A regional image displacement denoising method, system, computer device and medium

By performing feature point matching and random sampling dual-parameter estimation in complex dynamic scenes, a set of interior points that meet geometric constraints is selected, which solves the interference of factors such as illumination changes and texture loss on displacement monitoring and achieves high-precision and robust displacement data output.

CN120823114BActive Publication Date: 2025-11-21SHENZHEN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511341011.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-09-19
Publication Date
2025-11-21
Estimated Expiration
2045-09-19

AI Technical Summary

Technical Problem

Existing image displacement monitoring methods are easily affected by factors such as changes in illumination, occlusion, and missing textures in complex dynamic scenes, leading to the accumulation of feature matching errors and a decrease in the overall displacement vector accuracy. They also lack the adaptability of collaborative constraints on displacement direction consistency and fixed threshold mechanisms.

Method used

A regional image displacement denoising method is adopted. By extracting the initial and monitoring state feature point pairs of the target region, displacement vector decomposition and angle set generation are performed. Combined with a random sampling two-parameter estimation model, iterative calculation is carried out to select the set of interior points that meet the geometric constraints. Displacement threshold and angle threshold are used for two-parameter collaborative constraints to suppress error propagation.

Benefits of technology

It significantly improves the robustness and accuracy of displacement monitoring in complex dynamic scenarios, can adaptively suppress error propagation, and provides highly reliable displacement data output, making it suitable for high-precision scenarios such as engineering monitoring and dynamic target tracking.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120823114B_ABST
    Figure CN120823114B_ABST
Patent Text Reader

Abstract

The application provides a regional image displacement denoising method, system, computer device and medium, and belongs to the technical field of image processing and motion control. The method extracts an initial feature point pair set of a target region at an initial time and a monitoring time, decomposes the feature points in the initial feature point pair set into displacement vectors, and generates a displacement set and an angle set. A random sampling two-parameter estimation calculation model is established through random sampling, and iterative optimization is performed to screen an inlier set that meets geometric constraints, so as to extract a denoised feature point pair set, and then calculate the average displacement and average displacement direction angle of the target region. The regional image displacement denoising method, system, computer device and medium can effectively cope with displacement monitoring challenges in complex dynamic scenes, overcome the interference of factors such as light changes, occlusions and texture missing, improve the robustness and accuracy of displacement monitoring, and are suitable for various scenes requiring high-precision displacement measurement.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of image processing and motion control technology, and in particular to a method, system, computer device and medium for regional image displacement noise reduction. Background Technology

[0002] Image-based displacement monitoring technology faces numerous challenges in complex dynamic scenes. Existing methods typically rely on feature point matching to calculate the displacement of the target region. However, this method is susceptible to interference from factors such as changes in illumination, occlusion, and missing textures, leading to the accumulation of feature matching errors and interference from abnormal data.

[0003] Traditional error elimination strategies based on single parameters (such as displacement modulus) are difficult to effectively distinguish between real motion and noise, especially in non-rigid deformation or dynamic backgrounds, where the diffusion effect of single-point errors can significantly reduce the accuracy of the overall displacement vector.

[0004] Furthermore, existing methods lack collaborative constraints on the consistency of displacement direction, and the fixed threshold mechanism is insufficiently adaptable to scene changes, further limiting the robustness and accuracy of monitoring. Therefore, there is an urgent need for an efficient displacement optimization framework that can integrate multi-dimensional motion features and adaptively suppress error propagation to meet the stringent requirements for displacement data reliability in dynamic scenes. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, computer device, and medium for regional image displacement denoising, which can effectively address the challenges of displacement monitoring in complex dynamic scenes, overcome interference from factors such as changes in illumination, occlusion, and texture loss, improve the robustness and accuracy of displacement monitoring, and is applicable to various scenarios requiring high-precision displacement measurement.

[0006] To achieve the above objectives, the present invention provides a method for regional image displacement denoising, comprising the following steps:

[0007] Step S1: Extract the initial state and monitoring state of the target area at the initial time and the monitoring time respectively, and perform feature point matching to obtain the initial feature point pair set;

[0008] Step S2: Decompose the displacement vectors of the feature points in the initial feature point set to generate displacement sets and angle sets;

[0009] Step S3: Randomly sample the elements of the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the random sampling two-parameter estimation calculation model, and screen the set of interior points that satisfy the geometric constraints.

[0010] Step S4: Extract the set of noise-reduced feature point pairs based on the set of interior points;

[0011] Step S5: Calculate the average displacement and average displacement direction angle of the target region based on the set of interior points.

[0012] Preferably, in step S1, the initial set of feature point pairs is: ,in, Indicates the initial time. time One feature point, Indicates the initial time. Time The coordinates of each feature point in the image; Indicates the monitoring time Time and Corresponding feature points, Indicates the monitoring time Time The coordinates of each feature point in the image; Represents the set of positive integers.

[0013] Preferably, in step S2,

[0014] displacement set is , angle set is ,in, Indicates the first Displacement values ​​of a pair of feature points Indicates the first The angle values ​​of each feature point with respect to the displacement direction.

[0015] Preferably, step S3 includes the following steps:

[0016] Step S31: Randomly sample elements from the displacement set and angle set, with the number of samples being... Establish a random sampling two-parameter estimation calculation model;

[0017] Step S32: Filter interior points using displacement and angle thresholds:

[0018] Step S33: Repeat the iterations up to the maximum number of times, and select the optimization set with the largest number of interior points.

[0019] Preferably, in step S31, the random sampling two-parameter estimation calculation model is as follows:

[0020] ;

[0021] in, This represents the ideal displacement estimate. In this process, the mean displacement is selected as the ideal displacement estimate. Indicates the first Displacement values ​​of each sample, This represents the estimated ideal angle. In this process, the mean angle is selected as the estimated ideal angle. Indicates the first Each sampled angle value.

[0022] Preferably, in step S32, the displacement in-point condition is: the absolute difference between the displacement magnitude and the reference displacement value is less than the displacement threshold.

[0023] Interior point condition for angle: The minimum circumferential angle difference between the angle and the reference angle value is less than the angle threshold.

[0024] The present invention also provides a regional image displacement noise reduction system, comprising:

[0025] The feature extraction module is used to extract the initial state and monitoring state of the target area at the initial time and the monitoring time, respectively, and perform feature point matching to obtain an initial feature point pair set;

[0026] The displacement vector decomposition module is used to decompose the displacement vectors of feature points in the initial feature point pair set, generating displacement sets and angle sets.

[0027] The random sampling two-parameter estimation module is used to randomly sample elements from the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the calculation of the random sampling two-parameter estimation calculation model, and select the set of interior points that satisfy the geometric constraints.

[0028] The noise reduction feature point pair extraction module is used to extract the noise reduction feature point pair set based on the inlier set;

[0029] The average displacement and direction angle calculation module is used to calculate the average displacement and average displacement direction angle of the target area based on the set of interior points.

[0030] The present invention also provides a computer device, including a memory and a processor, wherein the memory is used to store instructions and the processor is used to execute the instructions to implement the region image displacement noise reduction method as described above.

[0031] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the region image displacement noise reduction method described above.

[0032] Therefore, this invention employs the aforementioned regional image displacement denoising method, system, computer equipment, and medium. By fusing a dual-parameter collaborative constraint mechanism of displacement modulus and orientation angle, combined with a random sampling iterative optimization strategy, it effectively overcomes the problems of excessive cumulative displacement monitoring errors caused by illumination changes, occlusion, texture loss, and intrusive white noise data. This significantly improves the robustness and accuracy of displacement monitoring in complex dynamic scenes. Specifically, it manifests as follows:

[0033] (1) Dual-dimensional error suppression: By utilizing the geometric consistency of displacement magnitude and direction angle, dual criteria (displacement threshold and angle threshold) are constructed, breaking through the limitations of traditional single-parameter criteria, accurately distinguishing real motion from abnormal noise, and is especially suitable for motion analysis under non-rigid deformation or dynamic background.

[0034] (2) Adaptive anti-interference capability: The iterative optimization model based on random sampling dynamically selects the set of internal points, avoiding the sensitivity of fixed thresholds to scene changes, and can adaptively suppress the spread of single-point errors and block the propagation of erroneous data;

[0035] (3) High reliability output: The feature point set after noise reduction is extracted by dual-parameter joint constraint. The final calculated average displacement and direction angle can accurately reflect the overall motion trend of the target area, providing reliable data support for high-precision displacement measurement scenarios such as engineering monitoring and dynamic target tracking. Attached Figure Description

[0036] Figure 1 This is a flowchart of a regional image displacement noise reduction method according to the present invention;

[0037] Figure 2 A schematic diagram of feature points in the monitoring area of ​​the motion platform;

[0038] Figure 3 This is a schematic diagram of the set of feature point pairs under the baseline state and the monitoring state;

[0039] Figure 4 This is a graph showing the longitudinal displacement change after noise reduction processing.

[0040] Figure 5 This is a graph showing the variation of horizontal displacement noise data. Detailed Implementation

[0041] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0043] Example 1

[0044] like Figure 1 As shown, a method for regional image displacement denoising specifically includes:

[0045] Step S1: Extract the initial state and monitoring state of the target area at the initial time and the monitoring time respectively, and perform feature point matching to obtain the initial feature point pair set. ,in, Indicates the initial time. time One feature point, Indicates the initial time. Time The coordinates of each feature point in the image; Indicates the monitoring time Time and Corresponding feature points, Indicates the monitoring time Time The coordinates of each feature point in the image; Represents the set of positive integers.

[0046] Step S2: Decompose the displacement vectors of the feature points in the initial feature point set to generate displacement sets and angle sets;

[0047] displacement set is , angle set is ,in, Indicates the first Displacement values ​​of a pair of feature points Indicates the first The angle values ​​of each feature point with respect to the displacement direction.

[0048] Step S3: Randomly sample the elements of the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the calculation of the model, and screen the set of interior points that satisfy the geometric constraints.

[0049] Step S31: Randomly sample elements from the displacement set and angle set, with the number of samples being... Establish a random sampling two-parameter estimation calculation model:

[0050] ;

[0051] in, This represents the ideal displacement estimate. In this process, the mean displacement is selected as the ideal displacement estimate. Indicates the first Displacement values ​​of each sample, This represents the estimated ideal angle. In this process, the mean angle is selected as the estimated ideal angle. Indicates the first Each sampled angle value.

[0052] Step S32: Filter interior points using displacement and angle thresholds:

[0053] Interior point condition for displacement: The absolute difference between the magnitude of the displacement and the reference displacement value is less than the displacement threshold; Interior point condition for angle: The minimum circumferential angle difference between the angle and the reference angle value is less than the angle threshold, as detailed below:

[0054] ;

[0055] in, Indicates the first Displacement values ​​of each feature point pair With displacement estimate The absolute value of the difference Indicates the first n Angle values ​​of each feature point pair With angle estimate The minimum difference, The threshold representing the displacement amplitude. The threshold value representing the angle.

[0056] Step S33: Repeat the iterations to the maximum number of times, and select the optimization set with the largest number of interior points as the interior point set, including the displacement set. With angle set .

[0057] Step S4: Process the obtained displacement set With angle set The elements in the original displacement set are arranged according to the original displacement set. With angle set Match the corresponding sequence number and extract the data from the set according to the index number. The corresponding feature point pairs are used to obtain the denoised feature point pair set. ;

[0058] Step S5: For the displacement set With angle set The mean of each element in the matrix is ​​calculated to obtain the current time. The average displacement of the corresponding target area Its direction angle is .

[0059] The experiment was conducted using a vertically reciprocating platform as the monitoring object, with the entire platform selected as the monitoring area.

[0060] First, video image data from the motion platform is acquired. The initial image frames are set as the baseline state, and each frame during the motion process is used as the monitoring state. Feature extraction algorithms (including SIFT, SURF, ORB, BRIEF, etc.) are used to obtain feature points of the monitoring area in both the baseline and each monitoring state, thereby obtaining feature points for different states (e.g., ...). Figure 2 (As shown).

[0061] Next, feature points in the monitoring area under the baseline state and the monitoring state are matched to obtain a set of feature point pairs (e.g., ...). Figure 3(As shown). Subsequently, the method proposed in this invention is applied to denoise the set of feature points, and the denoised longitudinal and lateral displacements are calculated in real time. Figure 4 The longitudinal displacement curve after noise reduction is shown, reflecting the displacement monitoring results of the target area under uniform longitudinal motion. This method effectively suppresses error accumulation, and the displacement curve conforms to the uniform motion trend, demonstrating the accuracy and robustness of the algorithm in the vertical direction. Figure 5 The diagram shows the displacement error variation curve in the horizontal direction. Although the target does not actually move horizontally, the error curve in the horizontal direction can be used to evaluate the stability of the algorithm due to the randomness of image noise in both the horizontal and vertical directions. Figure 5 As shown, in an 800-frame image sequence, the cumulative horizontal displacement error is controlled within approximately 0.1 pixels, indicating that the method has good anti-interference ability and stability, and can effectively suppress the error spread caused by noise.

[0062] Example 2

[0063] A region image displacement denoising system, comprising:

[0064] The feature extraction module is used to extract the initial state and monitoring state of the target area at the initial time and the monitoring time, respectively, and perform feature point matching to obtain an initial feature point pair set;

[0065] The displacement vector decomposition module is used to decompose the displacement vectors of feature points in the initial feature point pair set, generating displacement sets and angle sets.

[0066] The random sampling two-parameter estimation module is used to randomly sample elements from the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the calculation of the model, and select the set of interior points that satisfy the geometric constraints.

[0067] The noise reduction feature point pair extraction module is used to extract the noise reduction feature point pair set based on the inlier set;

[0068] The average displacement and direction angle calculation module is used to calculate the average displacement and average displacement direction angle of the target area based on the set of interior points.

[0069] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0070] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0071] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0072] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0073] Therefore, the present invention employs the above-mentioned method, system, computer equipment and medium for regional image displacement denoising, which can effectively address the challenges of displacement monitoring in complex dynamic scenes, overcome interference from factors such as changes in illumination, occlusion, and texture loss, improve the robustness and accuracy of displacement monitoring, and is applicable to various scenarios requiring high-precision displacement measurement.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for regional image displacement denoising, characterized in that, Includes the following steps: Step S1: Extract the initial state and monitoring state of the target area at the initial time and the monitoring time respectively, and perform feature point matching to obtain the initial feature point pair set; Step S2: Decompose the displacement vectors of the feature points in the initial feature point set to generate displacement sets and angle sets; Step S3: Randomly sample the elements of the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the random sampling two-parameter estimation calculation model, and screen the set of interior points that satisfy the geometric constraints. Step S4: Extract the set of noise-reduced feature point pairs based on the set of interior points; Step S5: Calculate the average displacement and average displacement direction angle of the target region based on the set of interior points; In step S2, the displacement set is , angle set is ,in, Indicates the first Displacement values ​​of a pair of feature points Indicates the first The angle value of each feature point with respect to the displacement direction Indicates the initial time. Time The coordinates of a feature point in the image. Indicates the monitoring time Time The coordinates of each feature point in the image; Step S3 includes the following steps: Step S31: Randomly sample elements from the displacement set and angle set, with the number of samples being... Establish a random sampling two-parameter estimation calculation model; Step S32: Filter interior points using displacement and angle thresholds: Step S33: Repeat the iterations up to the maximum number of times, and select the optimization set with the largest number of interior points.

2. The method for regional image displacement denoising according to claim 1, characterized in that, In step S1, the initial set of feature point pairs is: ,in, Indicates the initial time. time One feature point, Indicates the initial time. Time The coordinates of each feature point in the image; Indicates the monitoring time Time and Corresponding feature points, Indicates the monitoring time Time The coordinates of each feature point in the image; Represents the set of positive integers.

3. The method for regional image displacement denoising according to claim 2, characterized in that, In step S31, the random sampling two-parameter estimation calculation model is as follows: ; in, This represents the estimated ideal displacement. Indicates the first Displacement values ​​of each sample, This represents the estimated value of the ideal angle. Indicates the first Each sampled angle value.

4. The method for regional image displacement denoising according to claim 1, characterized in that, In step S32, the displacement interior point condition is: the absolute difference between the displacement magnitude and the reference displacement value is less than the displacement threshold. Interior point condition for angle: The minimum circumferential angle difference between the angle and the reference angle value is less than the angle threshold.

5. A regional image displacement noise reduction system, characterized in that, A method for performing region image displacement denoising as described in any one of claims 1-4, comprising: The feature extraction module is used to extract the initial state and monitoring state of the target area at the initial time and the monitoring time, respectively, and perform feature point matching to obtain an initial feature point pair set; The displacement vector decomposition module is used to decompose the displacement vectors of feature points in the initial feature point pair set, generating displacement sets and angle sets. The random sampling two-parameter estimation module is used to randomly sample elements from the displacement set and angle set, establish a random sampling two-parameter estimation calculation model, repeatedly iterate the calculation of the random sampling two-parameter estimation calculation model, and select the set of interior points that satisfy the geometric constraints. The noise reduction feature point pair extraction module is used to extract the noise reduction feature point pair set based on the interior point set; The average displacement and direction angle calculation module is used to calculate the average displacement and average displacement direction angle of the target area based on the set of interior points.

6. A computer device, characterized in that, It includes a memory and a processor, the memory being used to store instructions, and the processor being used to execute the instructions to implement the region image displacement noise reduction method as described in any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the region image displacement noise reduction method as described in any one of claims 1-4.

Citation Information

Patent Citations

  • Medical X-ray video sequence image noise reduction method and system and readable storage medium

    CN115526798A

  • Noise reduction model training method and related apparatus

    WO2022134971A1