A sliding window based visual displacement time series data filtering method and system

Abstract

The application provides a sliding window-based visual displacement time series data filtering method and system, which is applied to a motion controller, and the method comprises the following steps: acquiring a continuous image sequence of multiple frames, and obtaining a target position measurement value sequence based on the image sequence; based on the target position measurement value sequence, sequentially acquiring a historical measurement value sequence under a preset sliding window and a preset sliding step, and sequentially executing a preset filtering processing step based on the historical measurement value sequence until the sliding window traverses the entire target position measurement value sequence, thereby obtaining a plurality of filtering output values; wherein after each execution of the filtering processing step, the length of a residual measurement value sequence is obtained based on the target position measurement value sequence and the sliding window, and the sliding step of the sliding window is updated based on the length of the residual measurement value sequence; and based on the plurality of filtering output values, displacement state analysis processing is performed on the motion controller to obtain a displacement measurement result. The application can improve the stability of the displacement measurement result.

Description

Technical Field

[0001] This invention relates to the field of image signal processing technology, and in particular to a method and system for filtering visual displacement temporal data based on a sliding window. Background Technology

[0002] Traditional visual displacement measurement methods acquire consecutive frame images, extract features from each frame to obtain target position data, and then use this target position data for displacement measurement. However, due to the lack of temporal consistency between the obtained target position data, the displacement measurement results suffer from severe jumps and discontinuities, resulting in low reliability of the displacement measurement results.

[0003] To address the aforementioned issues, existing visual displacement measurement methods combine edge extraction and feature fitting algorithms to extract target position data from consecutive frames of images. However, these methods are susceptible to factors such as ambient light variations, thermal disturbances, imaging jitter, and unstable target textures, resulting in short-term jitter and high-frequency noise in the obtained target position data, severely impacting the stability and accuracy of the displacement measurement results. Furthermore, existing visual displacement measurement methods employ algorithms such as median filtering and Kalman filtering to filter the target position data; however, due to their high algorithm complexity and difficulty in parameter tuning, they exhibit poor real-time performance on resource-constrained devices, and the displacement measurement results remain relatively unstable. Summary of the Invention

[0004] The present invention aims to provide a method and system for filtering visual displacement time series data based on a sliding window, so as to solve the above-mentioned technical problems and improve the stability of displacement measurement results.

[0005] To address the aforementioned technical problems, this invention provides a visual displacement temporal data filtering method based on a sliding window, applied to a motion controller. The method includes the following steps:

[0006] Acquire a sequence of multiple consecutive frames of images, and based on the image sequence, obtain a sequence of target location measurement values;

[0007] Based on the target location measurement value sequence, historical measurement value sequences are sequentially acquired through a preset sliding window with a preset sliding step size. Preset filtering processing steps are then sequentially executed based on the historical measurement value sequences until the sliding window traverses the entire target location measurement value sequence, resulting in several filtered output values. After each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target location measurement value sequence and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length.

[0008] The motion controller is subjected to displacement state analysis based on several of the filtered output values ​​to obtain displacement measurement results.

[0009] In the above scheme, based on the target location measurement value sequence and the sliding window, several target location measurement values ​​in the target location measurement value sequence that have not been processed by the sliding window can be obtained, thus obtaining the total number of target location measurement values ​​that have not been processed by the sliding window, and using this total number as the length of the remaining measurement value sequence; and the process of the sliding window processing based on the target location measurement value sequence is implemented according to the sliding step size of the sliding window, which determines the distance the sliding window moves each time. The longer the sliding step size, the less time is required for the sliding window to traverse the entire target location measurement value sequence. Existing technology sets the sliding step size to a fixed value for processing. When processing the end of the target location measurement value sequence, if the length of the remaining measurement value sequence is less than the length of the sliding window, the sliding window may obtain incomplete data from the historical measurement value sequence, resulting in reduced reliability of the obtained filtered output value. Furthermore, if the fixed value is set too small, the sliding window will have a large overlapping area, resulting in a large amount of redundant calculation in subsequent data processing, reducing the overall computational efficiency. Therefore, this scheme dynamically updates the sliding window step size by changing the length of the remaining measurement value sequence. Compared with the existing technology, this improves the computational efficiency and the reliability of the obtained filtered output value, thereby improving the stability of the displacement measurement results.

[0010] Furthermore, the target position measurement sequence obtained by this scheme based on a sequence of consecutive multi-frame image sequences has temporal order, and the historical measurement sequence obtained sequentially under the sliding window also has temporal order. Therefore, there is good continuity between the several filtered output values ​​obtained in each filtering step, and there will be no data abrupt change, which improves the stability of displacement measurement results.

[0011] Further, obtaining the target location measurement value sequence based on the image sequence includes: performing feature extraction based on the image sequence to obtain an edge contour sequence; extracting an edge contour point set sequence based on the edge contour sequence, and performing fitting processing based on the edge contour point set sequence to obtain a center point coordinate parameter sequence; and obtaining the target location measurement value sequence based on the center point coordinate parameter sequence.

[0012] Further, the step of extracting features based on the image sequence to obtain an edge contour sequence includes: performing image enhancement based on the image sequence to obtain an enhanced image sequence; and performing edge extraction based on the enhanced image sequence to obtain an edge contour sequence.

[0013] This scheme enhances image sequences by suppressing random noise and improving contrast, resulting in clearer boundaries between the target location and other parts of the image. It reduces the impact of complex environments such as uneven lighting and image blurring on image processing, improving the accuracy of the obtained edge contours and consequently the accuracy of the target location measurement sequence. Furthermore, this scheme converts the edge contours into discrete edge contour point sets. Fitting these point sets allows for the identification of a stable and representative geometric center of the edge contour, yielding the coordinate parameters of the target feature center point in the image pixel space—the center point coordinate parameters. Edge contour sequences obtained from these center point coordinate parameters are more stable and reliable.

[0014] Furthermore, the filtering process includes: performing outlier removal processing based on the historical measurement value sequence to obtain a reliable dataset; and obtaining a filtered output value based on the reliable dataset under a preset filtering calculation mechanism.

[0015] In the above scheme, after outlier removal processing of the historical measurement value sequence, the high-frequency jumps caused by environmental interference, imaging noise or single-frame image anomalies in the historical measurement value sequence can be made smoother, resulting in a reliable dataset without abnormal mutations. The resulting filtered output value has better reliability, thereby improving the stability of subsequent displacement measurement results.

[0016] Furthermore, the outlier removal process based on the historical measurement value sequence to obtain a reliable dataset includes: sorting the historical measurement values ​​based on their numerical values ​​to obtain a historical measurement value sorting sequence; and removing outliers based on the historical measurement value sorting sequence under a preset measurement error tolerance to obtain a reliable dataset.

[0017] Further, the step of performing a removal process based on the historical measurement value sorting sequence under a preset measurement error tolerance to obtain a reliable dataset includes: obtaining a removal number value based on the preset measurement error tolerance; wherein the removal number value is less than the total number of historical measurement values ​​in the historical measurement value sorting sequence; and performing a removal process on the historical measurement value sorting sequence based on the removal number value to obtain a reliable dataset.

[0018] In the above scheme, the historical measurement value sequence is sorted by numerical value. Outliers with excessively small or large values ​​tend to cluster at both ends. Based on a preset measurement error tolerance, historical measurement values ​​with extreme values ​​at both ends of the sequence are considered outliers and removed. This results in a smaller variance in the obtained reliable dataset and a more stable and smooth data distribution. Therefore, this scheme avoids interference from individual outlier frames on the overall measurement results, improves resistance to noise and extreme measurement errors, and enhances the stability of subsequent displacement measurement results. Furthermore, the measurement error tolerance of this scheme can be adjusted adaptively according to different scenarios. Therefore, this scheme can be used in high-frequency measurements requiring data smoothing, as well as in medium- and low-frequency measurements requiring stable dynamic data tracking, demonstrating good versatility.

[0019] Further, obtaining the filtered output value based on the trusted dataset under a preset filtering calculation mechanism includes: obtaining an integration ratio based on the total number of trusted data in the trusted dataset and the total number of historical measurements in the historical measurement value sequence; performing summation processing based on the trusted dataset to obtain a total trusted historical measurement value; and performing calculation processing based on the total trusted historical measurement value and the integration ratio to obtain the filtered output value.

[0020] The above scheme can obtain reliable and stable filtered output values ​​by simply sorting and calculating. Compared with existing technologies that require complex filtering algorithms, this scheme can avoid complex modeling and parameter fitting processes and reduce computational complexity. Therefore, this scheme is more advantageous for application on resource-constrained equipment, can meet the real-time requirements of the equipment, and can also ensure the accuracy of the displacement measurement results.

[0021] Further, the step of obtaining the length of the remaining measurement value sequence based on the target position measurement value sequence and the sliding window, and updating the sliding step size of the sliding window based on the remaining measurement value sequence length, includes: obtaining the length of the remaining measurement value sequence based on the target position measurement value sequence and the sliding window; if the length of the remaining measurement value sequence is not less than the length of the sliding window, then updating the sliding step size of the sliding window to a preset step size; if the length of the remaining measurement value sequence is less than the length of the sliding window, then updating the sliding step size of the sliding window to 1 frame.

[0022] The above scheme dynamically updates the sliding window step size by adjusting the length of the remaining measurement value sequence. When there are many remaining measurement values, the sliding window moves faster and reduces redundant data processing. When there are few remaining measurement values, the sliding window moves slower to avoid incomplete data in the obtained historical measurement value sequence due to an excessively large sliding step size. Therefore, this scheme can improve computational efficiency, meet the real-time requirements of applications on resource-limited devices, and ensure the reliability of the obtained filtered output values, thereby improving the stability of displacement measurement results.

[0023] This invention also provides a visual displacement temporal data filtering system based on a sliding window, comprising: a target position measurement value acquisition module, used to acquire a sequence of consecutive multi-frame image images and obtain a target position measurement value sequence based on the image sequence; a sliding window filtering module, used to acquire historical measurement value sequences sequentially through a preset sliding window at a preset sliding step size based on the target position measurement value sequence obtained by the target position measurement value acquisition module, and to execute preset filtering processing steps sequentially based on the historical measurement value sequences until the sliding window traverses the entire target position measurement value sequence to obtain several filtered output values; wherein, after each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target position measurement value sequence obtained by the target position measurement value acquisition module and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length; and a displacement measurement module, used to perform displacement state analysis processing on the motion controller based on the several filtered output values ​​obtained by the sliding window filtering module to obtain displacement measurement results.

[0024] Furthermore, in the sliding window filtering module, the filtering process includes: performing outlier removal processing based on the historical measurement value sequence to obtain a reliable dataset; and obtaining a filtered output value based on the reliable dataset under a preset filtering calculation mechanism.

[0025] The above scheme can adjust the sliding step size of the sliding window in real time according to the length of the remaining measurement value sequence. This ensures that when the sliding window slides to the end of the target position measurement value, the obtained historical measurement value sequence will not have incomplete data. Therefore, this scheme can handle different data sequences well and has good stability, thereby improving the stability of displacement measurement results. Attached Figure Description

[0026] Figure 1 A flowchart illustrating the technical implementation of a visual displacement temporal data filtering method based on a sliding window, as provided in an embodiment of the present invention;

[0027] Figure 2This is a schematic diagram of a visual displacement time-series data filtering system architecture based on a sliding window, provided as an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0029] Please see Figure 1 This embodiment provides a visual displacement time-series data filtering method based on a sliding window, applied to a motion controller. The method includes the following steps:

[0030] Step S1: Obtain a sequence of multiple consecutive frames of images, and based on the image sequence, obtain a sequence of target location measurement values;

[0031] Step S2: Based on the target location measurement value sequence, historical measurement value sequences are sequentially acquired through a preset sliding window with a preset sliding step size, and preset filtering processing steps are sequentially executed based on the historical measurement value sequences until the sliding window traverses the entire target location measurement value sequence to obtain several filtered output values; wherein, after each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target location measurement value sequence and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length;

[0032] Step S3: Perform displacement state analysis processing on the motion controller based on several of the filtered output values ​​to obtain displacement measurement results.

[0033] In the above embodiments, based on the target location measurement value sequence and the sliding window, several target location measurement values ​​in the target location measurement value sequence that have not been processed by the sliding window can be obtained, thereby obtaining the total number of target location measurement values ​​that have not been processed by the sliding window, and using this total number as the length of the remaining measurement value sequence; and the process of the sliding window processing based on the target location measurement value sequence is implemented according to the sliding step size of the sliding window, which determines the distance the sliding window moves each time. The longer the sliding step size, the less time is required for the sliding window to traverse the entire target location measurement value sequence. The prior art sets the sliding step size to a fixed value for processing. When processing the end of the target location measurement value sequence, if the length of the remaining measurement value sequence is less than the length of the sliding window, the sliding window will obtain incomplete data from the historical measurement value sequence, resulting in reduced reliability of the obtained filtered output value. Moreover, if the fixed value is set too small, the sliding window will have a large overlapping area, resulting in a large amount of redundant calculation in subsequent data processing, reducing the overall computational efficiency. Therefore, this embodiment dynamically updates the sliding step size of the sliding window by the length of the remaining measurement value sequence. Compared with the prior art, this improves the computational efficiency and the reliability of the obtained filtered output value, thereby improving the stability of the displacement measurement results.

[0034] Furthermore, the target position measurement value sequence obtained from the image sequence of multiple consecutive frames in this embodiment has temporal order, and the historical measurement value sequence obtained sequentially under the sliding window also has temporal order. Therefore, there is good continuity between the several filtered output values ​​obtained in each filtering processing step, and there will be no data mutation problem, which improves the stability of displacement measurement results.

[0035] In one embodiment, the length of the sliding window is set to k, for storing the target position measurement sequence {a1, a2, ..., a...}. j ,…,a T} has k target position measurements, where a j Let be the j-th target location measurement value in the target location measurement value sequence, and T be the total number of target location measurement values ​​in the target location measurement value sequence, where T≥k. In this embodiment, during one filtering process, the historical measurement value sequence obtained by the sliding window is {a}. j ,a j+1 ,…,a j+k-1}, then after performing this filtering step, based on the target position measurement value sequence and the sliding window, the length of the remaining measurement value sequence is L = T - (j + k - 1), and the sliding step size of the sliding window is updated based on the length of the remaining measurement value sequence. In this embodiment, the updated sliding step size is z. Then, in the next filtering step, the historical measurement value sequence obtained by the sliding window through the updated sliding step size is {a j+z ,aj+1+z ,…,a j+k-1+z}

[0036] Further, obtaining the target location measurement value sequence based on the image sequence includes: performing feature extraction based on the image sequence to obtain an edge contour sequence; extracting an edge contour point set sequence based on the edge contour sequence, and performing fitting processing based on the edge contour point set sequence to obtain a center point coordinate parameter sequence; and obtaining the target location measurement value sequence based on the center point coordinate parameter sequence.

[0037] Further, the step of extracting features based on the image sequence to obtain an edge contour sequence includes: performing image enhancement based on the image sequence to obtain an enhanced image sequence; and performing edge extraction based on the enhanced image sequence to obtain an edge contour sequence.

[0038] This embodiment enhances the image sequence, suppressing random noise and enhancing contrast. This makes the boundary between the target location and other parts of the image clearer, reducing the impact of complex environments such as uneven lighting and image blurring on the image processing. This improves the accuracy of the obtained edge contours, and consequently, the accuracy of the target location measurement sequence. Furthermore, this embodiment converts the edge contours into discrete edge contour point sets. Fitting these point sets allows for the identification of a stable geometric center that best represents the edge contour. This yields the coordinate parameters of the target feature center point in the image pixel space, i.e., the center point coordinate parameters. The edge contour sequence obtained from these center point coordinate parameters is more stable and reliable.

[0039] Preferably, in this embodiment, the enhanced image sequence is obtained by filtering and binarizing the image sequence.

[0040] Furthermore, the filtering process includes: performing outlier removal processing based on the historical measurement value sequence to obtain a reliable dataset; and obtaining a filtered output value based on the reliable dataset under a preset filtering calculation mechanism.

[0041] In the above embodiments, after outlier removal processing of the historical measurement value sequence, the high-frequency jumps caused by environmental interference, imaging noise or single-frame image anomalies in the historical measurement value sequence can be made smoother, resulting in a reliable dataset without abnormal mutations. The resulting filtered output value has better reliability, thereby improving the stability of subsequent displacement measurement results.

[0042] Furthermore, the outlier removal process based on the historical measurement value sequence to obtain a reliable dataset includes: sorting the historical measurement values ​​based on their numerical values ​​to obtain a historical measurement value sorting sequence; and removing outliers based on the historical measurement value sorting sequence under a preset measurement error tolerance to obtain a reliable dataset.

[0043] Further, the step of performing a removal process based on the historical measurement value sorting sequence under a preset measurement error tolerance to obtain a reliable dataset includes: obtaining a removal number value based on the preset measurement error tolerance; wherein the removal number value is less than the total number of historical measurement values ​​in the historical measurement value sorting sequence; and performing a removal process on the historical measurement value sorting sequence based on the removal number value to obtain a reliable dataset.

[0044] In the above embodiments, the historical measurement value sorting sequence obtained by sorting the historical measurement values ​​shows that outliers with excessively small or large values ​​are concentrated at both ends. Based on a preset measurement error tolerance, historical measurement values ​​with extreme values ​​at both ends of the historical measurement value sorting sequence can be considered outliers and removed. This results in a smaller variance in the obtained reliable dataset and a more stable and smooth data distribution. Thus, this embodiment avoids the interference of measurement values ​​corresponding to individual outlier frames on the overall measurement results, improves the resistance to noise and extreme measurement errors, and enhances the stability of subsequent displacement measurement results. Furthermore, the measurement error tolerance of this embodiment can be adjusted adaptively according to different scenarios. Therefore, this embodiment can be used in scenarios where high-frequency measurements require data smoothing processing, as well as in scenarios where stable dynamic data tracking is required for medium- and low-frequency measurements, demonstrating good versatility.

[0045] Further, obtaining the filtered output value based on the trusted dataset under a preset filtering calculation mechanism includes: obtaining an integration ratio based on the total number of trusted data in the trusted dataset and the total number of historical measurements in the historical measurement value sequence; performing summation processing based on the trusted dataset to obtain a total trusted historical measurement value; and performing calculation processing based on the total trusted historical measurement value and the integration ratio to obtain the filtered output value.

[0046] The above embodiments can obtain reliable and stable filtered output values ​​by relying only on sorting and simple calculation processing. Compared with existing technologies that require complex filtering algorithms, this embodiment can avoid complex modeling and parameter fitting processes and reduce computational complexity. Therefore, this embodiment is more advantageous for application on resource-constrained equipment, can meet the real-time requirements of the equipment, and can also ensure the accuracy of the obtained displacement measurement results.

[0047] In one embodiment, the preset measurement error tolerance is set to n, so the number of values ​​to be eliminated is also n, and the historical measurement value sorting sequence {m1,m2,…,m i ,…,m g}, where g is the total number of historical measurements in the historical measurement sequence, and m i Let g be the i-th historical measurement value in the historical measurement value sorting sequence, and g > 2n; based on the removal count, the first n and last n historical measurement values ​​in the historical measurement value sorting sequence are regarded as outliers and removed. Then the reliable dataset obtained at this time is {m 1+n ,m 2+n ,…,m g-n}; and in this embodiment, the formula for calculating the filtered output value is:

[0048]

[0049] Where g-2n is the trusted dataset {m 1+n ,m 2+n ,…,m g-n The total number of credible data in} To integrate the proportions, The sum of trusted data in the trusted dataset, i.e., the sum of trusted historical measurements, O t This is the filtered output value at time t.

[0050] Further, the step of obtaining the length of the remaining measurement value sequence based on the target position measurement value sequence and the sliding window, and updating the sliding step size of the sliding window based on the remaining measurement value sequence length, includes: obtaining the length of the remaining measurement value sequence based on the target position measurement value sequence and the sliding window; if the length of the remaining measurement value sequence is not less than the length of the sliding window, then updating the sliding step size of the sliding window to a preset step size; if the length of the remaining measurement value sequence is less than the length of the sliding window, then updating the sliding step size of the sliding window to 1 frame.

[0051] The above embodiment dynamically updates the sliding window's step size by adjusting the length of the remaining measurement value sequence. This speeds up the sliding window's movement when there are many remaining measurement values ​​and reduces redundant data processing. Conversely, it slows down the sliding window's movement when there are few remaining measurement values ​​to prevent incomplete data in the obtained historical measurement value sequence due to an excessively large step size. Therefore, this embodiment improves computational efficiency, meets the real-time requirements of applications on resource-constrained devices, and ensures the reliability of the obtained filtered output values, thereby improving the stability of the displacement measurement results.

[0052] Please see Figure 2This embodiment also provides a visual displacement temporal data filtering system based on a sliding window, including: a target position measurement value acquisition module, used to acquire a sequence of consecutive multi-frame images and obtain a target position measurement value sequence based on the image sequence; a sliding window filtering module, used to acquire historical measurement value sequences sequentially through a preset sliding window at a preset sliding step size based on the target position measurement value sequence obtained by the target position measurement value acquisition module, and to execute preset filtering processing steps sequentially based on the historical measurement value sequences until the sliding window traverses the entire target position measurement value sequence to obtain several filtered output values; wherein, after each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target position measurement value sequence obtained by the target position measurement value acquisition module and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length; and a displacement measurement module, used to perform displacement state analysis processing on the motion controller based on the several filtered output values ​​obtained by the sliding window filtering module to obtain displacement measurement results.

[0053] Furthermore, in the sliding window filtering module, the filtering process includes: performing outlier removal processing based on the historical measurement value sequence to obtain a reliable dataset; and obtaining a filtered output value based on the reliable dataset under a preset filtering calculation mechanism.

[0054] The above embodiment can adjust the sliding step size of the sliding window in real time according to the length of the remaining measurement value sequence. This ensures that when the sliding window slides to the end of the target position measurement value, the obtained historical measurement value sequence will not have incomplete data. Therefore, this embodiment can handle different data sequences well and has good stability, thereby improving the stability of displacement measurement results.

[0055] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for filtering visual displacement temporal data based on a sliding window, characterized in that, Applied to motion controllers, the method includes the following steps: Acquire a sequence of multiple consecutive frames of images, and based on the image sequence, obtain a sequence of target location measurement values; Based on the target location measurement value sequence, historical measurement value sequences are sequentially acquired through a preset sliding window with a preset sliding step size. Preset filtering processing steps are then sequentially executed based on the historical measurement value sequences until the sliding window traverses the entire target location measurement value sequence, resulting in several filtered output values. After each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target location measurement value sequence and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length. Based on several of the filtered output values, the motion controller is subjected to displacement state analysis to obtain displacement measurement results. The filtering process includes: An outlier removal process is performed based on the historical measurement value sequence to obtain a reliable dataset. The integration ratio is obtained based on the total number of credible data in the credible dataset and the total number of historical measurements in the historical measurement sequence; The total sum of reliable historical measurements is obtained by performing summation processing based on the trusted dataset. The filtered output value is obtained by calculating the sum of the reliable historical measurements and the integration ratio.

2. The method for filtering visual displacement temporal data based on a sliding window according to claim 1, characterized in that, The step of obtaining the target location measurement value sequence based on the image sequence includes: Feature extraction is performed on the image sequence to obtain an edge contour sequence; Based on the edge contour sequence, an edge contour point set sequence is extracted, and a fitting process is performed on the edge contour point set sequence to obtain a center point coordinate parameter sequence. Based on the center point coordinate parameter sequence, the target position measurement value sequence is obtained.

3. The method for filtering visual displacement temporal data based on a sliding window according to claim 2, characterized in that, The step of extracting features based on the image sequence to obtain an edge contour sequence includes: Image enhancement is performed based on the image sequence to obtain an enhanced image sequence; Edge extraction is performed based on the enhanced image sequence to obtain an edge contour sequence.

4. The method for filtering visual displacement temporal data based on a sliding window according to claim 1, characterized in that, The outlier removal process based on the historical measurement sequence to obtain a reliable dataset includes: The historical measurement values ​​are sorted based on their numerical values ​​in the historical measurement value sequence to obtain a sorted historical measurement value sequence. Based on the historical measurement value sorting sequence, a rejection process is performed under a preset measurement error tolerance to obtain a reliable dataset.

5. The method for filtering visual displacement temporal data based on a sliding window according to claim 4, characterized in that, The historical measurement value sorting sequence is then processed by removing data within a preset measurement error tolerance to obtain a reliable dataset, including: Based on a preset measurement error tolerance, a rejection count is obtained; wherein, the rejection count is less than the total number of historical measurement values ​​in the historical measurement value sorting sequence; Based on the number of removals, the historical measurement value sorting sequence is processed to remove data to obtain a reliable dataset.

6. A method for filtering visual displacement temporal data based on a sliding window according to any one of claims 1 to 5, characterized in that, The step of obtaining the length of the remaining measurement value sequence based on the target location measurement value sequence and the sliding window, and updating the sliding step size of the sliding window based on the length of the remaining measurement value sequence, includes: Based on the target location measurement sequence and the sliding window, the length of the remaining measurement sequence is obtained. If the length of the remaining measurement value sequence is not less than the length of the sliding window, then the sliding step size of the sliding window is updated to a preset step size; if the length of the remaining measurement value sequence is less than the length of the sliding window, then the sliding step size of the sliding window is updated to 1 frame.

7. A visual displacement temporal data filtering system based on a sliding window, characterized in that, A method for implementing a visual displacement temporal data filtering method based on a sliding window as described in any one of claims 1 to 6 includes: The target location measurement value acquisition module is used to acquire a sequence of image values ​​for multiple consecutive frames, and obtain a sequence of target location measurement values ​​based on the image sequence; A sliding window filtering module is used to acquire historical measurement value sequences sequentially through a preset sliding window with a preset sliding step size based on the target position measurement value sequence obtained by the target position measurement value acquisition module, and to perform preset filtering processing steps sequentially based on the historical measurement value sequences until the sliding window traverses the entire target position measurement value sequence to obtain several filtered output values; wherein, after each execution of the filtering processing step, the remaining measurement value sequence length is obtained based on the target position measurement value sequence obtained by the target position measurement value acquisition module and the sliding window, and the sliding step size of the sliding window is updated based on the remaining measurement value sequence length; The displacement measurement module is used to perform displacement state analysis processing on the motion controller based on several filtered output values ​​obtained from the sliding window filtering module, and obtain displacement measurement results.

8. A visual displacement temporal data filtering system based on a sliding window according to claim 7, characterized in that, In the sliding window filtering module, the filtering process includes: An outlier removal process is performed based on the historical measurement value sequence to obtain a reliable dataset. Based on the trusted dataset, the filtered output value is obtained under a preset filtering calculation mechanism.

Images