An image background noise suppression method based on local contrast anisotropic filtering

Abstract

This invention provides an image background noise suppression method based on anisotropic filtering with local contrast, belonging to the field of image processing technology. The method includes: Step S1, calculating the average gray value of pixels in a central neighborhood within a first specific range centered on the current pixel, obtaining the average gray value of the central neighborhood; Step S2, calculating the average gray value of background pixels in eight surrounding background neighborhoods within a second specific range centered on the current pixel, obtaining the average gray value of each of the eight background neighborhoods; Step S3, calculating the differences between the average gray value of the eight background neighborhoods and the average gray value of the central neighborhood in four directions, based on the average gray value of the central neighborhood and the average gray value of the eight background neighborhoods; Step S4, determining the gray value of the current pixel based on the differences in the four directions. This invention provides stronger local signal-to-noise ratio gain, clearer target boundaries, lower average background brightness, and faster calculation speed, enabling it to respond to high frame rate processing requirements.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and more specifically, relates to an image background noise suppression method based on local contrast-based anisotropic filtering. Background Technology

[0002] Current weak target detection algorithms can be divided into multi-frame and single-frame methods. Multi-frame detection mainly utilizes the target's motion trajectory and correlation to detect weak targets, requiring more additional information and having higher computational complexity. Single-frame weak target detection methods mainly achieve detection through background noise suppression or target enhancement. Compared to multi-frame detection, they have the advantages of lower complexity, ease of implementation on engineering hardware platforms, and can be combined with other detection methods, making them more widely applicable. Background noise suppression and target enhancement mainly aim to improve the local signal-to-noise ratio of the target by suppressing background noise and enhancing target energy, providing a good foundation for subsequent target segmentation and extraction. Effective background noise suppression and target enhancement algorithms can improve the target detection rate, reduce the false alarm rate, and make target extraction more stable.

[0003] Most background noise suppression methods can be broadly categorized into spatial domain and transform domain approaches. Transform domain background modeling algorithms convert the original image to the transform domain, treating the slowly changing, uniform background as a low-frequency component and utilizing the target's high-frequency position within the transform domain. After processing, the target is highlighted, thus enabling the detection of small targets. Gaussian high-pass filters and ideal Gaussian high-pass filters are classic frequency domain background modeling algorithms, but both suffer from ringing artifacts. Spatial domain background noise suppression utilizes the difference in grayscale values ​​between the background and the target in the spatial domain to obtain a predicted background for the target. The predicted background image is then subtracted from the original image to suppress the background noise. Methods such as mean filtering and top-hat transformation are effective for background noise suppression in scenes with slow changes and small spans, but they suffer from edge blurring in scenes with abrupt changes. On the other hand, methods based on multi-feature fusion and spectral residuals, which utilize human visual characteristics, perform well in complex backgrounds, but are computationally slow and have poor real-time performance. Methods using deep learning for background noise suppression rely on large datasets and are highly complex. In summary, current background noise suppression methods cannot quickly and effectively suppress the background of images of small targets, thus creating obstacles in application scenarios where high frame rate cameras are used to detect small targets. Summary of the Invention

[0004] The purpose of this invention is to provide an image background noise suppression method based on local contrast anisotropic filtering, which solves the problem of target detection difficulties caused by insufficient background noise suppression in application scenarios using real-time high frame rate cameras to detect weak targets.

[0005] To achieve the above objectives, the present invention provides an image background noise suppression method based on local contrast anisotropic filtering. The method includes: step S1, calculating the average gray value of pixels in a central neighborhood of a first specific range centered on the current pixel, to obtain the average gray value of the central neighborhood; step S2, calculating the average gray value of background neighborhoods in eight surrounding second specific ranges centered on the current pixel, to obtain the average gray value of the eight background neighborhoods respectively; step S3, calculating the difference between the average gray value of the eight background neighborhoods and the average gray value of the central neighborhood in four directions based on the average gray value of the central neighborhood and the average gray value of the eight background neighborhoods respectively; step S4, determining the gray value of the current pixel based on the difference in the four directions.

[0006] The beneficial effects of this invention are as follows:

[0007] (1) It can quickly and effectively suppress the background of weak target images;

[0008] (2) It has a stronger signal-to-noise ratio gain for weak targets;

[0009] (3) Makes the target boundary clearer and the average brightness of the background lower. Attached Figure Description

[0010] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0011] Figure 1 This is a flowchart of the image background noise suppression method based on local contrast anisotropic filtering provided by the present invention;

[0012] Figure 2 This invention provides a preprocessed image and its grayscale statistical cube diagram.

[0013] Figure 3 This is an image and its statistical cubic diagram showing the effect of using the image background noise suppression method provided by this invention.

[0014] Figure 4 This is a pixel neighborhood position relationship diagram provided by the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other. To achieve the above objectives, this invention adopts the following technical solution.

[0016] Figure 1 This is a flowchart of the image background noise suppression method based on local contrast anisotropic filtering provided by the present invention, as shown below. Figure 1 As shown, the method includes:

[0017] Step S1: Calculate the average gray value of the pixels in the central neighborhood of the first specific range centered on the current pixel to obtain the average gray value of the central neighborhood.

[0018] Step S2: Calculate the average gray value of the eight background neighborhoods surrounding the current pixel in a second specific range, and obtain the average gray value of the eight background neighborhoods respectively.

[0019] Step S3: Based on the average gray value of the central neighborhood and the average gray values ​​of the eight background neighborhoods, calculate the difference between the average gray values ​​of the eight background neighborhoods and the average gray value of the central neighborhood in four directions.

[0020] Step S4: Determine the grayscale value of the current pixel based on the differences in the four directions.

[0021] Step S4 includes: sorting the differences in the four directions, and taking the minimum value among the differences that are greater than zero as the gray value of the current pixel.

[0022] Step S4 further includes: if the difference in all four directions is less than zero, then set the grayscale value of the current pixel to 0.

[0023] The image background noise suppression method based on local contrast anisotropic filtering provided by this invention further includes: step S5, performing steps S1 to S4 on all pixels in the image to obtain the processed enhanced grayscale values ​​of all pixels in the image, thereby obtaining the enhanced image. More specifically, steps S1 to S4 are repeated on all pixels in the image that can be effectively calculated, until all pixels in the image that can be effectively calculated are processed and the processed enhanced grayscale values ​​are obtained, thereby obtaining the enhanced image.

[0024] In step S2, the eight background neighborhoods of the second specific range are eight. Background neighborhood.

[0025] In step S1, the central neighborhood of the first specific range is Central neighborhood.

[0026] Specifically, based on the characteristics of weak targets, eight surrounding targets can be selected. Background neighborhood and The central neighborhood. It should be understood that those skilled in the art can adjust the values ​​of the first specific range (i.e., the range of the central neighborhood) and the second specific range (i.e., the range of the background neighborhood) according to the actual situation, although the present invention adopts Central Neighborhood and The background neighborhood is included, but the scope of protection of this invention is not limited to this. The central neighborhood of any first specific range and the background neighborhood of any second specific range are both within the scope of protection of this invention.

[0027] In step S1, the average gray value of the central neighborhood is calculated as follows:

[0028] ;

[0029] in, The mean gray value of the central neighborhood. This represents the current pixel coordinates. In pixel coordinates The pixel grayscale value below , All are integers.

[0030] In step S2, the average grayscale value of the eight background neighborhoods is calculated as follows:

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] ;

[0039] in, The average grayscale value of the top-left background neighborhood centered on the current pixel. The average grayscale value of the background neighborhood above the current pixel. The average grayscale value of the upper right background neighborhood centered on the current pixel. The mean gray value of the left-side background neighborhood centered on the current pixel. The mean gray value of the background neighborhood to the right of the current pixel. The mean grayscale value of the bottom-left background neighborhood centered on the current pixel. The mean grayscale value of the background neighborhood centered at the current pixel. The mean grayscale value of the bottom right background neighborhood centered on the current pixel. In pixel coordinates The pixel grayscale value below , All are integers.

[0040] The differences in the four directions include: the difference in the first direction, based on the average grayscale value of the upper-left background neighborhood centered on the current pixel. Compared with the average grayscale value of the bottom right background neighborhood centered on the current pixel The calculated second-direction difference is based on the average grayscale value of the background neighborhood above the current pixel. Compared to the average grayscale value of the background neighborhood centered on the current pixel The calculated third-direction difference is based on the average grayscale value of the left-side background neighborhood centered on the current pixel. Compared with the average grayscale value of the background neighborhood to the right centered on the current pixel The fourth-direction difference is calculated based on the average grayscale value of the upper-right background neighborhood centered on the current pixel. Compared with the average grayscale value of the lower left background neighborhood centered on the current pixel Calculated.

[0041] In step S3, the differences between the average gray values ​​of the eight background neighborhoods and the average gray value of the center neighborhood in four directions are calculated as follows:

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] in, Indicates the difference in the first direction. Indicates the difference in the second direction. Indicates the difference between the three directions. This represents the difference in the fourth direction. Furthermore, This indicates the direction formed by the center of the upper-left background neighborhood and the center of the lower-right background neighborhood selected with the current pixel as the center point; This indicates the direction formed by the center of the upper background neighborhood and the center of the lower background neighborhood selected with the current pixel as the midpoint; This indicates the direction formed by the center of the left background neighborhood and the center of the right background neighborhood selected with the current pixel as the midpoint; This indicates the direction formed by the center of the lower left background neighborhood and the center of the upper right background neighborhood, with the current pixel as the midpoint.

[0047] The present invention will be further illustrated below through embodiments. These embodiments use… Taking a single frame of a weak target image as an example, the original image and its grayscale statistical cube are as follows: Figure 2 As shown, Figure 2 This invention provides a preprocessed image and its grayscale statistical cubic diagram. From... Figure 2 As can be seen, the noise variance in the background region of the image is significantly higher than the signal fluctuation range. Small targets are obscured by background noise, lacking both obvious contour features and stable gray-level peaks, thus interfering with stable target extraction. Therefore, it is necessary to analyze target characteristics and suppress background noise. The image of a small target is very small and can be considered as a circle with Gaussian distribution characteristics and isotropic properties. The disturbed small target differs significantly from the background in at least two directions. Based on the imaging characteristics of small targets, this invention utilizes the difference between the mean gray-level value of the small target's local neighborhood and the mean gray-level value of the background neighborhood to provide a method for rapidly enhancing targets and suppressing background noise. The effect diagram and statistical cubic diagram of the technical solution provided by this invention are shown below. Figure 3 As shown, Figure 3 This is an image and its statistical cubic diagram showing the effect of using the image background noise suppression method provided by this invention.

[0048] The following is combined Figure 4 To describe the invention in more detail, Figure 4 This invention provides a pixel neighborhood position relationship diagram, and the specific implementation steps are as follows.

[0049] In step S1, the calculation is performed with the current pixel... Centered The average grayscale value of the neighboring pixels.

[0050] This embodiment uses Let's take a single frame of a small target image as an example. Assume the position of the first pixel in the upper left corner is... Since it is necessary to select the background neighborhood, the first pixel that can be effectively calculated is located first. , calculation Centered The mean gray value of the central neighborhood. central neighborhood, such as Figure 4 As shown in Figure 401.

[0051] In step S2, the calculation is performed with the current pixel... Eight around the center The average gray value of the background neighborhood, that is, the average gray value of the eight background neighborhoods. Figure 4 For example, the calculation is based on Centered The surrounding eight The average gray value of the background neighborhood, that is, the average gray value of the eight background neighborhoods, and the relative positions of the eight background neighborhoods are as follows: Figure 4 As shown in Figure 402.

[0052] In step S3, the differences between the mean gray values ​​of the eight background neighborhoods and the mean gray value of the central neighborhood are calculated in four directions. Based on the characteristic that weak targets have significant differences from the background in at least two directions, a local contrast saliency map is constructed in four directions. , , , like Figure 4 As shown in Figure 403.

[0053] In step S4, the differences in the four directions are sorted, and the minimum value among the differences greater than zero is taken as the grayscale value of the current pixel. If the differences in all four directions are less than zero, the grayscale value of the current pixel is set to 0. Through step S4, the grayscale value of the current pixel after filtering by the technical solution provided by this invention can be calculated.

[0054] In step S5, steps S1 to S4 are repeated for all pixels that can be effectively calculated until all pixels obtain the processed enhanced grayscale value, thus obtaining the enhanced image.

[0055] Subsequent pixels are processed through steps S1 to S4. Since the background area needs to be selected, the number of pixels that can be effectively calculated can be traversed up to the last pixel. This yields the image after background noise suppression using this method, and its effect and grayscale statistics are shown below. Figure 3 As shown, from Figure 3 As can be seen, the background noise has been effectively suppressed, the target is visible and the outline is clear. The statistical chart shows obvious gray-scale peaks of the target, and further normalization processing can be performed as needed.

[0056] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0057] This invention addresses the problem that weak targets in images are obscured by background noise, interfering with the stable extraction of targets. By leveraging the imaging characteristics of weak targets and utilizing the difference between the mean of the local neighborhood of the weak target and the background region, a method is provided to rapidly enhance the target and suppress background noise.

[0058] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not describe the various possible combinations separately.

[0059] Furthermore, various different implementations of the present invention can be combined arbitrarily, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed in the present invention.

Claims

1. A method of image background noise suppression based on local contrast- based anisotropic filtering, characterized in that, The method includes: Step S1: Calculate the average gray value of the pixels in the central neighborhood of the first specific range centered on the current pixel to obtain the average gray value of the central neighborhood. Step S2: Calculate the average gray value of the eight background neighborhoods surrounding the current pixel in a second specific range, and obtain the average gray value of the eight background neighborhoods respectively. Step S3: Based on the mean gray value of the central neighborhood and the mean gray values ​​of the eight background neighborhoods, calculate the difference between the mean gray values ​​of the eight background neighborhoods and the mean gray value of the central neighborhood in four directions. Step S4: Determine the grayscale value of the current pixel based on the differences in the four directions.

2. The image background noise suppression method based on local contrast-based anisotropic filtering according to claim 1, characterized in that, Step S4 includes: Sort the differences in the four directions, and take the minimum value among the differences that are greater than zero as the gray value of the current pixel.

3. The method of image background noise suppression based on local contrast- dependent anisotropic filtering according to claim 2, characterized in that, Step S4 further includes: If the difference in all four directions is less than zero, then the grayscale value of the current pixel is set to 0.

4. The method of image background noise suppression based on local contrast- dependent anisotropic filtering according to claim 3, characterized in that, The method also includes: Step S5: Perform steps S1 to S4 on all pixels in the image to obtain the processed enhanced grayscale values ​​of all pixels in the image, thereby obtaining the enhanced image.

5. The image background noise suppression method based on local contrast anisotropic filtering according to claim 4, characterized in that, In the step S2, the eight second specific ranges of background neighborhoods are eight background neighborhoods.

6. The image background noise suppression method based on local contrast anisotropic filtering according to claim 5, characterized in that, In said step S1, the center neighborhood of said first specific range is Center neighborhood.

7. The method of image background noise suppression based on local contrast- dependent anisotropic filtering according to claim 6, characterized in that, In step S1, the average gray value of the central neighborhood is calculated as follows: ； wherein, is the central neighborhood gray mean, is the current pixel coordinate value, is the pixel gray value at pixel coordinate , , wherein, is the central neighborhood gray mean, is the current pixel coordinate value, is the pixel gray value at pixel coordinate , , wherein, is the central neighborhood gray mean, is the current pixel coordinate value 8. The image background noise suppression method based on local contrast anisotropic filtering according to claim 7, characterized in that, In step S2, the average grayscale value of the eight background neighborhoods is calculated as follows: ； ； ； ； ； ； ； ； in, The average grayscale value of the top-left background neighborhood centered on the current pixel. The average grayscale value of the background neighborhood above the current pixel. The average grayscale value of the upper right background neighborhood centered on the current pixel. The mean gray value of the left-side background neighborhood centered on the current pixel. The mean gray value of the background neighborhood to the right of the current pixel. The mean grayscale value of the bottom-left background neighborhood centered on the current pixel. The mean grayscale value of the background neighborhood centered at the current pixel. The mean grayscale value of the bottom right background neighborhood centered on the current pixel. In pixel coordinates The pixel grayscale value below , All are integers.

9. The image background noise suppression method based on local contrast anisotropic filtering according to claim 8, characterized in that, The differences in the four directions include: The first directional difference is based on the average grayscale value of the upper-left background neighborhood centered on the current pixel. Compared with the average grayscale value of the bottom right background neighborhood centered on the current pixel Calculated; The second directional difference is based on the average grayscale value of the background neighborhood above the current pixel. Compared to the average grayscale value of the background neighborhood centered on the current pixel Calculated; The third directional difference is based on the average grayscale value of the left-side background neighborhood centered on the current pixel. Compared with the average grayscale value of the background neighborhood to the right centered on the current pixel Calculated; The fourth directional difference is based on the average grayscale value of the upper right background neighborhood centered on the current pixel. Compared with the average grayscale value of the lower left background neighborhood centered on the current pixel Calculated.

10. The image background noise suppression method based on local contrast anisotropic filtering according to claim 9, characterized in that, In step S3, the differences between the average gray values ​​of the eight background neighborhoods and the average gray value of the center neighborhood in four directions are calculated as follows: ； ； ； ； in, Indicates the difference in the first direction. Indicates the difference in the second direction. Indicates the difference between the three directions. This represents the difference in the fourth direction.

Images