A method for processing images of cell droplets

Abstract

This invention provides a method for processing cell droplet images. The method includes the following steps: S1: Image sharpness enhancement processing; S2: Droplet edge processing, including the following steps: Step 21: Dividing the enhanced image into multiple sub-images; Step 22: Processing the contours in the sub-images, including background inversion, opening, and closing operations on the edge lines to obtain smooth contours. The processed contours from the multiple sub-images are then merged into a complete contour to obtain a complete droplet contour. This invention first effectively suppresses imaging interference such as optical path noise, uneven illumination, and background noise through image sharpness enhancement and image block preprocessing, improving the contrast between the droplet and the background and avoiding problems such as detail loss and edge truncation in traditional global image processing. Through image inversion, fixed threshold segmentation, and opening / closing morphological operations, the droplet edges are accurately extracted and smoothed.

Description

Technical Field

[0001] This invention relates to cell droplet image processing technology, specifically to a cell droplet image processing method for obtaining a clear outline image and performing centering correction on a droplet image. Background Technology

[0002] Single-cell analysis is a technology in the life sciences that can resolve cellular heterogeneity and microscopic life mechanisms at single-cell resolution. Currently, single-cell genomics, epigenomics, transcriptomics, and proteomics technologies are well-developed and widely used in basic life research, disease diagnosis, and drug development.

[0003] Compared to the aforementioned omics technologies, the development of single-cell metabolomics has lagged behind. Metabolomics is the terminal expression of multi-omics, directly reflecting the real-time physiological state of cells and serving as a core basis for elucidating cellular life activities. However, limited by the diverse types of metabolites, their extremely low concentrations, high dynamism, and poor stability, coupled with the cumbersome operation, low throughput, and significant sample loss of traditional detection methods, the development of single-cell metabolomics has been slow, becoming a key bottleneck restricting the improvement of single-cell multi-omics systems.

[0004] Existing single-cell metabolomics detection platforms mainly include three categories: microscopy, spectroscopy, and mass spectrometry. Among them, surface-enhanced Raman spectroscopy (SERS) technology, with its advantages of not requiring complex sample pretreatment, high sensitivity and specificity, no sample damage, low detection cost, and in-situ real-time detection, is highly suitable for the detection needs of trace metabolites in single cells and is currently the mainstream core detection technology in single-cell metabolomics.

[0005] However, existing SERS single-cell metabolic detection systems suffer from core defects such as low detection throughput and insufficient automation, limiting the large-scale application of the technology. Microfluidic droplets are the core carrier for single-cell SERS detection. Accurate segmentation, contour extraction, and recognition of droplet images are crucial preliminary steps for single-cell localization, spectral matching, and high-throughput data analysis. The accuracy and efficiency of image processing directly determine the accuracy and automation level of the overall detection system.

[0006] Current conventional cell droplet image processing techniques are ill-suited for high-throughput automated SERS detection, exhibiting several technical shortcomings. Firstly, traditional algorithms employ a uniform threshold segmentation and fixed contour extraction mode, failing to differentiate between droplet imaging quality variations. Microfluidic droplet imaging is susceptible to optical path noise, uneven illumination, and impurity interference, resulting in a mixture of clear and blurry droplets. Traditional methods struggle to maintain extraction accuracy for both types of droplets simultaneously, leading to issues such as incomplete contours, target loss, and noise-induced misidentification.

[0007] Current technologies lack specific extraction solutions for low-resolution droplets. Traditional brightness threshold segmentation methods cannot effectively identify dark, blurry droplets under low light and low contrast conditions, making it difficult to distinguish droplets from background areas, which can easily lead to missed samples and significantly reduce detection throughput.

[0008] In summary, there is an urgent need for a cell droplet image processing method that can adapt to droplets of different resolutions, intelligently optimize contours, balance accuracy and efficiency, and support fully automated batch processing. This method would provide accurate image support for high-throughput automated single-cell SERS metabolomics detection, and promote the development and large-scale application of single-cell metabolomics technology. Summary of the Invention

[0009] To address the problems existing in the above-mentioned technologies, the present invention provides a method for processing cell droplet images, which includes the following steps: S1: Image sharpness enhancement processing; S2: Droplet edge processing, including the following steps: Step 21: Divide the enhanced image into multiple sub-images; Step 22: Process the contours in the sub-images, including background inversion of the edge lines, opening operation and closing operation processing; to obtain smooth contours, merge the processed contours in multiple sub-images into a complete contour to obtain a complete droplet contour.

[0010] Preferably, in step 22: the roundness of the merged complete droplet is calculated, the clarity of the droplet is determined, and droplets above a set threshold are defined as clear droplets, and droplets below the set threshold are defined as unclear droplets.

[0011] Preferably, the following steps are performed for images of clear droplets; Step S221: Detect the black edges of the droplets in the sub-image, invert the image to convert the black edges to white, and keep the background black; Step S222: Set a fixed high threshold Preserve the brightest part of the inverted image. Threshold segmentation is achieved in the specified region. Step S223: Perform morphological operations on the segmented image: Morphological operations include closing and opening operations; Step S224: Intelligent merging of small outlines: Iterate through all detected small contours and merge only those contours that meet the following conditions: the distance is less than a set pixel distance; the area of ​​the small contour is greater than a set area pixel. Overall droplet sphericity after merging ; Indicates the roundness of the merged shape. This indicates the initial roundness; after merging, the complete droplet profile is obtained. ; The steps for extracting unclear droplet contours are as follows: Step S31: Extract using HSV color space, adaptively calculate color threshold, and obtain the dark part inside the droplet; Step S32: Select the initial outline with the largest area from the dark region, and use the same intelligent merging strategy as the clear droplet, i.e., perform step S224 above, to improve the roundness of the outline and obtain a complete droplet outline. .

[0012] Preferably, the set threshold is 0.95, when Determined to be a clear droplet, The droplet was determined to be unclear. In step S221: the white pixel value is 255, and the black background pixel value is 0; Step S222: Fix the high threshold , which represents the pixel values ​​of the inverted image; The set pixel distance is 5, which satisfies the condition. The small outline area is set to 50 pixels, which satisfies the condition. Pixel.

[0013] Preferably, step S223 includes the following steps: The closing operation involves the following steps: using a 3×3 rectangular structural element to fill the outline gap, using the following formula: ,in For expansion operation, For corrosion operation, For structural elements; The opening operation involves: removing minute noise points and smoothing the edges, using the following formula: .

[0014] Preferably, step S1 includes the following steps: S11: Read the microscopic droplet image and convert the image into RGB three-channel data format; S12: Apply the CLAHE algorithm to each channel to enhance local contrast. The calculation formula is as follows: , in, For a single channel of an RGB image (c=R, G, B). For adaptive histogram equalization operators, The contrast clipping threshold. This refers to the size of the grid.

[0015] Preferably, in step 21, the enhanced image is uniformly divided into 8 sub-images, i.e., a 2×4 matrix distribution, and contour detection is performed on each sub-image to reduce noise interference in the global processing.

[0016] Preferably, the roundness is determined using the following standard calculation formula: , in, The outline area is based on the number of pixels. The perimeter of the outline is based on the number of pixels. , It is a perfect circle.

[0017] Preferably, the method further includes a step of identifying the droplet center and the image center: Step 1: Calculate the center coordinates of the complete droplet profile The centroid formula is used to calculate the mean of the coordinates of all pixels in the contour. Step 2: Calculate the coordinates of the image center ; Step 3: Calculate the offset between the droplet center and the image center. : , .

[0018] Preferably, the method further includes step 4: calculating the motion path of the two-dimensional motion platform. The pixel-to-physical distance conversion factor of the 2D motion platform is Then the physical distance that the two-dimensional motion platform needs to move is: , Two-dimensional motion platform according to The droplet moves along a linear path to align its center with the image center, thus achieving automatic centering.

[0019] This invention addresses the shortcomings of existing cell droplet image processing technologies, such as low contour extraction accuracy, poor ability to identify blurred droplets, severe contour fragmentation, and high reliance on manual intervention, making them unsuitable for high-throughput automated detection in single-cell SERS metabolomics. It designs a differentiated droplet recognition and intelligent contour optimization processing strategy, combining image enhancement and block processing, morphological operations, and a multi-condition contour merging mechanism, resulting in the following significant technical advantages: 1. This invention first enhances image sharpness and performs image segmentation preprocessing to effectively suppress imaging interference such as optical path noise, uneven illumination, and background noise, thereby improving the contrast between droplets and the background and avoiding problems such as detail loss and edge truncation in traditional global image processing. Simultaneously, it innovatively introduces a circularity discrimination mechanism to distinguish between clear and unclear droplets and employs differentiated processing schemes: for clear droplets, image inversion, fixed threshold segmentation, and opening / closing morphological operations are used to accurately extract and smooth droplet edges; for low-contrast, blurry droplets that cannot be adapted to traditional brightness thresholds, an HSV color space adaptive threshold extraction method is used to accurately locate the effective droplet area, completely solving the problems of poor adaptability to complex imaging scenes, missed detection of blurry droplets, and incomplete contour extraction in traditional algorithms, achieving full coverage and high-precision recognition of various droplet samples.

[0020] 2. Through multi-constraint intelligent contour merging, the fragmentation problem of the contour is effectively solved, ensuring the regularity of the droplet contour: This invention abandons the crude processing method of blindly merging or directly eliminating fragmented contours in traditional algorithms, and establishes an intelligent merging mechanism with triple constraints of pixel distance, minimum contour area, and post-merging circularity optimization. Only fragmented small contours that meet the conditions are selectively merged. This effectively repairs incomplete droplet contours caused by imaging defects, maximizing the retention of effective sample information, and avoids contour distortion and morphological distortion caused by merging invalid noise points. This significantly improves the integrity and circularity of the final droplet contour, providing a reliable image foundation for subsequent precise single-cell localization and region matching, reducing image processing errors from the source. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 These are clear images of the droplets.

[0023] Figure 2 The image shows a clear droplet outline obtained by processing the method provided in this invention.

[0024] Figure 3 It is an image of an unclear droplet.

[0025] Figure 4 The image is an image of an unclear droplet outline obtained by processing the method provided in this invention. Detailed Implementation

[0026] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1:

[0028] The cell droplet image processing method of this embodiment includes the following steps: S1: Image sharpness enhancement processing; S2: Droplet edge processing, including the following steps: Step 21: Divide the enhanced image into multiple sub-images; Step 22: Process the contours in the sub-images, including background inversion of the edge lines, opening operation and closing operation processing; to obtain smooth contours, merge the processed contours in multiple sub-images into a complete contour to obtain a complete droplet contour.

[0029] This method first divides the entire enhanced image into multiple small sub-images, and performs background inversion and opening / closing morphological processing on each sub-image separately: background inversion can highlight the faint black edges of the droplets, making the edge features more noticeable; The opening operation removes small noise and isolated specks, the closing operation fills in tiny gaps and gaps in the outline, and finally all the processed sub-image outlines are reassembled to obtain the overall droplet outline.

[0030] Example 2:

[0031] In step 22: the roundness of the merged complete droplet is calculated, and the clarity of the droplet is determined. Droplets above a set threshold are defined as clear droplets, and those below the set threshold are defined as unclear droplets.

[0032] The edges of clear droplets exhibit strong black-and-white contrast and distinct edge lines, primarily identified by differences in brightness / darkness. Therefore, this method is suitable for image inversion and high-brightness-threshold segmentation, and when combined with conventional morphological processing, it offers high speed and accuracy.

[0033] For unclear droplets, i.e., those that are blurry, dimly lit, and have low contrast, the droplets and background can be distinguished by their color depth and chromaticity differences, and the HSV color space can be switched for extraction; this method yields better results.

[0034] For clear images of droplets, perform the following steps; Step S221: Detect the black edges of the droplets in the sub-image, and invert the image to convert the black edges to white while keeping the background black. Originally, the droplets had black edges and a bright background, but after inversion, they became white edges and a black background, making the droplet edge lines extremely prominent and facilitating precise edge capture.

[0035] Step S222: Set a fixed high threshold Preserve the brightest part of the inverted image. Threshold segmentation is achieved in the specified region. Step S223: Perform morphological operations on the segmented image: Morphological operations include closing and opening operations; Step S224: Intelligent merging of small outlines: Iterate through all detected small contours and merge only those contours that meet the following conditions: the distance is less than a set pixel distance; the area of ​​the small contour is greater than a set area pixel. Overall roundness after merging After merging, a complete droplet profile is obtained. ; In summary, only fragments that are close together, have a large enough area, and become more rounded after merging are pieced together, avoiding random merging that could lead to deformation, resulting in a regular and standard circular droplet outline.

[0036] The steps for extracting unclear droplet contours are as follows: Step S31: Extract the dark parts inside the droplet using the HSV color space and adaptively calculate the color threshold; H is hue, S is saturation, and V is brightness. By switching to the HSV space, the dark color features of the droplet that are not visible in ordinary grayscale images can be separated. Then, through adaptive color thresholding, the algorithm automatically calculates the most suitable color filtering range based on the lighting conditions of each image, accurately extracting the main dark area of ​​the droplet, separating blurry, low-contrast, and low-light droplets from the background, and solving the problem of not being able to identify blurry droplets.

[0037] An adaptive color thresholding method is used to calculate the color threshold based on a linear mapping of the image's own brightness and darkness range, thus extracting the dark interior region of a droplet. First, the maximum and minimum pixel values ​​of a single-channel image are calculated to obtain the actual brightness and darkness range of the entire image. Then, a preset base color value is normalized to a standard color level of 0-255 as a scaling factor, and the offset is calculated by combining this with the image's brightness and darkness range. This offset is then superimposed with the minimum pixel value of the image to obtain a segmentation threshold specific to the current image. Since the threshold dynamically changes with the brightness and contrast of each image and is unaffected by external lighting, pixels are ultimately selected based on this threshold to accurately separate the dark interior region of the droplet.

[0038] The following formula is used to calculate the color range specific to an image: lower_blue (lower threshold of the blue channel) = base value / 255 * (current maximum value - current minimum value) + current minimum value; upper_blue (upper limit threshold of the blue channel) = base value / 255 * (current maximum value - current minimum value) + current minimum value.

[0039] Step S32: Select the initial outline with the largest area from the dark region, and use the same intelligent merging strategy as the clear droplet, i.e., perform step S224 above, to improve the roundness of the outline and obtain a complete droplet outline. .

[0040] Example 3: Set the threshold to 0.95, when Determined to be a clear droplet, It was determined to be an unclear droplet.

[0041] In step S221: the white pixel value is 255, and the black background pixel value is 0.

[0042] Step S222: Fix the high threshold , These are the pixel values ​​of the inverted image. Retain pixels with brightness higher than [a certain value] in the inverted image. Threshold segmentation is performed on the original black edge region. The original droplet has a black edge and a bright background; after inversion, it becomes a white edge and a black background. The edge of the inverted droplet is the brightest pixel in the entire image. By setting a fixed high threshold of 0.875 times the maximum pixel brightness, a strict threshold segmentation is performed on the inverted image, retaining only the bright real droplet edge region and filtering out low-brightness background stray light, ghosting, and weak noise. This achieves accurate and interference-free initial extraction of the droplet edge, ensuring the baseline accuracy for subsequent morphological processing and contour merging.

[0043] The set pixel distance is 5, which satisfies the condition. The small outline area is set to 50 pixels, which satisfies the condition. Pixel.

[0044] Example 4: Step S223 includes the following steps: The closing operation involves the following steps: using a 3×3 rectangular structural element to fill the outline gap, using the following formula: ,in For expansion operation, For corrosion operation, For structural elements; The opening operation involves: removing minute noise points and smoothing the edges, using the following formula: .

[0045] Intelligent merging of small contours: Iterates through all detected small contours and merges only those contours that meet the following condition: pixel distance from the edge of the large contour. Small outline area Pixels (filtering out minor noise); overall circularity after merging (Improving roundness); after merging, a complete droplet profile is obtained. .

[0046] Setting a pixel distance d < 5 (distance threshold): In an image with many fragmented droplets, only small fragments that are very close to the main outline (less than 5 pixels) are considered to be the original broken or incomplete parts of the droplet and are allowed to be merged. Fragments that are more than 5 pixels away are directly considered to be distant noise or impurities and are not merged.

[0047] Small outline area Pixel (area threshold); filtering out ultra-small fragments and noisy pixels. Only small outlines with an area greater than 50 pixels are considered valid droplet fragments and have merging value; all small spots with an area less than 50 pixels are judged as imaging noise and tiny specks, ignored, and not included in merging.

[0048] Example 5: The S1 includes the following steps: S11: Read the microscopic droplet image and convert the image into an RGB three-channel data format.

[0049] S12: Apply the CLAHE (Contrast-Limited Adaptive Histogram Equalization) algorithm to each channel to enhance local contrast. The algorithm parameters are set as follows: , The calculation formula is as follows: , in, For a single channel of an RGB image (c=R, G, B). For adaptive histogram equalization operators, The contrast clipping threshold. This determines the size of the grid. By limiting the clipping threshold of the local histogram, excessive noise amplification is avoided, while improving detail in dark areas.

[0050] Here, the CLAHE algorithm is used to divide the image into countless 8×8 local grids, and the contrast of each grid is individually and adaptively adjusted. Local enhancement is performed using 8×8 pixels as the smallest unit, which fits the microscopic imaging size of droplets and accurately improves local edge details; Noise limiting threshold to prevent local over-enhancement.

[0051] Separate channel processing: Each of the R, G, and B channels is individually CLAHE equalized and then merged back into the complete image. Compared to single-channel grayscale enhancement, this method preserves the color gradation and dark area details of the droplets to the greatest extent, highlighting the subtle edges of the blurred droplets and details in dark areas, while keeping the background clean.

[0052] In step 21, the enhanced image is uniformly divided into 8 sub-images, i.e., a 2×4 matrix distribution, and contour detection is performed on each sub-image to reduce noise interference in the global processing.

[0053] Calculate the roundness of the initial contour in each sub-image to determine the sharpness of the droplet: The roundness is determined using the following standard calculation formula: , in, The outline area is based on the number of pixels. The perimeter of the outline is based on the number of pixels. , It is a perfect circle.

[0054] Example 6: It also includes steps for identifying the droplet center and the image center: Step 1: Calculate the center coordinates of the complete droplet profile The centroid formula is used to calculate the mean of the coordinates of all pixels in the contour. Step 2: Calculate the coordinates of the image center ; Step 3: Calculate the offset between the droplet center and the image center. : , .

[0055] In a preferred embodiment, step 4 is further included: motion path calculation for the two-dimensional motion platform. The pixel-to-physical distance conversion factor of the 2D motion platform is Then the physical distance that the two-dimensional motion platform needs to move is: , Two-dimensional motion platform according to The droplet moves along a linear path to align its center with the image center, thus achieving automatic centering.

[0056] Figure 1 and Figure 3 These are images of a clear droplet and an unclear droplet, respectively. Figure 2 and Figure 4Images showing clear and unclear droplet outlines, respectively, obtained through the method provided in this invention, are shown, with the droplet outlines represented by solid red lines. Compared to... Figure 1 Clear images of the droplets. Figure 3 The captured images of droplets are blurry, with indistinct droplet outlines and inconsistent color brightness. The method provided by this invention can not only capture clear droplet images (such as...) Figure 1 Accurate positioning of the outline of the droplet, even for unclear droplet images (such as...). Figure 3 After processing using the method provided by this invention, the droplet outline can also be accurately identified. That is to say, the method provided by this invention can reduce the requirement for droplet image sharpness; it can accurately identify the droplet outline regardless of whether the droplet image is sharp or blurry, avoiding the problem of inaccurate droplet outline identification due to insufficient image sharpness (e.g., locating a non-droplet outline area). The method provided by this invention results in a smoother and clearer droplet outline. Furthermore, since there is a distance between the center of the droplet outline and the image center, the path between the droplet outline center and the image center is calculated using the above embodiment. Then, a two-dimensional motion platform moves the droplet outline (sample to be detected) to coincide with the image center, facilitating subsequent spectral analysis (such as surface-enhanced Raman spectroscopy).

[0057] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for processing cell droplet images, characterized in that, Includes the following steps: S1: Image sharpness enhancement processing; S2: Droplet edge processing, including the following steps: Step 21: Divide the enhanced image into multiple sub-images; Step 22: Process the contours in the sub-images, including background inversion of the edge lines, opening operation and closing operation processing; to obtain smooth contours, merge the processed contours in multiple sub-images into a complete contour to obtain a complete droplet contour.

2. The cell droplet image processing method according to claim 1, characterized in that, In step 22: the roundness of the merged complete droplet outline is calculated, and the clarity of the droplet is determined. Droplets above a set threshold are defined as clear droplets, and those below the set threshold are defined as unclear droplets.

3. The cell droplet image processing method according to claim 2, characterized in that, For clear images of droplets, perform the following steps; Step S221: Detect the black edges of the droplets in the sub-image, invert the image to convert the black edges to white, and keep the background black; Step S222: Set a fixed high threshold Preserve the brightest parts of the inverted image. The region is used to achieve threshold segmentation; Step S223: Perform morphological operations on the segmented image: Morphological operations include closing and opening operations; Step S224: Intelligent merging of small outlines: Iterate through all detected small contours and merge only those contours that meet the following conditions: the distance is less than a set pixel distance; the area of ​​the small contour is greater than a set area pixel. Overall droplet sphericity after merging ; Indicates the roundness of the merged shape. This indicates the initial roundness; after merging, the complete droplet profile is obtained. ; The steps for extracting unclear droplet contours are as follows: Step S31: Extract using HSV color space, adaptively calculate color threshold, and obtain the dark part inside the droplet; Step S32: Select the initial outline with the largest area from the dark region, and use the same intelligent merging strategy as the clear droplet, i.e., perform step S224 above, to improve the roundness of the outline and obtain a complete droplet outline. .

4. The cell droplet image processing method according to claim 3, characterized in that, The set threshold is 0.95, when Determined to be a clear droplet, The droplet was determined to be unclear. In step S221: the white pixel value is 255, and the black background pixel value is 0; Step S222: Fix the high threshold , These are the pixel values ​​of the inverted image; The set pixel distance is 5, which satisfies the condition. The small outline area is set to 50 pixels, which satisfies the condition. Pixel.

5. The cell droplet image processing method according to claim 3, characterized in that, Step S223 includes the following steps: The closing operation involves the following steps: using a 3×3 rectangular structural element to fill the outline gap, using the following formula: ,in For expansion operation, For corrosion operation, For structural elements; The opening operation involves: removing minute noise points and smoothing the edges, using the following formula: .

6. The cell droplet image processing method according to claim 1, characterized in that, The S1 includes the following steps: S11: Read the microscopic droplet image and convert the image into RGB three-channel data format; S12: Apply the CLAHE algorithm to each channel to enhance local contrast. The calculation formula is as follows: ， in, For a single channel of an RGB image (c=R, G, B). For adaptive histogram equalization operators, The contrast clipping threshold. This refers to the size of the grid.

7. The cell droplet image processing method according to claim 1, characterized in that, In step 21, the enhanced image is uniformly divided into 8 sub-images, i.e., a 2×4 matrix distribution, and contour detection is performed on each sub-image to reduce noise interference in the global processing.

8. The cell droplet image processing method according to claim 2, characterized in that, The roundness is determined using the following standard calculation formula: ， in, The outline area is based on the number of pixels. The perimeter of the outline is based on the number of pixels. , It is a perfect circle.

9. The cell droplet image processing method according to claim 1, characterized in that, It also includes steps for identifying the droplet center and the image center: Step 1: Calculate the center coordinates of the complete droplet profile The centroid formula is used to calculate the mean of the coordinates of all pixels in the contour. Step 2: Calculate the coordinates of the image center ; Step 3: Calculate the offset between the droplet center and the image center. : , .

10. The cell droplet image processing method according to claim 9, characterized in that, It also includes step 4: motion path calculation for the two-dimensional motion platform: The pixel-to-physical distance conversion factor of the 2D motion platform is Then the physical distance that the two-dimensional motion platform needs to move is: , Two-dimensional motion platform according to The droplet moves along a linear path to align its center with the image center, thus achieving automatic centering.

Images