Stage runout coupled closed-loop control method, system, and medium

Abstract

The application discloses a kind of object table jump coupling closed loop control method, system and medium, belong to digital pathology scanning technical field.The present object table jump detection relies on single mechanical threshold, to solve the problem that it leads to misjudgment, extensive and cannot distinguish abnormal type of supplementary scanning range.The gist is that: theoretical trajectory, feedback trajectory and tile image are synchronously acquired, motion residual and image residual are calculated, and jump influence fingerprint is constructed by combining tissue diagnosis area weight fusion;Based on the fingerprint, the abnormal type is identified, and the corresponding compensation, replacement, fallback or minimum supplementary scanning strategy is selected;According to the jump risk of completed scanning area, the subsequent scanning parameters are closed loop adjusted.The application is used for digital pathology scanner, can accurately distinguish jump type and splicing pseudo abnormality, differentiates processing according to the importance difference of diagnosis area, reduces invalid supplementary scanning rate, improves scanning efficiency and image quality.

Description

Technical Field

[0001] This invention relates to the field of digital pathological scanning imaging technology, and in particular to a stage jump coupling closed-loop control method, system and medium. Background Technology

[0002] Digital pathology scanners typically require continuous or stepwise image acquisition of tissue regions on glass slides under high magnification conditions such as 20X and 40X. Whole-slice images are then generated through tile stitching, pyramid generation, and image quality control. Due to the small field of view and high pixel resolution of high magnification, even minute fluctuations in the stage during high-speed start-up and stop, uniform movement, line changes, retracements, or localized re-scanning can be magnified into image-level misalignment, localized defocusing, directional ghosting, or abnormal stitching seams.

[0003] Currently, there are three main approaches to handling stage movement issues during digital pathology scanning. The first is based on sensor or encoder-based mechanical anomaly detection, triggering a rescan when mechanical vibration or positional error exceeds a preset threshold. The second involves image quality assessment after or during scanning, rescanning defective areas such as blurriness, noise, folds, and bubbles. The third corrects misalignment between adjacent tiles using overlapping area stitching algorithms. However, the first approach, relying solely on mechanical vibration thresholds, cannot distinguish whether the movement actually affects the diagnostic quality of the current pathological image, easily leading to unnecessary rescanning of blank or weakly textured areas. The second approach, relying only on image sharpness or defect detection, struggles to accurately distinguish between Z-axis movement, instantaneous XY shifts, exposure blur, or stitching artifacts caused by tissue sparseness. While the third approach can correct image coordinates, it cannot feed back the movement risk to subsequent scanning motion parameters, making it difficult to prevent similar movements from recurring.

[0004] Therefore, there is an urgent need for a stage jump coupling closed-loop control method, system, and medium to solve the problems existing in the current technology. Summary of the Invention

[0005] This invention provides a stage jump coupling closed-loop control method, system, and medium, which addresses the problems of existing technologies that cannot distinguish whether stage jump actually affects the pathological diagnostic image quality of the current tile, easily generate invalid supplementary scans in blank areas or weak texture areas, and lack the ability to perform differentiated processing based on the importance of the diagnostic area.

[0006] The core technology of this invention is to fuse motion residuals, image residuals, and tissue diagnostic region weights into a multidimensional jitter-affected fingerprint. Based on this fingerprint, anomaly types are identified and differentiated processing strategies are matched. At the same time, closed-loop adjustments are made to subsequent scanning parameters according to regional jitter risks.

[0007] In a first aspect, the present invention provides a stage bounce coupling closed-loop control method, the method comprising the following steps: Simultaneously acquire the theoretical trajectory, feedback trajectory, and current tile image of the stage, and calculate the motion residual reflecting the deviation between the actual motion and the expected motion; Based on the current tile image and its overlapping area image with adjacent tiles, calculate the image residual. The image residual includes at least the overlap area matching confidence, the abrupt change in image sharpness, and the directional motion blur index. Calculate the tissue diagnostic region weights that characterize diagnostic importance based on the tissue features of the current tile image; Multidimensional feature fusion is performed on motion residuals, image residuals, and tissue diagnostic region weights to construct the jitter impact fingerprint of the current tile, and the processing priority is calculated based on the jitter impact fingerprint. Based on the abnormal types of the fingerprint recognition platform's movement caused by the movement, and according to the abnormal type and processing priority, the corresponding processing strategy is selected for processing. Based on the jitter impact fingerprint of the completed scan area, a regional jitter risk is generated, and the motion parameters of subsequent scans are adjusted in a closed loop according to the regional jitter risk.

[0008] Furthermore, the motion residuals include at least the following three items: The XY residual is calculated based on the deviation between the encoder feedback position and the theoretical position at the midpoint of exposure in the XY plane. Z-residual is calculated based on the difference between the Z-axis position fed back by the encoder and the predicted focal plane height; The trajectory fluctuations during exposure are calculated based on the maximum deviation of the actual trajectory from the fitted trajectory within the exposure time window.

[0009] Furthermore, the reliability of overlapping area matching is obtained by weighted fusion after performing phase correlation detection, feature matching intra-point rate calculation and edge continuity evaluation on the overlapping area of ​​the current tile and the adjacent tiles; The image sharpness mutation is obtained by comparing the current tile's sharpness index with the median sharpness of the same row or neighborhood; Directional motion artifacts are obtained by statistically analyzing the difference ratio of high-frequency edge energy distribution in the direction of scanning motion and its perpendicular direction.

[0010] Furthermore, the tissue diagnostic region weights are calculated based on a weighted combination of at least one of the following tissue features: effective tissue area ratio, edge density, texture intensity, nuclear density proxy index, and effective staining intensity; Among them, the nuclear density proxy index is switched to the corresponding diagnostic signal channel index according to the imaging mode or staining type: for immunohistochemical images, the density of positive regions or the distribution density of positive cells is used; for fluorescence images, the local structural density of the target fluorescence channel is used; and for special staining images, the distribution density or area ratio of the target staining structure is used.

[0011] Furthermore, the jitter impact fingerprint of the current tile is constructed, and the processing priority is calculated based on the jitter impact fingerprint, specifically including: The normalized motion residual, image residual, tissue diagnostic region weight, and anomaly occurrence stage identifier are combined into a multi-dimensional vector form of the jitter effect fingerprint. The basic score for jitter impact is obtained by weighted summation of the inverse values ​​of the normalized motion residuals, image residuals, and stage risk factors. A diagnostic region weight amplification factor is generated based on the weight of the diagnostic region. The processing priority is obtained by multiplying the basic score of the fluctuation impact by the diagnostic region weight amplification factor. Among them, the diagnostic region weight amplification factor makes the treatment priority of diagnostically sensitive regions higher than the base score, and the treatment priority of low diagnostic value regions lower than the base score.

[0012] Furthermore, the anomaly types include at least compensable XY instantaneous offset, Z-axis focal plane disturbance, exposure-time ghosting, start-stop residual vibration, clamping looseness, and sparse splicing pseudo-anomalies. Each anomaly type is matched and identified according to the numerical combination relationship of each dimension in the fingerprint affected by the jump and the preset anomaly identification rules.

[0013] Furthermore, the corresponding handling strategy is selected based on the anomaly type, specifically including: If the anomaly type is a compensable instantaneous XY offset, the tile coordinates are compensated by fusing the relative displacement of the overlapping area image registration and the relative trajectory error of the encoder feedback, and the mosaic image is optimized in the local tile window. If the anomaly type is Z-axis focal plane disturbance, select a high-resolution image with consistent texture and color from the neighboring focal plane candidate images to replace it, or perform local refocusing scan on the affected area. If the anomaly type is exposure-time ghosting, determine the minimum supplementary scan window smaller than the entire line or complete scan band based on the location of the abnormal tile, the affected overlapping edge, and the weight of the tissue diagnostic area, and perform supplementary scan. If the anomaly type is a sparse patching pseudo-anomaly, a mechanical pose backtracking strategy is adopted, and the coordinates are not updated using low-confidence image matching results, but are restored to the physical coordinates of the theoretical trajectory and the feedback trajectory fusion.

[0014] Furthermore, regional-level jitter risk is generated, and the motion parameters of subsequent scans are adjusted based on the closed-loop adjustment of regional-level jitter risk, specifically including: After each row, scan strip, or local area is completed, the regional-level judder risk is calculated based on the weighted combination of the mean, peak, and recurrence anomaly score of the processing priority of each tile in that area; the recurrence anomaly score is used to quantify the degree to which anomalies recur in the same spatial location, the same motion phase, or the same speed segment. The motion parameters of subsequent scans are adjusted according to the regional level jolt risk. The motion parameters include at least one of the following: scan speed, acceleration, exposure trigger point, stabilization window, overlap amount, and stitching search range. The closed-loop adjustment has a hysteresis mechanism. Parameter changes are only executed after the regional-level fluctuation risk of multiple consecutive scan lines meets the adjustment trigger condition, and the original parameters are only restored after multiple consecutive scan lines meet the recovery condition.

[0015] Secondly, the present invention provides a stage jump coupling closed-loop control system, comprising: The trajectory synchronization acquisition module is used to simultaneously acquire the theoretical trajectory, feedback trajectory, exposure time window, and tile image data of the stage during high-magnification scanning. The tile quality analysis module is used to calculate the overlap area matching confidence, image sharpness mutation amount, directional ghosting index, and tissue diagnosis region weight of the current tile. The jitter impact fingerprint generation module is used to fuse motion residuals and image residuals and combine them with tissue diagnostic region weights to generate jitter impact fingerprints and processing priorities. Anomaly type identification module, used to identify anomalies based on the impact of vibration on the fingerprint recognition platform; The adaptive processing module is used to execute the corresponding processing strategies in coordinate compensation, mosaic optimization, focal plane replacement, mechanical pose reversal, or minimum scan based on the anomaly type and processing priority. A closed-loop motion control module is used to adjust the motion parameters of subsequent scans based on the regional-level jolt risk. The quality map output module is used to output the bounce quality map, tile quality labels, and processing records.

[0016] Thirdly, the present invention provides a readable storage medium storing a computer program, the computer program including program code for controlling a process to execute the process, the process including the stage jump coupling closed-loop control method described above.

[0017] The main contributions and innovations of this invention are as follows: 1. This invention constructs a motion-affected fingerprint, coupling motion trajectory residuals with multi-dimensional information such as the reliability of overlapping areas at the image level, sharpness abrupt changes, and ghosting indicators, and combining this with the weights of tissue diagnostic regions for comprehensive judgment. This approach avoids misjudgment based on a single mechanical vibration threshold or a single image quality indicator, accurately distinguishing different types of anomalies such as XY shift, Z-axis perturbation, and exposure ghosting. It prioritizes and rigorously controls diagnostically sensitive areas such as dense kernels and rich edges, improving image quality and the reliability of diagnostic data.

[0018] 2. This invention addresses the low reliability of stitching caused by sparse or blank areas of tissue, identifying it as a "spurious anomaly in sparse tissue stitching" and employing a mechanical pose backoff strategy to avoid blindly triggering rescanning due to lack of texture. Simultaneously, for compensable XY offsets, coordinate compensation or stitching map optimization is used to avoid unnecessary rescanning; for cases requiring rescanning, a minimum rescanning window is calculated based on the anomaly range and tissue weight, requiring only the affected tiles and overlapping edges to be rescanned, reducing the rescanning area and significantly shortening the overall scanning time.

[0019] 3. This invention establishes a row-level or region-level closed-loop control mechanism for runout risk. Based on historical runout fingerprints, a runout risk map is generated, dynamically adjusting parameters such as speed, acceleration, exposure trigger point, stabilization window, and overlap during subsequent scanning processes. This feedback mechanism effectively suppresses repetitive runout anomalies and improves the stability of the entire scan.

[0020] 4. The present invention outputs full slice metadata including quality map, tile quality label and processing record, providing detailed data basis for equipment debugging, after-sales maintenance, scanning quality control analysis and AI diagnostic input quality control.

[0021] Details of one or more embodiments of the present invention are set forth in the following drawings and description, so that other features, objects and advantages of the invention will be more readily understood. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of a stage jump coupling closed-loop control method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the system module composition according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the coupled fingerprint, processing priority, and closed-loop control calculation according to an embodiment of the present invention; Figure 4This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0023] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0024] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0025] This invention provides a stage vibration coupling closed-loop control method, system, and medium for detecting, identifying, differentiating, and controlling minute stage vibrations during continuous or step-by-step image acquisition of tissue regions on a glass slide under high magnification using a digital pathology scanner. The technical solution of this invention is described in detail below with reference to the accompanying drawings.

[0026] See appendix Figure 1 The diagram illustrates the overall flowchart of the stage bounce coupling closed-loop control method of the present invention. See Appendix. Figure 2 The diagram illustrates the modular composition of the system of the present invention. See Appendix. Figure 3 The diagram illustrates the impact of fingerprint movement on processing priority and closed-loop control calculations. The method and system of this invention will now be described in detail with reference to the accompanying drawings.

[0027] Before describing the specific implementation methods in detail, the core terms involved in this invention will be defined first.

[0028] Tile Images The local image obtained by the camera at the i-th acquisition position during high-magnification scanning can be a single image from an area scan camera, a composite image from a line scan camera, or a multi-frame fused image.

[0029] Theoretical Trajectory The control system obtains the desired X, Y, and Z positions and velocities based on the scan path planning, including position, velocity, acceleration, and exposure trigger time.

[0030] Feedback trajectory The actual position acquired by the encoder, servo drive, or position feedback module, which may include multiple points within the exposure time window.

[0031] Image-based displacement estimation The relative displacement estimated by image registration of the overlapping area of ​​adjacent tiles can be calculated by methods such as phase correlation, feature matching, or edge continuity.

[0032] Motion residual R: The degree of inconsistency between the theoretical trajectory and the feedback trajectory, and the image estimated displacement, including XY residual, Z residual, and trajectory fluctuations during exposure.

[0033] Impact of movement on fingerprints A multidimensional vector consisting of motion residuals, image overlap residuals, sharpness abrupt changes, ghosting indicators, and tissue diagnostic region weights is used to identify jitter types and select processing strategies.

[0034] Tissue diagnostic region weight The weight representing the importance of the current tile diagnosis is calculated from factors such as tissue area, nuclear density proxy, edge density, texture intensity, and staining intensity.

[0035] Minimum scan window The minimum spatial range required to re-acquire data to correct jitter is typically smaller than an entire row, column, or full scan band.

[0036] Example 1 This embodiment provides a closed-loop control method for stage bounce coupling, including the following steps: Step 1: Data Synchronization and Acquisition See attached document Figure 1 The step of "synchronously acquiring motion trajectory and tile images" in the document, combined with the attached... Figure 2 The trajectory synchronization acquisition module in the system simultaneously acquires the motion trajectory data of the stage and the current tile image data during high-magnification scanning. Specifically, for the current tile image... The system establishes scan metadata records. The record includes at least: tile number, row number, column number, theoretical coordinates, feedback coordinates, focal plane coordinates, exposure start time, exposure end time, camera trigger time, scan speed, acceleration, overlap edge information, and current tile image.

[0037] Among them, theoretical trajectory The control system obtains the desired X, Y, and Z positions and velocities based on the scan path planning, including position, velocity, acceleration, and exposure trigger time. Feedback trajectory. This refers to the actual position acquired by the encoder, servo drive, or position feedback module. During continuous motion imaging, the feedback trajectory is preferably sampled at multiple points within the exposure time window, forming a set of feedback trajectories during the exposure period.

[0038] If the hardware can only provide positional feedback near the exposure time, then at least the actual position before and after the exposure trigger time should be recorded to estimate displacement fluctuations during the exposure.

[0039] Optionally, the system can also connect to an accelerometer, gyroscope, or rack vibration sensor, but these signals are only used as auxiliary inputs to assess the impact of fingerprint movement on the fingerprint, and not as the sole condition for triggering a rescan. This design aims to avoid misjudgments caused by data from a single sensor and ensure the accuracy of movement detection.

[0040] Step 2: Calculation of motion residuals See attached document Figure 1 The "Calculate Motion Residual" step describes the deviation between the actual motion and the expected motion of the stage, and includes at least three items: XY residual, Z residual, and trajectory fluctuation during exposure.

[0041] For the current tiles At the midpoint of exposure Calculate XY residuals Its expression is:

[0042] in, and These are the X-axis and Y-axis positions fed back by the encoder at the midpoint of the exposure, respectively. and These are the theoretical positions at that moment.

[0043] For the Z direction, the Z residual is calculated using the difference between the current feedback Z position and the predicted focal plane or local focal plane model. Its expression is:

[0044] in, To determine the Z-axis position fed back by the encoder at the midpoint of the exposure, This is the focal height predicted from the focal plane model based on the current position. The focal plane model can be obtained by polynomial fitting or spline interpolation based on the focus data of the preceding scan lines.

[0045] For instantaneous jumps during exposure, calculate the maximum deviation of the actual trajectory from the fitted trajectory within the exposure window. Its expression is:

[0046] FitLine(·) is a set of multi-point sampling trajectories within the exposure window. The reference trajectory function obtained after least squares linear or polynomial fitting, where t is the time of evaluation on the fitted trajectory; This represents the maximum deviation between the actual trajectory and the fitted trajectory.

[0047] The above residuals can be converted to image pixel units for easier evaluation with subsequent overlap region registration results. When the camera pixel corresponds to a sample plane resolution of s micrometers / pixel, the XY residual in pixel units is:

[0048] Step 3: Image Residual Calculation See attached document Figure 1 The "Calculate Image Residuals" step in the document, combined with the attached... Figure 2 The tile quality analysis module uses image residuals to determine whether motion anomalies have manifested as image defects, including at least the confidence level of overlap region matching. Sharpness mutation ΔF and directional motion index .

[0049] (1) Reliability of overlapping area matching The overlapping area of ​​the current tile with the adjacent tiles on the left, top, or front and back is subjected to phase correlation, feature matching, and edge continuity detection, expressed as follows:

[0050] in, This is the normalized result of the phase correlation peak value and the sidelobe ratio; This is the ratio of the number of points matched within a feature to the total number of matches. The score represents the continuity of overlapping edges. All values ​​are normalized to the interval [0,1]. The low confidence threshold is set to 0.55, meaning that values ​​below this threshold require further evaluation based on motion residuals and tissue weights.

[0051] (2) Sharpness mutation ΔF. The current tile sharpness is calculated using Tenengrad, Laplace variance, or high-frequency energy. And compare it with the median of the same row or neighborhood, its expression is:

[0052] in, is the set of sharpness values of the current tile row or neighboring tiles, and ε is a small constant to avoid division by zero. A higher ΔF indicates that the sharpness of the current tile decreases more obviously relative to the neighborhood. The moderate abnormality threshold of ΔF is set to 0.25.

[0053] (3) Directional smearing indicator . The distribution difference of edge or high-frequency energy in the scanning direction and its vertical direction is counted. If the edge is elongated along the motion direction and the high-frequency energy decreases, then increases, and its expression is:

[0054] wherein, clip(x, a, b) represents a clipping function that restricts x to the interval [a, b], if x<a, a is taken; if x>b, b is taken, otherwise x is taken; is high-frequency energy or edge intensity in the scanning direction, is high-frequency energy or edge intensity perpendicular to the scanning direction. The smearing threshold is set to 0.35.

[0055] Step 4: Weight calculation of tissue diagnosis region Refer to Appendix Figure 1 for the step of "calculating weight of tissue diagnosis region", and in combination with the "diagnosis region weight" calculation mechanism shown in Appendix Figure 3 , the tissue diagnosis region weight is used to characterize the importance of the current tile in pathological diagnosis. This indicator is not used to determine whether tissue exists, but to determine whether the region needs higher priority repair once there is a jitter effect. Its expression is:

[0056] wherein, is the proportion of effective tissue area; is edge density; is texture or local high-frequency change intensity; is a nuclear density proxy indicator; is effective staining intensity. All the above items are normalized to the interval [0, 1].

[0057] For different staining types or imaging modes, the nuclear density proxy indicator can be replaced with the corresponding diagnostic signal channel density indicator, so that the re-scanning decision is consistent with the real diagnostic focus object of the current image. Specifically: In the bright-field hematoxylin-eosin (H&E) staining imaging mode, the core morphological features that pathological clinical diagnosis pays most attention to are the polymorphism, differentiation and arrangement aggregation of cell nuclei. Therefore, in this scenario, The proxy configured as the density of blue-purple cell nuclear channels: pure hematoxylin channels are separated by color space deconvolution, and multi-scale local nonmaximum suppression is applied to detect the small connected domains of the main peak of the cell nucleus. The absolute spatial aggregation density of hematoxylin in the tissue region is statistically analyzed as the core diagnostic structural density index for final normalization, thereby generating a high-value response to the highly infiltrated lesion area of ​​tumor cells and a low-value response to the connective tissue or pure background area.

[0058] For immunohistochemical (IHC) images, Replace with DAB-positive region density, positive cell local distribution density, or DAB-positive boundary density. Specifically, after reading the high-magnification multi-channel original image, the IHC-specific deconvolution calculation model is called to decouple the optical signal into a pure hematoxylin channel and a pure brown-yellow DAB-positive color development channel. Within the extracted true DAB color development channel, positive pixel clusters are extracted using local thresholding and morphological connected component labeling operators, and the proportion of the positive color development region area or boundary distribution length relative to the total tissue entity is statistically analyzed.

[0059] For fluorescence images, Replace with the density of cell nuclear channels, the local structural density of target fluorescence channels, or the density of multi-channel co-localization regions. Specifically, after independently collecting the fluorescence of the marker through the bandpass channels, the channel signals are processed at high frequency using the Laplacian of Gaussian (LoG) second-order high-pass differential operator to extract the total amount of high-frequency fluorescent micro-dots, filament membranes, or co-localization patches, and the normalized fluorescence density is obtained.

[0060] For specially stained images Replace with the area ratio, edge density, or structure distribution density of the target stained structure.

[0061] Its technical effect is to improve the priority of compensation or supplementary scanning in diagnostic sensitive areas and reduce invalid supplementary scanning in negative areas, background areas or non-target channels.

[0062] Step 5: Impact of jitter on fingerprint construction and processing priority calculation See attached document Figure 1 The step of "constructing a fingerprint of bouncing effects" in the text, combined with the attached... Figure 2 The vibration affects the fingerprint generation module and its attachments Figure 3 The system implements multi-dimensional feature coupling in the processes of "calculating the basic score of the impact of fluctuations" and "diagnostic weight adjustment processing priority".

[0063] The motion residual, image residual, trajectory fluctuation during exposure, and tissue diagnostic region weights are combined to form the jitter-affected fingerprint J_i, whose expression is:

[0064] in, , and The normalized residual; To identify the stage where an anomaly occurs, enumerated values ​​for stages such as the starting segment, constant speed segment, line break segment, supplementary sweep segment, and ending segment can be used. These enumerated values ​​do not directly participate in the subsequent numerical calculation of the base score, but are mapped to numerical stage risk factors through the stage risk mapping function G(·). It is then used in the calculation.

[0065] The preferred normalization function is a piecewise linear clipping function:

[0066] Where L and U are the lower and upper bounds of normalization, respectively. A set of implementable normalization parameters is as follows: The normalization lower limit is L=0.25 pixels and the upper limit is U=2.0 pixels. Pixels below 0.25 pixels are considered negligible, and pixels above 2.0 pixels are considered obvious XY anomalies. The normalization lower limit L = 0.3 micrometers and the upper limit U = 1.2 micrometers can be adjusted in combination with objective depth of field, focal plane model residuals and Z-axis accuracy; The normalization lower limit L = 0.2 pixels and upper limit U = 1.2 pixels are used to identify the risk of instantaneous jitter or ghosting during exposure.

[0067] The Sigmoid function can also be used to reduce the impact of occasional noise.

[0068] To fully cover the needs of different application scenarios, the system supports adjustments to the normalization limits: For large-scale clinical rapid scanning applications with high load and fast start-stop, the lower limit of the normalization for planar position residuals can be relaxed to L=0.10 pixels and the upper limit to U=3.0 pixels, while the lower limit of the normalization for normal residuals can be adjusted to L=0.1 micrometers and the upper limit to U=2.5 micrometers; For precision scientific research analysis scenarios with 40X or higher objectives and zero tolerance for minute misalignments, the lower limit of the normalization for planar residuals can be tightened to L=0.05 pixels and the upper limit fixed to U=0.8 pixels, while the upper limit of the normal focal plane residuals can be reduced to U=0.5 micrometers.

[0069] Stage risk factors involved in the calculation of the base score D_i Depend on The mapping is obtained, and the mapping relationship is as follows:

[0070] Wherein, G(·) is the stage risk mapping function, which is determined by the actual equipment's factory calibration data, historical scan logs, and quality control anomaly records, and stored in the scan control parameter table. Generally, the stage risk of the starting segment, the line break segment, the supplementary scan segment, and the ending segment is higher than that of the constant speed segment, but the specific risk value of each stage is not fixed in this invention and can be updated according to different equipment structures, motion control parameters, and scan modes.

[0071] Jumping affects the base score The calculation expression is:

[0072] Processing priority Further, this involves calculating the weight of the diagnostic region. First, a weight amplification factor for the diagnostic region is defined. :

[0073] in, The value range is [0.70, 1.30]. Processing priority. The calculation expression is:

[0074] This represents the priority for exception handling; it does not indicate a probability value and is not required to be limited to the [0,1] interval. For diagnostically sensitive regions ( (Higher), allowed Greater than D_i, even when the base score is high. A value greater than 1 is used to increase the priority of compensation or additional scanning; for background areas or areas with low diagnostic value ( (lower), allowed Less than This reduces unnecessary fill-in scans. If used for quality map display or unified normalization output, it can be adopted... , It is only used for display or normalization of output and does not change the actual processing decision logic.

[0075] when At lower levels, even The lower reliability could also be due to blank areas or weak texture areas causing low stitching confidence, preventing the system from directly triggering a rescan; when At higher resolutions, even slight decreases in sharpness or ghosting will increase processing priority.

[0076] A set of feasible judgment thresholds is as follows: The threshold for mild cases is 0.35, initiating coordinate compensation or local stitching optimization. The moderate threshold is 0.55, and the process proceeds to focal plane replacement, local verification, or overlapping edge supplementary scanning. The severity threshold is 0.75, triggering the minimum scan window.

[0077] Step 6: Anomaly Type Identification See attached document Figure 1 The "Identify Anomaly Type" step in the document, combined with the attached... Figure 2 The anomaly type identification module in the system identifies fingerprints based on the impact of vibration. The numerical combinations of each dimension are matched with preset anomaly identification rules to identify anomaly types. Each type corresponds to different identification conditions and processing strategies, avoiding a uniform rescan of all anomalies. Anomaly types include at least the following seven: Class A: Compensable for instantaneous XY offset. Identification criteria are: rise, The trend is decreasing but a stable match still exists; ΔF is low. Low. The corresponding processing strategy is to estimate the displacement based on the image and re-estimate the tile coordinates based on the encoder residual, perform local mosaic optimization, and not perform supplementary scanning.

[0078] Type B: Z-axis focal plane perturbation. Identification criteria are: Or, as the focal plane prediction residual increases, ΔF increases. There is no obvious directional misalignment. The corresponding processing strategy is to select a replacement from the nearby focal plane, the focal plane model of the adjacent tile, or the local refocus image, and perform minimum rescanning if necessary.

[0079] Category C: Image blur during exposure. Identification criteria are... rise, As the image is raised, the edges stretch along the direction of motion, resulting in a decrease in sharpness. The corresponding processing strategy is to mark the current tile as uncompensable, calculate the affected overlapping edges, and perform minimum rescanning.

[0080] Class D: Aftershock during start-stop operation. Identification criteria: anomalies are concentrated in the first few tiles at the beginning, end, or after a line break within each row. Increase. The corresponding handling strategy is to first compensate for the current mild anomaly, and then extend the stabilization window, reduce the initial acceleration, or adjust the exposure trigger point in subsequent rows.

[0081] Class E: Loose clamping or micro-displacement of the slide. Identification criteria include repeated abnormalities at the same spatial location across multiple rows, with the abnormality direction not entirely consistent with the direction of movement. The corresponding handling strategy is to mark the clamping risk, reduce the local velocity, and if the diagnostic area has a high weight, perform a follow-up scan and prompt for re-examination.

[0082] Class F: Sparse tissue splicing pseudo-anomaly. Identification criteria are... Low, but , , All are low, and Low. The corresponding processing strategy is to adopt a mechanical pose reversal strategy, not to perform supplementary scanning, and to reduce the stitching weight of this region.

[0083] Class G: Composite anomalies. Identification criteria include the simultaneous fulfillment of two or more anomaly conditions, such as XY offset superimposed with Z-axis sharpness reduction. The corresponding processing strategy is a combination of processing based on the principles of "prioritizing non-compensable areas, prioritizing diagnostic regions, and prioritizing the smallest possible area."

[0084] Step 7: Processing Strategy Execution See attached document Figure 1 The "Select Processing Strategy" step in the document, combined with the appendix Figure 2 The adaptive processing module in the system employs differentiated processing strategies for different exception types: (1) Coordinate compensation strategy. For Class A compensable instantaneous XY offset, the system fuses the image estimated displacement and the encoder feedback displacement to update the tile coordinates:

[0085] in, This represents the relative offset obtained from image registration; This represents the offset of the encoder feedback relative to the theoretical trajectory; λ is based on the weights of the tissue diagnostic region. and the credibility of overlapping areas Dynamic adjustment, when and When the value is high, λ is taken as 0.6 to 0.8, indicating greater trust in image registration; when... lower or At lower values, λ is set to 0.2 to 0.5, indicating greater trust in the mechanical pose.

[0086] (2) Graph optimization strategy. Treat adjacent tiles as graph nodes, and treat the matching results of overlapping areas, mechanical coordinate constraints and jump impact scores as edge weights. Perform graph optimization on local 3×3 or 5×5 tile windows, reduce the weight of low-confidence edges and increase the weight of edges in high-diagnosis areas.

[0087] (3) Focal plane replacement strategy. For B-type Z-direction focal plane perturbation, if there is an adjacent Z-layer candidate image for the current tile, the image with the highest clarity and consistent with the color and texture of the neighborhood is selected for replacement; if there is no candidate image, only local refocusing and rescanning are performed on the current tile and the affected overlapping edge.

[0088] (4) Mechanical pose backoff strategy. For F-type tissue sparse splicing pseudo-anomalies, the system does not use low-confidence image matching results to update coordinates, but backoffs to the theoretical trajectory and encoder feedback fused coordinates to prevent misregistration of blank areas from causing the entire line splicing to drift.

[0089] (5) Minimum scan strategy. For Class C exposure trailing, severe Z-axis perturbation, or compound anomalies, the system calculates the minimum scan window based on the abnormal tile, affected overlapping edge, motion direction, tissue diagnostic region weight, and neighborhood propagation range. .when When the value is below 0.15 and the anomaly does not affect the tissue area, the system may not perform a follow-up scan, but only write a low-risk quality label; when Higher than 0.60 and When the value is higher than 0.55, even if the abnormal range is small, the affected overlapping edges will be scanned first.

[0090] Step 8: Row-level / Region-level Closed-Loop Control See attached document Figure 1 The steps of "generating a fluctuation risk map" and "adjusting scan parameters in a closed loop" are combined with the appendix. Figure 2 Closed-loop motion control module and accessories Figure 3 In the "formation of row-level or region-level risks" process, after each row, scan band, or local supplementary scan area is completed, the system counts the fingerprint impact of jitter in that area to form row-level or region-level jitter risks. :

[0091] Where RepeatScore(·) is the repeat anomaly scoring function. To handle priority, `location` is a spatial location identifier (such as the row and column grid coordinates of the tile), and `phase` is a motion phase identifier (such as the initial phase, constant velocity phase, etc.). This function describes whether anomalies recur in the same spatial location, the same motion phase, or the same velocity phase. `RepeatScore` can be viewed in the recent history window. Internal statistics show that It can retrieve the current row and several rows preceding it, the current scan band, or the most recent scan records from the same spatial grid. The specific window size is determined by device calibration or operation logs. The calculation method is as follows:

[0092] in, This indicates that within the same spatial grid, the same motion phase, or the same velocity range, The number of times the trigger threshold is exceeded and the exception is determined to be valid; This represents the total number of tiles included in the statistics within this window. To emphasize consecutive anomalies, the number of consecutive times exceeding the threshold can be added as a cumulative item to RepeatScore.

[0093] Closed-loop adjustment methods include, but are not limited to, the following: When the initial segment has N consecutive tiles When the value is greater than 0.55, reduce the initial acceleration and extend the stabilization window by 5 to 20 ms to reduce the residual oscillations during start-stop.

[0094] When the constant speed segment When continuously increasing the exposure speed, reduce the local speed by 5% to 20% and adjust the exposure trigger point to reduce motion blur during exposure.

[0095] When Z-axis disturbances occur repeatedly in a certain spatial region, local focal plane checkpoints are added or local refocusing is enabled to reduce the probability of local defocusing.

[0096] When High area When fluctuations occur, increase the overlap by 5 to 20 pixels and expand the stitching search window to improve the stitching stability of diagnostic sensitive areas.

[0097] When low area When the image is low but the motion is normal, reduce the image matching weight and prioritize mechanical pose backtracking to avoid misregistration of blank areas and invalid supplementary scanning.

[0098] Closed-loop control parameters do not need to be adjusted for every row; a hysteresis mechanism can be set. For example, for two consecutive rows... The speed is reduced only when the value is above 0.55; the speed is restored only when the value is below 0.30 for three consecutive lines, in order to avoid frequent parameter fluctuations.

[0099] Step 9: Output the quality chart See attached document Figure 2 The quality map output module in the system ultimately outputs a full-slice image, tile quality labels, and a bounce quality map. The bounce quality map contains the processing priority, anomaly type label, processing record, and compensation information for each tile, which is written into the full-slice metadata for use in equipment debugging, after-sales maintenance, scanning quality control, and AI diagnostic input quality control.

[0100] This embodiment illustrates the application process of the method in Embodiment 1 in conjunction with a specific scanning scenario.

[0101] Example 1: Processing of Type A Compensable Instantaneous XY Shift in 20X Bright Field Scan Continuous motion imaging was performed using 20X bright-field scanning, with the camera outputting 2448×2048 pixel tiles, with a 192-pixel overlap between tiles. The system recorded the theoretical trajectory, encoder feedback, and exposure time window for each tile. At the 35th tile in a certain row, the system detected the following indicators: =0.48 (below the low confidence threshold of 0.55) =0.70 (significantly higher than normal level), ΔF=0.08 (below the moderate abnormality threshold of 0.25). =0.05 (far below the ghosting threshold of 0.35) =0.72 (high, indicating a diagnostically important area).

[0102] The system determines that the anomaly type of the tile is Class A, compensable XY instantaneous offset, and updates the tile coordinates using the coordinate compensation formula, λ according to... =0.72 and =0.48 was adjusted to 0.65 to give more confidence to the image registration results. Subsequently, the mosaic image was re-optimized in a local 5×5 tile window using the overlap area matching degree, mechanical coordinate constraints, and jitter baseline scores as edge weights, without performing any additional scanning operations.

[0103] Example 2: Handling of Z-direction focal plane perturbation in type B At the 62nd tile of the same scan line, the system detected the following metric: ΔF = 0.42 (significantly higher than the moderate anomaly threshold of 0.25). =0.68 (relatively high, indicating significant deviation in the Z direction) =0.73 (relatively normal, no obvious directional misalignment) =0.04 (low, no ghosting) =0.81 (weight of high diagnostic regions).

[0104] The system determines that the anomaly type of the tile is a Class B Z-axis focal plane perturbation. If there is a candidate image of the adjacent focal plane in the area, the candidate image with the highest sharpness and color consistency that meets the threshold is selected as the replacement; if there is no candidate image, a local refocusing scan task is generated for the current tile and its right overlapping edge.

[0105] Example 3: Processing of Type D start-stop afterimages superimposed with Type C exposure motion blur In the first four tiles after the start of a certain scan band, the system detected all of them. >0.65 (significantly higher than normal level) >0.40 (above the ghosting threshold of 0.35), and the anomaly occurs in the initial segment ( (Elevation). The system classifies it as a composite anomaly (Type G) of Class D start-stop aftershock superimposed with Class C exposure ghosting. Following the principles of "prioritizing non-compensable areas, prioritizing diagnostic regions, and prioritizing the smallest affected areas," the affected high-altitude areas are... The tile performs minimum refresh scan, while reducing the starting acceleration of the next line by 15%, increasing the stabilization window by 10ms, and delaying the exposure trigger point until the encoder speed residual stabilizes.

[0106] Example 4: Handling pseudo-anomalies of sparse splicing in type F tissues In areas with fat or blank background, the system detected the following indicators: =0.32 (low), but =0.05 (extremely low) =0.04 (extremely low) =0.03 (extremely low), ΔF=0.02 (extremely low) =0.08 (far below the diagnostic importance threshold). The system determined it to be a pseudo-anomaly of sparse splicing in type F tissue. It adopted a mechanical pose backtracking strategy, which did not use the low-confidence image matching results to update the coordinates, but instead backtracked to the fused coordinates of the theoretical trajectory and encoder feedback. No additional scanning was performed, and it was marked as "low-texture mechanical pose passed" in the quality map.

[0107] Example 2 See attached document Figure 2 This embodiment provides a stage jump coupling closed-loop control system for executing the method of Embodiment 1 described above. The system includes the following modules: The trajectory (and exposure) synchronous acquisition module is used to simultaneously acquire the theoretical trajectory, feedback trajectory, exposure time window, and camera trigger information of the stage during high-magnification scanning. The module's inputs are motion control data and camera trigger data, and its output is scan metadata. .

[0108] The tile quality analysis module calculates the sharpness, motion blur index, overlap area matching confidence, and tissue diagnostic region weights for the current tile and its adjacent tiles. The input to this module is a tile image. And neighboring tiles, output as , ΔF, and .

[0109] The motion-affected fingerprint generation module is used to fuse motion residuals and image residuals to generate motion-affected fingerprints. and processing priority The input to this module is scanned metadata. The output includes image quality metrics, and measures the impact of jitter on fingerprints and processing priorities.

[0110] An anomaly type identification module is used to identify anomalies based on the impact of vibration on the fingerprint recognition platform. The input to this module is... The output is an exception type label.

[0111] The adaptive processing module is used to execute corresponding processing strategies among coordinate compensation, mosaic optimization, focal plane replacement, mechanical pose reversal, or minimum rescanning based on the anomaly type and processing priority. The module's inputs are the anomaly type and strategy parameters, and its output is the corrected tile or rescanning task.

[0112] A closed-loop motion control module is used to assess row-level or region-level fluctuation risks. Adjust the speed, acceleration, exposure trigger point, stabilization window, and overlap for subsequent scans. The input to this module is... or The output is the parameters for the next scan segment.

[0113] The quality map output module is used to write the bounce quality map, tile quality labels and compensation records into the full tile metadata, and output quality reports, post-sales analysis data and AI input quality control tags.

[0114] The data flow between the above modules is as follows: The trajectory synchronization acquisition module sends its output scan metadata to the tile quality analysis module and the jitter impact fingerprint generation module respectively; the tile quality analysis module sends its output image quality index to the jitter impact fingerprint generation module; the jitter impact fingerprint generation module sends its output jitter impact fingerprint to the anomaly type recognition module, and simultaneously sends the processing priority to the adaptive processing module; the anomaly type recognition module sends its output anomaly type label to the adaptive processing module; the adaptive processing module feeds back its execution result to the closed-loop motion control module; the closed-loop motion control module sends the adjusted scan parameters to the underlying motion controller; and the quality map output module obtains the processing records from each module and generates the final output.

[0115] Example 3 This embodiment also provides an electronic device, see reference. Figure 2 It includes a memory 402 and a processor 401, the memory 402 storing a computer program and the processor 401 being configured to run the computer program to perform the steps in any of the above method embodiments.

[0116] Specifically, the processor 401 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0117] The memory 402 may include a mass storage device for data or instructions. For example, and not limitingly, the memory 402 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 402 may include removable or non-removable (or fixed) media. Where appropriate, the memory 402 may be internal or external to a data processing device. In a particular embodiment, the memory 402 is non-volatile memory. In a particular embodiment, the memory 402 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random Access Memory (FPMDRAM), Extended Data Out Dynamic Random Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0118] The memory 402 can be used to store or cache various data files that need to be processed and / or communicated, as well as possible computer program instructions executed by the processor 401.

[0119] The processor 401 reads and executes computer program instructions stored in the memory 402 to implement any of the stage jump coupling closed-loop control methods in the above embodiments.

[0120] Optionally, the electronic device may further include a transmission device 403 and an input / output device 404, wherein the transmission device 403 is connected to the processor 401 and the input / output device 404 is connected to the processor 401.

[0121] The transmission device 403 can be used to receive or send data via a network. Specific examples of the network described above may include wired or wireless networks provided by the communication provider of the electronic device. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 403 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0122] Input / output device 404 is used for inputting or outputting information. It can be a speaker, microphone, monitor, or keyboard.

[0123] Example 4 This embodiment also provides a readable storage medium storing a computer program, the computer program including program code for controlling a process to execute the process, the process including the stage jump coupling closed-loop control method according to Embodiment 1.

[0124] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0125] Generally, various embodiments can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects of the invention can be implemented in hardware, while others can be implemented by firmware or software executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flowcharts, or using some other graphical representation, it should be understood that, by way of non-limiting example, these blocks, apparatuses, systems, techniques, or methods described herein can be implemented in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0126] Embodiments of the present invention can be implemented by computer software, which may be executable by a data processor of a mobile device, such as a processor entity, or by hardware, or by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets, and / or macros can be stored in any device-readable data storage medium, and they include program instructions for performing specific tasks. The computer program product may include one or more computer-executable components configured to perform the embodiments when the program is run. The one or more computer-executable components may be at least one piece of software code or a portion thereof. Additionally, it should be noted in this respect that, as Figure 1 Any box in the logical flow can represent a program step, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. Software can be stored on physical media such as memory chips or blocks of storage implemented within a processor, magnetic media such as hard disks or floppy disks, and optical media such as DVDs and their data variants, CDs, etc. The physical medium is a non-transient medium.

[0127] Those skilled in the art should understand that the technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0128] The above embodiments are merely illustrative of several implementations of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the appended claims.

Claims

1. A closed-loop control method for stage bounce coupling, characterized in that, Includes the following steps: Simultaneously acquire the theoretical trajectory, feedback trajectory, and current tile image of the stage, and calculate the motion residual reflecting the deviation between the actual motion and the expected motion; Based on the current tile image and its overlapping area image with adjacent tiles, calculate the image residual, which includes at least the overlap area matching confidence, the image sharpness abrupt change, and the directional motion blur index; Based on the tissue features of the current tile image, calculate the tissue diagnostic region weights that characterize diagnostic importance; The motion residual, the image residual, and the tissue diagnostic region weights are fused using multidimensional features to construct the jitter impact fingerprint of the current tile, and the processing priority is calculated based on the jitter impact fingerprint. Based on the abnormal type of the vibration affecting the fingerprint recognition platform, and according to the abnormal type and the processing priority, the corresponding processing strategy is selected for processing. Based on the jitter impact fingerprint of the completed scan area, which is a multi-dimensional vector that integrates motion deviation, image quality defects and diagnostic importance, a region-level jitter risk is generated, and the motion parameters of subsequent scans are adjusted in a closed loop according to the region-level jitter risk.

2. The method as described in claim 1, characterized in that, The motion residuals include at least the following three items: The XY residual is calculated based on the deviation between the encoder feedback position and the theoretical position at the midpoint of exposure in the XY plane. Z-residual is calculated based on the difference between the Z-axis position fed back by the encoder and the predicted focal plane height; The trajectory fluctuations during exposure are calculated based on the maximum deviation of the actual trajectory from the fitted trajectory within the exposure time window.

3. The method as described in claim 1, characterized in that, The reliability of the overlapping area matching is obtained by weighted fusion after performing phase correlation detection, feature matching intra-point rate calculation and edge continuity evaluation on the overlapping area between the current tile and the adjacent tiles. The image sharpness mutation amount is obtained by comparing the current tile's sharpness index with the median sharpness of the same row or neighborhood; The directional motion blur index is obtained by statistically analyzing the difference ratio of high-frequency edge energy distribution in the direction of scanning motion and its perpendicular direction.

4. The method as described in claim 1, characterized in that, The tissue diagnostic region weights are calculated based on a weighted combination of at least one of the following tissue features: effective tissue area ratio, edge density, texture intensity, nuclear density proxy index, and effective staining intensity. The nuclear density proxy index is switched to the corresponding diagnostic signal channel index according to the imaging mode or staining type: for immunohistochemical images, positive region density or positive cell distribution density is used; for fluorescence images, local structure density of the target fluorescence channel is used; and for special staining images, distribution density or area ratio of the target staining structure is used.

5. The method as described in claim 1, characterized in that, The process of constructing the bounce effect fingerprint of the current tile and calculating the processing priority based on the bounce effect fingerprint specifically includes: The normalized motion residual, image residual, tissue diagnostic region weight, and anomaly occurrence stage identifier are combined into a multi-dimensional vector form of the jitter effect fingerprint. The basic score for jitter impact is obtained by weighted summation of the inverse values ​​of the normalized motion residuals, image residuals, and stage risk factors. A diagnostic region weight amplification factor is generated based on the weight of the tissue diagnostic region, and the processing priority is obtained by multiplying the basic score of the jump effect with the diagnostic region weight amplification factor. The diagnostic region weight amplification factor prioritizes diagnostically sensitive regions over the base score, and prioritizes low-diagnostic-value regions over the base score.

6. The method as described in claim 1, characterized in that, The anomaly types include at least compensable XY instantaneous offset, Z-axis focal plane disturbance, exposure-time ghosting, start-stop residual vibration, clamping looseness, and sparse tissue splicing pseudo-anomalies. Each anomaly type is matched and identified according to the numerical combination relationship of each dimension in the fingerprint affected by the jump and the preset anomaly identification rules.

7. The method as described in claim 6, characterized in that, Selecting the corresponding handling strategy based on the anomaly type specifically includes: If the anomaly type is a compensable instantaneous XY offset, the tile coordinates are compensated by fusing the relative displacement of the overlapping area image registration and the relative trajectory error of the encoder feedback, and the stitching image is optimized in the local tile window; If the anomaly type is Z-axis focal plane disturbance, select a high-resolution image with consistent texture and color from the neighboring focal plane candidate images to replace it, or perform local refocusing scan on the affected area. If the anomaly type is motion blur during exposure, determine a minimum supplementary scan window smaller than the entire line or complete scan band based on the position of the abnormal tile, the affected overlapping edge, and the weight of the tissue diagnostic region, and perform supplementary scan. If the anomaly type is a pseudo-anomaly caused by sparse tissue splicing, a mechanical pose backtracking strategy is adopted, and the coordinates are not updated using low-confidence image matching results, but are restored to the physical coordinates fused from the theoretical trajectory and the feedback trajectory.

8. The method as described in claim 1, characterized in that, The process of generating regional-level jitter risk and adjusting the motion parameters of subsequent scans in a closed loop based on the regional-level jitter risk specifically includes: After each row, scan band, or local area is completed, the regional-level jitter risk is calculated based on the weighted combination of the mean, peak, and recurrence anomaly score of the processing priority of each tile in that area; the recurrence anomaly score is used to quantify the degree to which anomalies recur in the same spatial location, the same motion phase, or the same speed segment. The motion parameters of subsequent scans are adjusted according to the regional-level jitter risk, and the motion parameters include at least one of the following: scan speed, acceleration, exposure trigger point, stabilization window, overlap amount, and stitching search range; The closed-loop adjustment is equipped with a hysteresis mechanism, which requires that the parameter change be performed only after the regional level jump risk of multiple consecutive scan lines meets the adjustment trigger condition, and the original parameters be restored only after the recovery condition is met for multiple consecutive scan lines.

9. A stage jumping coupled closed-loop control system, characterized in that, include: The trajectory synchronization acquisition module is used to simultaneously acquire the theoretical trajectory, feedback trajectory, exposure time window, and tile image data of the stage during high-magnification scanning. The tile quality analysis module is used to calculate the overlap area matching confidence, image sharpness mutation amount, directional ghosting index, and tissue diagnosis region weight of the current tile. The jitter impact fingerprint generation module is used to fuse motion residuals and image residuals and combine them with tissue diagnostic region weights to generate a jitter impact fingerprint and processing priority; the jitter impact fingerprint is a multi-dimensional vector that fuses motion deviation, image quality defects and diagnostic importance; An anomaly type identification module is used to identify the anomaly type affecting the fingerprint recognition platform's movement. The adaptive processing module is used to execute the corresponding processing strategies among coordinate compensation, mosaic optimization, focal plane replacement, mechanical pose reversal, or minimum scan based on the anomaly type and the processing priority. A closed-loop motion control module is used to adjust the motion parameters of subsequent scans based on the regional-level jolt risk. The quality map output module is used to output the bounce quality map, tile quality labels, and processing records.

10. A readable storage medium, characterized in that, The readable storage medium stores a computer program, the computer program including program code for controlling a process to execute the process, the process including the stage jump coupling closed-loop control method according to any one of claims 1 to 8.

Images