Micro-droplet uniformity on-line detection and alarm method and application

Abstract

The invention discloses a droplet uniformity on-line detection and alarm method and application. The method comprises the following steps: acquiring an original image through an industrial camera, carrying out image type conversion on the original image to obtain a droplet grayscale image, setting different corrosion values according to different background pollution conditions to carry out morphological operation on a solder ball image, and carrying out online detection on the droplet uniformity according to a preset connectivity value. Eliminating a part smaller than a connected value to obtain a binary image with only the microdroplet, extracting contour coordinate information of the microdroplet by using a Canny algorithm, performing first correction on the contour coordinate information by using open operation, and then performing second coordinate information correction by using a least square ellipse fitting algorithm to obtain a binary image with the microdroplet; and finally, the particle size and the roundness of the microdroplet are calculated according to the contour coordinate information of the microdroplet, and when the software detects that the threshold values of the particle size and the roundness of the microdroplet are lower than a preset alarm threshold value, the alarm gives an alarm to remind an operator to adjust parameters of equipment for preparing the microdroplet in time, so that the final product quality and the production efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of droplet image processing, and in particular to a method and application of online detection and alarm of droplet uniformity. Background Art

[0002] With the increasing demand for microdroplet products in modern life and industry, microdroplet preparation technology has been widely used. Microdroplet preparation technology achieves precise dispensing of microfluidics by generating droplets of uniform size. This technology offers the advantages of a simple process flow and high controllable precision. Droplet sizes range from nanometers to micrometers and are primarily used in a variety of key fields, including electronic packaging, 3D printing of micro-components, inkjet printing, microfluidic chip technology, and biochemical testing. In these fields, the quality of droplet preparation technology directly impacts its application. Real-time monitoring of the droplet preparation process is crucial. However, ensuring droplet uniformity is challenging and requires high-precision detection technology and effective control measures. For example, in microfluidic chips, subtle variations in channel dimensions, fluctuations in fluid flow rate, and uneven surface tension can lead to variations in droplet size and shape. In inkjet printing, problems such as nozzle clogging and unstable droplet ejection velocity can also affect droplet uniformity.

[0003] To detect droplet uniformity, traditional methods often rely on offline sampling. This approach is not only inefficient but also fails to reflect real-time changes in uniformity during droplet generation, making it difficult to meet the real-time and accuracy requirements of modern industrial production and scientific research. Furthermore, offline detection can damage droplets, affecting their subsequent use. Machine vision systems, on the other hand, can rapidly acquire droplet images in a non-contact manner. Using image processing algorithms, they analyze and process these images, enabling accurate measurement of parameters such as droplet size, shape, and distribution.

[0004] Existing machine vision-based droplet detection technology still has some shortcomings. First, in terms of image acquisition, due to the rapid generation speed of droplets and the small spacing between droplets, high requirements are placed on the frame rate and resolution of the image acquisition equipment. At the same time, the stability and uniformity of lighting conditions will also have a significant impact on image quality. If the lighting is uneven or there are shadows, it may lead to inaccurate feature extraction of droplets. Secondly, in terms of image processing algorithms, there may be large differences in the shape and size of droplets, which brings difficulties to image segmentation, feature extraction and parameter calculation. Existing image processing algorithms often have problems with insufficient accuracy and robustness when dealing with droplets with complex shapes and size distributions. The computational efficiency of the algorithm is also a key factor. For real-time online detection, it is necessary to increase the running speed of the algorithm as much as possible while ensuring accuracy. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and application for online detection and alarm of droplet uniformity, and to construct a detection device based on this technology, which can timely and accurately monitor the uniformity of droplets during the preparation process. By combining machine vision with image processing, the acquisition and processing of droplet images are realized, and finally an alarm instruction is sent through serial communication, thereby realizing real-time monitoring and alarm of droplet uniformity, greatly improving the efficiency of droplet preparation.

[0006] The present invention is implemented as follows: a method for online detection and alarm of droplet uniformity, comprising the following steps:

[0007] (1) Capturing images of falling droplets using an industrial camera;

[0008] (2) performing image type conversion on the droplet image and performing image preprocessing on the initial image;

[0009] (3) Feature detection is performed based on the pre-processed droplet image. The coordinate information of the droplet contour is first extracted using the Canny algorithm. The measurement error is then repaired using the least squares ellipse fitting operator. Finally, the roundness and particle size are calculated based on the repaired contour information.

[0010] (4) Threshold analysis is performed on the obtained roundness and particle size results. If the results are lower than the pre-set threshold, an alarm is issued.

[0011] Furthermore, the capturing of the falling droplet image in step (1) is capturing a color image, extracting the red channel image component of the color image, and obtaining the initial droplet image based on the red channel image component.

[0012] Furthermore, in step (2), the initial image is preprocessed, including the following steps:

[0013] (2-1) If the initial image contains salt and pepper noise, median filtering is performed on the initial image to obtain an initial grayscale image without noise; if the initial image contains Gaussian noise, Gaussian filtering is performed on the initial image to obtain an initial grayscale image without noise;

[0014] (2-2) Binarizing the filtered initial grayscale image without noise, and performing threshold binarization using the OTSU algorithm to obtain a droplet binary image;

[0015] (2-3) If the binary image is a stain-containing image, a demorphological operation is performed on the stain image. According to the preset connectivity value, the part less than the connectivity value is eliminated to obtain a binary image with only droplets. If the binary image is a stain-free image, step (3) is directly performed.

[0016] Furthermore, in step (3), the operator for extracting the edge of the droplet is the Canny operator, thereby obtaining the contour coordinate information of the droplet; a first correction is performed based on the extracted droplet contour coordinate information, and an opening operation is used to perform a morphological operation on the droplet, wherein the eroded structure is a circular structure operator; based on the result of the first contour coordinate information correction, a second contour coordinate information correction is performed using the least squares ellipse fitting algorithm; based on the droplet contour coordinate information, the roundness and particle size of the droplet are calculated.

[0017] A computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the above method.

[0018] A computer device comprises a memory, a processor and a program stored and executable on the memory, wherein the program implements the steps of the above method when executed by the processor.

[0019] The above-mentioned method of online detection and alarm of droplet uniformity is applied in the preparation of solder balls under gas control mode.

[0020] Compared with existing detection technologies, the present invention has the following beneficial effects: it greatly improves detection efficiency and saves a lot of manpower and time compared to manual detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The present invention is a flow chart of the detection method.

[0022] Figure 2 Schematic diagram of the structure of a uniform droplet preparation device in gas control mode using the method of the present invention.

[0023] Figure 3 The figure is a schematic diagram of the software interface developed using the method of the present invention.

[0024] Figure 4 This is a comparison diagram of the original droplet salt and pepper noise image and the median filtered image in the method of the present invention.

[0025] Figure 5 This is a comparison diagram of the original droplet image containing Gaussian noise and the image after Gaussian filtering in the method of the present invention.

[0026] Figure 6 This is a binarized droplet image using the method of the present invention.

[0027] Figure 7 This is an image of a droplet after decontamination in the method of the present invention.

[0028] Figure 8It is the droplet edge image after Canny edge detection in the method of the present invention.

[0029] Figure 9 This is the droplet edge image after least squares ellipse fitting in the method of the present invention.

[0030] Figure 10 This is a schematic diagram of the final result of software detection in the method of the present invention. DETAILED DESCRIPTION

[0031] 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. 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.

[0032] In the description of the present invention, it should be noted that the terms "upper", "middle", "lower", "inside", "outside", "both sides", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings. They are only simplified descriptions for the convenience of describing the present invention, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on the present invention.

[0033] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "provided with," "connected," etc., should be understood in a broad sense. For example, "connected" may refer to a fixed connection, a detachable connection, or an integral connection; it may refer to a mechanical connection or an electrical connection; it may refer to a direct connection or an indirect connection through an intermediate medium; it may refer to internal communication between two components. Those skilled in the art will be able to understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0034] like Figure 1 As shown, the method for online detection and alarm of droplet uniformity of the present invention includes the following steps:

[0035] (1) Capturing images of falling droplets using an industrial camera;

[0036] (2) performing image type conversion on the droplet image and performing image preprocessing on the initial image;

[0037] (3) Feature detection is performed based on the pre-processed droplet image. The coordinate information of the droplet contour is first extracted using the Canny algorithm. The measurement error is then repaired using the least squares ellipse fitting operator. Finally, the roundness and particle size are calculated based on the repaired contour information.

[0038] (4) Threshold analysis is performed on the obtained roundness and particle size results. If the results are lower than the pre-set threshold, an alarm is issued.

[0039] Furthermore, the capturing of the falling droplet image in step (1) is capturing a color image, extracting the red channel image component of the color image, and obtaining the initial droplet image based on the red channel image component.

[0040] Furthermore, in step (2), the initial image is preprocessed, including the following steps:

[0041] (2-1) If the initial image contains salt and pepper noise, median filtering is performed on the initial image to obtain an initial grayscale image without noise; if the initial image contains Gaussian noise, Gaussian filtering is performed on the initial image to obtain an initial grayscale image without noise;

[0042] (2-2) Binarizing the filtered initial grayscale image without noise, and performing threshold binarization using the OTSU algorithm to obtain a droplet binary image;

[0043] (2-3) If the binary image is a stain-containing image, a demorphological operation is performed on the stain image. According to the preset connectivity value, the part less than the connectivity value is eliminated to obtain a binary image with only droplets. If the binary image is a stain-free image, step (3) is directly performed.

[0044] Furthermore, in step (3), the operator for extracting the edge of the droplet is the Canny operator, thereby obtaining the contour coordinate information of the droplet; a first correction is performed based on the extracted droplet contour coordinate information, and an opening operation is used to perform a morphological operation on the droplet, wherein the eroded structure is a circular structure operator; based on the result of the first contour coordinate information correction, a second contour coordinate information correction is performed using the least squares ellipse fitting algorithm; based on the droplet contour coordinate information, the roundness and particle size of the droplet are calculated.

[0045] The present invention

[0046] A computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the above method.

[0047] A computer device comprises a memory, a processor and a program stored and executable on the memory, wherein the program implements the steps of the above method when executed by the processor.

[0048] The above-mentioned method of online detection and alarm of droplet uniformity is applied in the preparation of solder balls under gas control mode.

[0049] The software interface diagram developed by the method of the present invention is as follows Figure 3 shown.

[0050] The method of the present invention is described below by taking the solder ball preparation process in the gas control mode as an example. Figure 2 As shown, the solder ball welding device includes a disturbance generator, an air control system, an elastic diaphragm, a liquid storage chamber, a spray chamber, a nozzle, a detection system and an alarm device. The connection assembly of the device is as follows: the liquid storage chamber 1 is at the top of the solder ball preparation device, and the top of the liquid storage chamber is connected to a vacuum tube 12 for connecting to the air control system. The air control system consists of a pressure valve and a nitrogen tank; a pressure regulating valve 11 is connected to the left side of the vacuum tube 12, and a nitrogen tank 10 is connected below the pressure valve 11; the constant air pressure generated by the air control system acts on the surface of the distilled water in the liquid storage chamber 1, forcing the distilled water to fill the spray chamber 4; a disturbance chamber 2 is provided below the liquid storage chamber 1, and the disturbance chamber 2 is filled with a gallium-indium Ga-In alloy solution. An elastic diaphragm 3 is provided at the bottom of the disturbance chamber 2. The function of the elastic diaphragm 3 is to transmit the pulse electromagnetic force 9 to the jet liquid column when a pulse electromagnetic force 9 is generated in the disturbance chamber 2. At the same time, the elastic diaphragm 3 also has the function of isolating the liquid metal solution and the solution to be sprayed; a spray chamber 4 is provided below the elastic diaphragm 3. 4 is filled with a distilled water solution to be sprayed; a nozzle 5 is provided below the center of the spray chamber 4, and the aperture sizes of the nozzle 5 are 0.2 mm, 0.3 mm, 0.4 mm, 0.5 mm, 0.6 mm and above. According to the parameters of gas pressure and flow rate, a suitable nozzle 5 is selected, and generally a nozzle 5 with an aperture of 0.4 mm is selected; an image acquisition device is provided below the nozzle 5, and the image acquisition device consists of a coaxial light source 13 and an industrial camera 14, wherein the coaxial light source 13 model is HL-COPL30P from Haoli Automation, the lens model of the industrial camera is MML4-HR65D-VI from Daheng Image, and the camera model is ME2P-900-13GC-P; a computer 15 is connected to the right side of the industrial camera 14, and the computer 15 performs image processing. After processing by the computer 15, if it is found that the droplets are uneven, a signal will be transmitted to the alarm 16, and the alarm 16 will issue a warning at this time to remind the staff to adjust the preparation parameters and gas pressure in time. Below the liquid storage chamber 1 lies the core of the disturbance generator, which consists of a disturbance chamber 2, a liquid metal Ga-In alloy, an electrode 6, a permanent magnet 7, a signal source 17, and a power amplifier 19. A uniform magnetic field is generated between two parallel permanent magnets 7. The periodic pulsed electrical signal 18 generated by the signal source 17 is amplified by the power amplifier 19 to produce a frequency-adjustable pulsed current 20 with a current amplitude of 15A. When this pulsed electrical signal flows through the liquid metal, the variable electrical signal interacts with the liquid metal in a constant magnetic field to generate a pulsed electromagnetic force. This pulsed electromagnetic force, as a disturbance wave, is transmitted through the elastic diaphragm to the jet liquid column, thereby controlling the breakage of the droplets.

[0051] The method for online detection and alarm of droplet uniformity is as follows:

[0052] Step 1: The original image data source used by the method of the present invention is collected by a high-resolution industrial microscopy imaging device, and after digital conversion, a grayscale image with a spatial resolution of 2480×2048 pixels is generated, and its quantization accuracy meets the 8-bit deep grayscale matrix standard. In the digital image processing architecture, the filtering algorithm is a key preprocessing link, and has complex functions such as multimodal noise suppression, edge feature preservation, and image quality enhancement. According to its mathematical principles, it can be divided into two categories: time domain operations and frequency domain analysis: the former realizes spatial domain processing by directly operating the pixel neighborhood, and the latter performs frequency component regulation based on Fourier transform; from the algorithm construction dimension, it can be divided into two paradigms: linear system (such as Gaussian convolution) and nonlinear operator (such as median filtering).

[0053] The main goal of image filtering technology is to achieve the following: noise elimination mechanism, statistical modeling and numerical correction of random noise (including but not limited to Gaussian distribution noise, Poisson noise and impulse noise) introduced by the imaging system during the photoelectric conversion and signal transmission process.

[0054] If the original image contains salt and pepper noise, the original image with salt and pepper noise is shown in Figure 4 In the left image, the selected filtering method is median filtering. As a typical nonlinear digital filtering method, the core mechanism of this technology is to extract the numerical median of the neighborhood pixels as the output value. Compared with linear filtering, this processing strategy can effectively eliminate discrete pulse interference while better maintaining contour clarity and significantly reducing the image blur caused by traditional mean filtering. The formula g(x,y)=median{f(x±i,y±j)}, (i,j)∈S represents median filtering. f(x,y)——the pixel value of the original grayscale image; g(x,y)——the grayscale value of the filtered image pixel; S——the filter window.

[0055] In this image filtering, a 5×5 pixel rectangular processing unit is selected as the basic operation window. This structural unit traverses the entire digital image in a pixel-by-pixel sliding manner. For the pixel to be processed at the coordinate (x, y), the algorithm automatically constructs a 25-dimensional grayscale data set with the point as the center and in the spatial domain i, j∈[-2,2]. After ordering, the statistical median with a rank of 13 is taken as the output value. This operation process is implemented by calling the medfilt2 function of the MATLAB numerical computing platform. Its parameterized configuration includes the kernel matrix dimension declaration and index mode optimization settings. The processing results are shown in the attached figure. Figure 4 As shown in the right figure, the discretely distributed salt and pepper noise components are effectively filtered out, while key features such as the droplet geometric outline remain intact.

[0056] If the original image contains Gaussian noise, the original image with Gaussian noise is Figure 5In the left image, the selected filtering method is Gaussian filtering. Gaussian filtering is a linear filtering method that smooths the image by performing a weighted average calculation using a convolution kernel generated by a Gaussian function. Gaussian filtering effectively suppresses random noise that follows a normal distribution, retaining more low-frequency information while smoothing the image. This method offers advantages in balancing denoising and preserving the overall image characteristics. The formula for Gaussian filtering is: f(x,y)—the pixel value of the original grayscale image; g(x,y)—the grayscale value of the filtered image pixel; G—the Gaussian convolution kernel.

[0057] In this image filtering, a 5×5 filter window is selected and a Gaussian kernel G with a standard deviation of σ is defined. The convolution operation is performed on the image. The weighted average of each pixel in the filter window is calculated (the weight is determined by the Gaussian function) and the result is used as the new value of the center pixel of the window. This process is repeated for each pixel in the image until the entire image is processed. In MATLAB, the "imgaussfilt" function in the toolbox is called to complete the Gaussian filtering function. The image after Gaussian filtering is shown in Figure 5 As shown in the picture on the right.

[0058] Step 2: A key step in binarization is threshold segmentation. The threshold is the basis for converting a grayscale image into a binary image. By selecting an appropriate threshold, each grayscale value in the image can be divided into 0 or 255, that is, the effect of a black and white image, achieving the separation of the target and the background. In this detection method, the OTSU method, that is, the maximum inter-class variance method, is used for threshold segmentation. That is, a segmentation threshold τ should be found to maximize the variance of the two classes A and B. This can be seen as a problem of finding the maximum value. Use τ to traverse all integers from 0 to 255, find all corresponding variances, and take the τ corresponding to the maximum variance. The specific steps are as follows:

[0059] When the threshold is k, the probability of the pixel being classified into A is p A (k), the average grayscale of the pixels assigned to A is m A (k). Similarly, the probability that a pixel is classified into B is p B (k), the average grayscale of the pixels assigned to A is m B (k). There is also the cumulative mean m of gray level k, the gray value m of the entire image G . Then we have:

[0060] p A (k)×m A (k)+p B (k)×m B (k) = m G (1)

[0062] p A (k)+pB (k)=1

[0063] (2) According to the concept of variance, the expression of variance is written as:

[0064] σ 2 =p A (k)(m A (k)-m G ) 2 +p B (k)(m B (k)-m G ) 2

[0065] (3) Substitute (1) into (3):

[0066] σ 2 =p A (k)p B (k)(m A (k)-m B (k)) 2

[0067] (4) Of which:

[0068]

[0069]

[0070] For the convenience of calculation, we can write it as:

[0071]

[0072] Next, we only need to find the gray level k that maximizes (5), which is the OTSU threshold τ.

[0073] Step 3: Image background decontamination is to remove small areas in the image background. This step can effectively remove unnecessary elements. Use the connected domain segmentation operator to remove area objects with small preset connectivity values from the binary image, and then obtain a binary image with only droplets. The effect is shown in the figure. Figure 7 shown.

[0074] Step 4: Use Canny edge segmentation to extract the edge of the droplet and obtain the contour coordinate information of the droplet. The specific steps of the Canny edge detection algorithm are as follows:

[0075] The Canny edge detection operator first uses a Gaussian filter to smooth the image, and then uses the first-order partial derivative operator to find the image grayscale along the horizontal direction G x and vertical direction G yThe partial derivative of is obtained and the magnitude and orientation of the gradient are calculated using the following formulas, as shown in formulas (6) and (7).

[0076]

[0077] Here, G represents the gradient magnitude, and θ represents the azimuth angle. Non-maximum suppression is then performed on the gradient magnitude, essentially finding the local gradient maximum. The method involves comparing a given pixel P with two pixels along the gradient line within a 3x3 neighborhood window. If the gradient magnitude of P is not greater than the gradient magnitudes of these two pixels, then P is set to 0; otherwise, the original magnitude is retained. Here, we classify the gradient directions into four directions: horizontal, vertical, 45°, and -45°, to compare the strength of the gradient magnitudes. A dual-threshold algorithm is used to detect and link edges. The method states that any value greater than a high threshold T1 is an edge, while any value less than a low threshold T2 is not an edge. If the detection result is both greater than the low threshold and less than the high threshold, the pixel's neighboring pixels are checked to see if there are edge pixels greater than the high threshold. If so, it is an edge; otherwise, it is not.

[0078] The final output is the refined edge image, see Figure 8 As shown, the coordinate information of the identified edge is output at the same time.

[0079]

[0080] Step 5: Use the least squares method to fit the detected droplet edge and perform secondary correction on the coordinate information. The specific steps are as follows:

[0081] First, determine the elliptic curve algebraic equation. Formula (8) is expressed as the elliptic curve algebraic equation.

[0082] F(x,y)=ax 2 +bxy+cy 2 +dx+ey+f=0 (8)

[0083] Where a, b, c, d, e, and f are the coefficients in the ellipse equation. During the fitting process, the contour points after the droplet edge detection are processed by least squares to minimize the sum of the algebraic distances from each point to the equation, thereby obtaining the coefficients in the equation as shown in Equation (9). i ,y i ) to the elliptic equation can be expressed as F(x i ,y i ).

[0084]

[0085] i——sequence of contour points; n——number of contour points.

[0086] The final fitting effect is shown in Figure 9 As shown, the coordinate information before and after fitting is compared.

[0087]

[0088] The final result diagram of the software detection in the method of the present invention is as follows Figure 10 shown.

[0089] Compared with other detection technologies, the present invention is based on machine vision-based droplet uniformity detection:

[0090] 1. Advantages

[0091] Non-contact detection: Machine vision detection does not require direct contact with the droplets, avoiding physical interference and contamination of the droplets.

[0092] High precision and resolution: Capable of capturing subtle features and differences in droplets, providing more accurate uniformity assessments. For example, in the detection of tiny droplets, differences of a few microns or even smaller can be clearly distinguished, while other technologies may not be able to achieve such high precision.

[0093] Rapid detection: Ability to detect a large number of droplets in a short period of time, improving detection efficiency; for the droplet generation process on industrial production lines, it can quickly monitor quality in real time, detect problems in time and adjust production parameters.

[0094] Good repeatability: Based on the set algorithms and parameters, the test results are highly consistent and repeatable; regardless of the number of tests, similar test results can be stably provided, reducing human errors.

[0095] Multi-parameter detection: Simultaneously acquire multiple characteristic parameters of droplets, such as size and shape, to comprehensively evaluate uniformity.

[0096] 2. Improvement

[0097] Significant improvement in detection accuracy: Compared with traditional detection methods based on physical or chemical principles, machine vision can more accurately identify the morphology and characteristic changes of droplets.

[0098] Expanded detection range: Various types and materials of droplets can be detected, including transparent, translucent, and opaque droplets. Whether it is liquid drug droplets or industrial special material droplets, effective uniformity detection can be performed.

[0099] Real-time monitoring and feedback: The droplet generation process is monitored in real time, and test results are promptly fed back. Process parameters can be adjusted instantly based on real-time test data to ensure droplet uniformity remains within the ideal range.

[0100] Lower testing costs: Although the initial equipment investment may be high, in the long run, its efficiency, accuracy, and repeatability can reduce the costs and resource waste caused by defective products. For example, in large-scale droplet production, reducing the scrap rate can significantly save production costs.

[0101] The present invention can not only produce uniform droplets below 0.5 mm, but also use machine vision to monitor the uniformity of the droplets during the preparation process. When non-uniform droplets are detected, an alarm can be issued to remind the staff to adjust the preparation parameters in time.

[0102] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (e.g., infrared, wireless, microwave, etc.)) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape) or an optical medium.

[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for online detection and alarm of droplet uniformity, characterized in that: The following steps are involved: (1) Capturing images of falling droplets using an industrial camera; (2) performing image type conversion on the droplet image and performing image preprocessing on the initial image; (3) Feature detection is performed based on the pre-processed droplet image. The coordinate information of the droplet contour is first extracted using the Canny algorithm. The measurement error is then repaired using the least squares ellipse fitting operator. Finally, the roundness and particle size are calculated based on the repaired contour information. (4) Threshold analysis is performed on the obtained roundness and particle size results. If the results are lower than the pre-set threshold, an alarm is issued.

2. The method for online detection and alarm of droplet uniformity according to claim 1, characterized in that: The step (1) of capturing the falling droplet image is to capture a color image, extract the red channel image component of the color image, and obtain the initial droplet image based on the red channel image component.

3. The method for online detection and alarm of droplet uniformity according to claim 1, characterized in that: In step (2), the initial image is preprocessed, including the following steps: (2-1) If the initial image contains salt and pepper noise, median filtering is performed on the initial image to obtain an initial grayscale image without noise; if the initial image contains Gaussian noise, Gaussian filtering is performed on the initial image to obtain an initial grayscale image without noise; (2-2) Binarizing the filtered initial grayscale image without noise, and performing threshold binarization using the OTSU algorithm to obtain a droplet binary image; (2-3) If the binary image is a stain-containing image, a demorphological operation is performed on the stain image. According to the preset connectivity value, the part less than the connectivity value is eliminated to obtain a binary image with only droplets. If the binary image is a stain-free image, step (3) is directly performed.

4. The method for online detection and alarm of droplet uniformity according to claim 1 or 3, characterized in that: In step (3), the operator for extracting the edge of the droplet is the Canny operator, thereby obtaining the contour coordinate information of the droplet; the first correction is performed based on the extracted droplet contour coordinate information, and the droplet is morphologically operated using the opening operation, wherein the eroded structure is a circular structure operator; based on the result of the first contour coordinate information correction, the least squares ellipse fitting algorithm is used to perform the second contour coordinate information correction; based on the droplet contour coordinate information, the roundness and particle size of the droplet are calculated.

5. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 4.

6. A computer device, characterized in that: The computer device includes a memory, a processor, and a program stored and executable on the memory, and when the program is executed by the processor, the steps of the method according to any one of claims 1 to 4 are implemented.

7. Application of the method for online detection and alarm of droplet uniformity according to any one of claims 1 to 4 in solder ball preparation under gas control mode.

Images