Linear clutter filtering method based on adaptive median filtering

Abstract

The invention discloses a three-dimensional image linear clutter filtering method based on self-adaptive median filtering, belongs to the technical field of computational imaging, and aims to overcome the defect that in an existing imaging method, linear clutters possibly occur in an image due to jitter or other reasons in the imaging process, and the image reconstruction quality is affected. According to the method, the position where linear artifacts possibly exist is detected through one-way variation, then linear clutters are filtered out through a self-adaptive median filter, and finally total variation regular reconstruction is carried out, so that the linear clutters are accurately and effectively eliminated, and the reconstruction quality of an image is improved.

Description

Technical Field

[0001] This invention belongs to the field of computational imaging and relates to a linear clutter removal method based on adaptive median filtering. By using unidirectional variational methods to detect possible locations of linear artifacts, the method then uses an adaptive median filter to remove the linear clutter. Finally, a total variational regularized reconstruction is performed, effectively eliminating linear clutter and improving image reconstruction quality. Background Art

[0002] Linear clutter is widely present in ultrasonic imaging and some imaging methods that generate mechanical jitter due to scanning, and it has a significant impact on image quality. Median filtering is a nonlinear signal processing technology based on sorting statistics theory that can effectively suppress noise. Its basic principle is to replace the value of a point in a digital image or digital sequence with the median of the values of all points in a neighborhood of that point, so that the surrounding pixel values are close to the true value, thereby eliminating isolated noise points. The variational image denoising algorithm is different from the previous isotropic models such as Gaussian and mean filtering. The variational model is an anisotropic model that relies on the gradient descent method to smooth the image. By establishing a noise model, using the optimization algorithm to solve the module, and through a continuous iterative process, the restored image is infinitely close to the ideal denoised image. Summary of the Invention

[0003] A brief overview of the present invention is provided below to provide a basic understanding of certain aspects of the present invention. It should be understood that this overview is not an exhaustive overview of the present invention. It is not intended to identify key or important aspects of the present invention, nor is it intended to limit the scope of the present invention. Its purpose is simply to present certain concepts in a simplified form as a prelude to the more detailed description discussed later.

[0004] In view of this, the present invention provides a linear clutter filtering method based on adaptive median filtering. By using unidirectional variation to detect the possible locations of linear artifacts, the linear clutter is then filtered out using an adaptive median filter wave, and finally a total variation regularized reconstruction is performed, thereby accurately and effectively eliminating linear clutter and improving the reconstruction quality of the image.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] Solution 1: The present invention provides a linear clutter filtering method based on adaptive median filtering. The microscopic imaging process includes the following steps:

[0007] Step a: Obtain and input the image to be processed.

[0008] Step b: Calculate the unidirectional first-order variation of each pixel in the entire image in three dimensions.

[0009] Step c: For each pixel point, determine its variation value in three directions. If the variation in one direction is significantly greater than that in the other two directions, mark it as an abnormal pixel point.

[0010] Step d: For each abnormal pixel, a plane is selected based on the two directions with smaller variational values.

[0011] Step e: On this plane, perform median filtering with a window size of n*n on the abnormal pixel point.

[0012] Step f: Repeat steps c to d until all pixels in the image are filtered.

[0013] Step g: optimize and reconstruct the image using total variation as a regularization term.

[0014] Preferably, the input image in step a should be a three-dimensional image.

[0015] Preferably, in step b, for the functional:

[0016]

[0017] Fix the two endpoints, and denote the function when the functional S reaches its extreme value as g(x). Define a function "close" to this function, h(x) = g(x) + δg(x), where δg(x) is called the variation of the function g(x).

[0018] In practice, the change in x is usually 1 pixel.

[0019] Preferably, the process of calculating the one-way variation in step b needs to be performed once in each of the three dimensions.

[0020] Preferably, the marked abnormal points in step c should satisfy:

[0021] δg(x1)>k·δg(x2), δg(x1)>k·δg(x3)

[0022] Where k is the coefficient of the judgment threshold, δg(x1), δg(x2), δg(x3) are the variational values of different dimensions.

[0023] Preferably, the median filtering in step d only processes the abnormal pixel points selected in the previous step.

[0024] Preferably, the median filter window in step d should be two-dimensional and located in the plane formed by the directions of the two smaller variation values.

[0025] Preferably, the median filtering process in step f can be expressed as:

[0026] g(x,y)=med{f(xi,yi)},(i,j)∈S

[0027] Among them, g(x, y), f(x, y) are pixel grayscale values, and S is the template window.

[0028] Preferably, the total variation regularization model described in step g can be expressed as:

[0029]

[0030] where f is the observed noisy image and k is the reconstructed image.

[0031] Beneficial effects:

[0032] The median filter, a nonlinear signal processing technique based on sorting statistics theory that effectively suppresses noise, is the basis of this invention. Based on this, the invention incorporates an adaptive screening method using one-way variational methods, and further optimizes the image using total variational methods. By using one-way variational methods to detect possible locations of linear artifacts, an adaptive median filter is then used to remove these linear artifacts. Finally, a total variational regularized reconstruction is performed, effectively eliminating these linear artifacts and improving image reconstruction quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The figure is a workflow diagram of the linear clutter filtering method based on adaptive median filtering of the present invention. DETAILED DESCRIPTION

[0034] Exemplary embodiments of the present invention are described below with reference to the accompanying drawings. For the sake of clarity and conciseness, not all features of an actual implementation are described in this specification. However, it should be understood that in the process of developing any such actual implementation, many implementation-specific decisions must be made in order to achieve the developer's specific goals, such as meeting those constraints related to the system and business, and these constraints may vary from implementation to implementation. In addition, it should be understood that although the development work may be very complex and time-consuming, it is a routine task for those skilled in the art who benefit from the disclosure of the present invention.

[0035] It is also necessary to explain here that, in order to avoid obscuring the present invention due to unnecessary details, the accompanying drawings only show the device structure and / or processing steps closely related to the solution according to the present invention, while other details that are not closely related to the present invention are omitted.

[0036] Example 1: The present invention provides a method for filtering linear clutter based on adaptive median filtering. The microscopic imaging process includes the following steps:

[0037] Step a: Obtain and input the image to be processed.

[0038] Step b: Calculate the unidirectional first-order variation of each pixel in the entire image in three dimensions.

[0039] Step c: For each pixel point, determine its variation value in three directions. If the variation in one direction is significantly greater than that in the other two directions, mark it as an abnormal pixel point.

[0040] Step d: For each abnormal pixel, a plane is selected based on the two directions with smaller variational values.

[0041] Step e: On this plane, perform median filtering with a window size of n*n on the abnormal pixel point.

[0042] Step f: Repeat steps c to d until all pixels in the image are filtered.

[0043] Step g: optimize and reconstruct the image using total variation as a regularization term.

[0044] More specifically, the input image in step a should be a three-dimensional image.

[0045] More specifically, in step b, for the functional:

[0046]

[0047] Fix the two endpoints, and denote the function when the functional S reaches its extreme value as g(x). Define a function "close" to this function, h(x) = g(x) + δg(x), where δg(x) is called the variation of the function g(x).

[0048] In practice, the change in x is usually 1 pixel.

[0049] More specifically, the one-way variation calculation process described in step b needs to be performed once in each of the three dimensions.

[0050] More specifically, the marked outliers in step c should satisfy:

[0051] δg(x1)>k·δg(x2), δg(x1)>k·δg(x3)

[0052] Where k is the coefficient of the judgment threshold, δg(x1), δg(x2), δg(x3) are the variational values of different dimensions.

[0053] More specifically, the median filtering described in step d only processes the abnormal pixel points selected in the previous step.

[0054] More specifically, the median filter window in step d should be two-dimensional and located in the plane formed by the directions of the two smaller variation values.

[0055] More specifically, the median filtering process described in step f can be expressed as:

[0056] g(x,y)=med{f(xi,yi)},(i,j)∈S

[0057] Among them, g(x, y), f(x, y) are pixel grayscale values, and S is the template window.

[0058] More specifically, the total variation regularization model described in step g can be expressed as:

[0059]

[0060] where f is the observed noisy image and k is the reconstructed image.

Claims

1. A method for removing linear clutter from three-dimensional images based on adaptive median filtering, characterized in that: The rebuild process consists of the following steps: Step a: obtaining and inputting an image to be processed; Step b: Calculate the unidirectional first-order variation of each pixel in the entire image in three dimensions respectively; Step c: For each pixel, determine its variation value in three directions. If the variation in one direction is significantly greater than that in the other two directions, mark it as an abnormal pixel. Step d: For each abnormal pixel, a plane is selected based on the two directions with smaller variation values; Step e: On this plane, perform median filtering on the abnormal pixel with a window size of n*n; Step f: repeat steps c to d until all pixels in the image are filtered. Step g: optimize and reconstruct the image using total variation as a regularization term.

2. The method for removing linear clutter from three-dimensional images based on adaptive median filtering according to claim 1, characterized in that: The input image described in step a should be a three-dimensional image.

3. The method for removing linear clutter from three-dimensional images based on adaptive median filtering according to claim 1, wherein: In step b for the functional: Fix the two endpoints and denote the function when the functional S reaches its extreme value as g(x). Define a function "close" to this function, h(x) = g(x) + δg(x), where δg(x) is called the variation of the function g(x). In practice, the change in x is usually 1 pixel.

4. The method for removing linear clutter based on adaptive median filtering according to claim 1, wherein: The one-way variation calculation process described in step b needs to be performed once in each of the three dimensions.

5. The linear clutter filtering method based on adaptive median filtering according to claim 1, characterized in that: The marked outliers in step c should meet the following requirements: δg(x1)>k·δg(x2), δg(x1)>k·δg(x3) Where k is the coefficient of the judgment threshold, δg(x1), δg(x2), δg(x3) are the variational values of different dimensions.

6. The method for removing linear clutter based on adaptive median filtering according to claim 1, characterized in that: The median filtering described in step d only processes the abnormal pixel points selected in the previous step.

7. The method for removing linear clutter based on adaptive median filtering according to claim 1, characterized in that: The median filter window in step d should be two-dimensional and located in the plane formed by the directions of the two smaller variation values.

8. The method for removing linear clutter based on adaptive median filtering according to claim 1, characterized in that: The median filtering process described in step f can be expressed as: g(x,y)=med{f(xi,yi)},(i,j)∈S Among them, g(x, y), f(x, y) are pixel grayscale values, and S is the template window.

9. The method for removing linear clutter based on adaptive median filtering according to claim 1, characterized in that: The total variation regularization model described in step g can be expressed as: where f is the observed noisy image and k is the reconstructed image.

Images