilitate the subsequent introduction of multi-path imaging.

[0144] In terms of aerosol measurement methods, existing aerosol particle size and shape measurement technologies are mostly based on the principle of light scattering. However, the inversion accuracy of this technology is not high and the device is more complicated. The present invention is based on the principle of microscopic imaging. Unlike light scattering technology, it adopts the light obstruction method to obtain the contour information of aerosol particles and uses image processing to calculate the aerosol particle size and shape. However, the existing microscopic imaging technology is limited by the size of the field of view and the depth of field, and is also limited by the computational efficiency and accuracy in image processing. It usually outputs few parameters and has a slow calculation speed. Due to the existence of optical imaging limitations, there are few studies on determining the measurement target as aerosol particles. In addition, the existing dynamic image analysis technology is mostly used for the analysis of large particles. Its magnification is low, and there is no need to consider the depth of field and field of view. These technologies can also output some particle size and shape parameters. Compared with existing measurement methods, the present invention has made improvements in technical means for measuring aerosol particles of 10nm-100μm (including large pollen particle aerosols). While considering the depth of field and field size, it improves the resolution of image processing and outputs sufficiently detailed aerosol particle size and particle shape parameters. The entire process also realizes parallel computing.

[0145] In general, the present invention develops a system parameter calibration method and a method for aerosol particle extraction and measurement based on the principle of microscopic imaging, and proposes a combination of image denoising, edge detection, defocus detection, and duplicate particle removal, which largely avoids the problem of depth of field limitation, and adopts a video processing method to solve the limitation of a small field of view and a small number of particles to be acquired; in addition, the present invention also improves the traditional particle size and particle shape algorithms into parallel operations, fully utilizes computer resources, and realizes high-speed aerosol video processing technology under the collaboration of CPU and GPU, which can accurately, quickly, stably and long-term complete the measurement of aerosol particle size and particle shape.

[0146] Example 2

[0147] This embodiment provides a method for measuring aerosol particle size and particle shape, which is provided with the aerosol particle size and particle shape measurement system of the above embodiment 1. The method includes the following steps:

[0148] Obtain aerosol profile information based on transmission microscopy;

[0149] Collect aerosol particle images, analyze each frame of image data to generate image grayscale values, and plot the corresponding grayscale values ​​on each pixel of the image to obtain a complete aerosol particle image;

[0150] Calibrate system parameters, including background field calibration, pixel size calibration, depth of field feature grayscale value calibration, and depth of field size calibration;

[0151] Acquire particulate matter from aerosol particle images, including image denoising, aerosol labeling, defocus detection, and duplicate particle removal;

[0152] Calculate aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, maximum Fret diameter, and minimum Fret diameter.

[0153] Calculate aerosol particle shape. The output aerosol particle shape parameters include: particle convexity, particle circularity, particle roundness, and particle sphericity.

[0154] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. An aerosol particle size and shape measurement system, It is characterized in that include: Image acquisition module, aerosol microscopy imaging device, system parameter calibration module, particle acquisition module, particle size calculation module, particle shape calculation module; The aerosol microscopic imaging device is used to obtain the contour information of the aerosol based on the transmission microscopic imaging method; The image acquisition module acquires aerosol particle images through an aerosol microscopic imaging device, analyzes each frame of image data acquired to generate image grayscale values, and plots the corresponding grayscale values ​​on each pixel of the image to obtain a complete aerosol particle image; The system parameter calibration module calibrates the system parameters, including: background field calibration, pixel size calibration, depth of field characteristic gray value calibration and depth of field size calibration; The particle acquisition module is used to acquire particles of aerosol particle images, specifically including: image denoising, aerosol marking, defocus detection, and repeated particle removal; The particle size calculation module is used to calculate the aerosol particle size, and the output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, and Fret minimum diameter; The particle shape calculation module is used to calculate the aerosol particle shape, and the output aerosol particle shape parameters include: particle convexity, particle circularity, particle roundness, and particle sphericity.

2. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that The aerosol microscopic imaging device comprises: an aerosol generator, an aerosol dryer, an aerosol neutralizer, an imaging cavity, an LED light source, a microscope objective lens, a tube lens, an extension tube, and a CMOS camera; The imaging cavity is provided with an air inlet and an air outlet in the vertical direction, and flat glass is provided on both sides of the imaging cavity in the horizontal direction; The aerosol generator generates aerosol particles, which pass through an aerosol dryer and an aerosol neutralizer in sequence, and the output port of the aerosol neutralizer is connected to the air inlet of the imaging cavity; The LED light source is arranged on one side of the imaging cavity, the light beam is incident from the flat glass on one side and emitted from the flat glass on the opposite side, the light beam carrying aerosol particles is magnified by the microscope objective lens, and the light is converged onto the photosensitive surface of the CMOS camera through the tube lens and the extension tube.

3. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that The background field calibration specifically includes: Remove the background field from each frame of the image collected, leaving the foreground aerosol particles, and obtain the background value of each pixel in multiple background images. During calibration, collect multiple frames of empty background images without the target object, obtain the sub-minimum grayscale value at each pixel point in all background images as the background value at that point, and cyclically calculate the background value of each pixel point and output the background image; The pixel size calibration specifically includes: The pixel size is measured by using an optical graticule with a set line spacing, the position of the optical graticule is adjusted, and the optical graticule image is acquired; Binarization and morphological optimization denoising are performed on the optical graticule image; Marking the line pattern of the optical reticle using a line detection algorithm; The corner points of the optical graticule image are detected through the coordinates of the straight lines; The pixel size is calculated based on the distances between the two farthest corner points in the pixel coordinate system and the world coordinate system, and the pixel size is obtained after multiple calibrations; The depth of field characteristic grayscale value calibration specifically includes: The optical graticule is used as the target object, and the average gray value of the foreground area on the graticule image is calculated according to the weight on the image. The calculation formula is expressed as: Among them, I(x i ,y i ) represents the gray value of the i-th pixel on the straight line of the graticule in the foreground area of ​​the image. This formula assigns a smaller weight to the pixel with a smaller gray value; Move the optical graticule to find the maximum average grayscale value as the characteristic grayscale value of the depth of field, and take the average value after completing multiple calibrations; The depth of field calibration specifically includes: Move the optical graticule, calculate the line width and line length of the graticule in real time on the image, and record the farthest and nearest optical graticule positions within the set measurement error based on the clarity evaluation algorithm. The distance between the two is the depth of field. After completing multiple calibrations, take the average value.

4. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that The particle acquisition module is used to acquire particles in aerosol images, specifically including: image denoising, aerosol marking, defocus detection, and repeated particle removal; The image denoising specifically includes: Obtaining the aerosol particle foreground image after removing the background field image, performing bilateral filtering on the image, binarizing the filtered image, processing the binarized image with morphological closing operation, performing median filtering to remove the remaining noise points, and obtaining the aerosol particle image; The aerosol marking specifically includes: Perform edge detection on the denoised image to obtain the position of aerosol particles and the coordinates of the contour points of aerosol particles; The defocus detection specifically includes: The vertical distance and horizontal distance of the aerosol particles are calculated according to the contour coordinates of the aerosol particles, and a local image of the aerosol particles is constructed. The width of the local image is twice the horizontal distance, the height is twice the vertical distance, and the center point coordinate is the average value of the horizontal and vertical coordinates of the particle contour point; The maximum grayscale value in the local image area is obtained. If the value is less than the calibrated characteristic grayscale value of the depth of field, the aerosol particle is determined to be a defocused particle. The defocus of aerosol particles is calculated based on the defocus algorithm. The image variance value of the local image is calculated for the aerosol particles detected by the defocus, and the defocus particles with variance values ​​less than the set range are eliminated. The repeated particle removal specifically includes: Arrange and combine the aerosol particles, inputting two aerosol particles that have not been calculated each time; The frame difference range of repeated particles is set according to the gas flow rate and the camera frame rate, and it is determined whether the number of frames in which the two aerosol particles appear is less than the maximum frame difference. If the conditions are met, a minimum bounding rectangle is constructed for each of the two aerosol particles, and the size of the minimum bounding rectangle is used to determine whether the two aerosol particles are similar in size. If the aerosol particles are similar in size, the local images of the two aerosol particles are respectively input into the similarity evaluation algorithm to obtain the similarity index. The possibility of the two particles being repeated is determined based on the similarity index, and all permutations and combinations are calculated cyclically. For each group of repeated particles, only the first particle is retained and the remaining particles are discarded.

5. The aerosol particle size and shape measurement system according to claim 4, It is characterized in that The defocus algorithm evaluates the defocus of aerosol particles by combining the gray value difference in the local image of aerosol particles and the spatial distance between pixels. The calculation formula is as follows: Where i refers to the index of a pixel in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), and (x i ,y i ) represents the pixel coordinate value of pixel point i, (x max ,y max ) represents the pixel coordinate value of the pixel with the largest gray value in the local image, H and W are the height and width of the local image respectively, and n is the total number of pixels in the local image; The image variance value of the local image is calculated by the calculation formula: Where i refers to the index of a pixel in the local image of aerosol particles, I(x,y) is the grayscale value at point (x,y), and (x i ,y i ) represents the pixel coordinate value of pixel i, μ is the average gray value of the local image, and n is the total number of pixels in the local image.

6. The aerosol particle size and shape measurement system according to claim 4, It is characterized in that The similarity evaluation algorithm is specifically expressed as: Among them, p 1 and p 2 represents two aerosol particles, a(p 1 ,p 2 ) represents the angle between the line connecting two aerosol particles and the X-axis, (x 1 ,y 1 ) and (x 2 ,y 2 ) represent the center coordinates of the two aerosol particles.

7. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that The particle size calculation module is used to calculate the aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, Fret minimum diameter. The specific calculation method includes: The vertical distance and horizontal distance represent the maximum distance between particle contour points in the vertical and horizontal directions respectively; The equivalent circle diameter refers to the diameter of a circle with the same area as the particle area; The Fret diameter is determined by a set of two parallel tangent distances tangent to the particle. All tangents of the current angle are obtained at multiple angles. The largest tangent distance is taken as the Fret diameter at this angle by permutation and combination. The maximum Fret diameter and the minimum Fret diameter are selected.

8. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that The particle shape calculation module is used to calculate the aerosol particle shape. The output aerosol particle shape parameters include: particle convexity, particle roundness, particle roundness, and particle sphericity. The specific calculation methods include: By detecting whether the connecting line between the aerosol contour points is contained in the interior of the aerosol particle, its convex hull region is constructed and the convex hull area is calculated. The particle convexity is obtained by dividing the particle area by the convex hull area of ​​the particle. The particle circularity is calculated by the particle area and perimeter. By obtaining points with a curvature radius within a set range on the particle contour, the particle roundness is obtained by selecting the average curvature radius and dividing it by the minimum circumscribed circle radius of the particle; The sphericity of a particle is expressed as the ratio of the particle's equivalent circular circumference to its actual circumference.

9. The aerosol particle size and shape measurement system according to claim 1, It is characterized in that A parallel computing module is also provided, which is based on the CPU and GPU to coordinate and parallelly calculate the particle size and particle shape, specifically including: Parallel calculation parameters on the CPU side, including: minimum circumscribed rectangle, minimum circumscribed circle, convex hull and particle area; Construct parallel logic on the GPU side, and build a separate thread for each contour point of all aerosol particles when calculating horizontal distance, vertical distance, particle circumference, curvature radius, tangent slope, and roundness parameters. The calculation of each parameter occupies an independent CUDA stream and kernel function, and uses a synchronization mechanism to transmit data; When calculating the Fret diameter, a thread block is constructed for each particle, and a thread is constructed for each angle in each thread block to achieve parallel calculation, thereby synchronously outputting the Fret diameters of all particles at each angle.

10. A method for measuring aerosol particle size and particle shape, It is characterized in that The steps include: Obtain aerosol profile information based on transmission microscopy; Collect aerosol particle images, analyze each frame of image data collected to generate image grayscale values, and draw the corresponding grayscale values ​​on each pixel of the image to obtain a complete aerosol particle image; Calibrate system parameters, including background field calibration, pixel size calibration, depth of field characteristic grayscale value calibration and depth of field size calibration; Obtain particulate matter from aerosol particle images, including: image denoising, aerosol labeling, defocus detection, and duplicate particle removal; Calculate aerosol particle size. The output aerosol particle size parameters include: vertical distance, horizontal distance, equivalent circle diameter, Fret maximum diameter, and Fret minimum diameter; Calculate aerosol particle shape, and the output aerosol particle shape parameters include: particle convexity, particle roundness, particle roundness, and particle sphericity.

Images