SAR image strong scattering point elimination method based on image threshold segmentation

Abstract

An SAR image strong scatter elimination method based on image threshold segmentation analyzes a pixel value relation between strong scatter points and a background area through an image threshold segmentation method, realizes effective elimination of the strong scatter points, and comprises the following steps: firstly, carrying out quantization and downsampling processing on an image, then carrying out uniform windowing processing on the image, and finally, carrying out high-resolution processing on the image. The method comprises the following steps: firstly, obtaining a window, then calculating an average value and a standard deviation in the window, judging the relationship between the average value and the standard deviation of a small area, marking strong scattering points, then performing expansion corrosion operation on a binary image, and finally performing logic AND operation on the binary image and an original image, thereby marking the strong scattering points in the image. According to the method, the strong scattering points in the SAR image can be effectively removed, and the wind wave inversion precision is effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a method for eliminating strong scattering points of SAR images based on image threshold segmentation. Background Art

[0002] Synthetic Aperture Radar (SAR) images are widely used in fields such as ocean wind and wave inversion. However, strong scattering points in SAR images, such as ships, islands, and special ocean phenomena, can interfere with image analysis and processing. For example, in wind and wave inversion, the wind speed inversion result is too large because the backscattering coefficient in the strong scattering point area is too large. Existing technologies are difficult to effectively remove these strong scattering points, which affects the image quality and application effect.

[0003] At present, many strong scatter removal methods have been proposed. The research team of Tsinghua University proposed to identify the parameters of AR(1) model through Chirp-Z transform, and then use linear phase equiripple notch filter to filter out strong scattering signals. This method has advantages in maintaining image resolution, statistical characteristics and polarization scattering characteristics, and its effectiveness has been verified by simulation and measured data processing. The research team of Xiangtan University proposed a deep learning network for integrated speckle removal and super-resolution in multi-temporal SAR (ISSMSAR). The network improves the image restoration quality by fusing multi-temporal features, and outperforms the speckle removal and super-resolution methods based on single tasks in terms of subjective perception and objective evaluation indicators.

[0004] However, high-precision strong scatter point methods often require complex algorithms as support, and have poor processing timeliness. At the same time, the background of SAR images is complex, and commonly used algorithms are difficult to effectively guarantee the dual optimization of strong scatter point segmentation accuracy and processing efficiency, and cannot be effectively deployed on the terminal processing platform. Therefore, how to achieve low-complexity and efficient removal of strong scatter points is an urgent problem to be solved in the field of SAR image strong scatter point removal technology. Summary of the invention

[0005] In order to address the deficiencies in the prior art, the present invention provides a method for removing strong scattering points in SAR images based on image threshold segmentation. By analyzing the relationship between the pixel values ​​of strong scattering points and the background area through image threshold segmentation, the strong scattering points can be effectively removed, thereby effectively improving the accuracy of wind and wave inversion.

[0006] The present invention provides a method for removing strong scattered points from SAR images based on image threshold segmentation, comprising the following steps:

[0007] Step (1), image quantization. Convert the original SAR image to an 8-bit image for subsequent processing;

[0008] Step (2), image sampling. Downsample the image processed in step (1) to reduce the number of pixels by sampling at alternate points to speed up the processing;

[0009] Step (3), sliding window processing. Use a sliding window on the image downsampled in step (2) to perform uniform sliding window processing on the entire image;

[0010] Step (4), threshold determination. Calculate the mean and standard deviation of the pixels in the window, traverse each small area, determine whether its mean is greater than the window mean plus N times the standard deviation, mark the strong scattering points, and obtain a binary image of the strong scattering points;

[0011] Step (5), dilation and corrosion operation. Dilation and corrosion operation is performed based on the binary image obtained in step (4) to remove noise in the image;

[0012] Step (6), mapping the original image. Use the image obtained in step (5) to perform a logical AND operation with the original image to mark the strong scattering points in the original image.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) by judging the threshold value of pixels in the image area, strong scattering points can be effectively eliminated quickly; (2) by downsampling, the number of pixels is reduced, thereby accelerating the processing speed; (3) by performing expansion and corrosion operations, noise in the image is removed; (4) the data is quantized from 32 bits to 8 bits to improve the processing efficiency, and a data stretching enhancement module is added to enhance the contrast between strong scattering points and the sea background; (5) the method has a simple flow and is easy to implement, and is convenient for engineering deployment, and can effectively eliminate strong scattering points in SAR images on the terminal processing platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The above and other objects, features and advantages of the present disclosure will become more apparent through a more detailed description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present disclosure.

[0015] Figure 1 is a flow chart of an exemplary embodiment according to the present disclosure;

[0016] Figure 2 This is an example of inversion results without removing strong scattering points;

[0017] Figure 3 This is an example of inversion results after removing strong scattering points. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0019] The present invention provides a method for removing strong scattering points in SAR images based on image threshold segmentation. First, the image is quantized and downsampled, and then the image is uniformly windowed. After that, the average value and standard deviation in the window are calculated, and the relationship between the mean value of a small area and the average value and the standard deviation is judged to mark the strong scattering points. Then, the binary image is expanded and eroded, and finally the binary image is logically ANDed with the original image, so as to mark the strong scattering points in the image.

[0020] Figure 1 A flow chart of an exemplary embodiment according to the present disclosure is shown, comprising the following steps:

[0021] Step (1), image quantization: convert the original image into an 8-bit image. The original imaging data is 32-bit floating point type, and the effective data width of image processing is 8 bits. In order to improve the processing efficiency, the data is quantized from 32 bits to 8 bits, and a data stretching enhancement module is added to enhance the contrast between the strong scattered points and the sea background;

[0022] Step (2), image sampling: downsample the image, that is, reduce the number of pixels in the image. This is achieved by sampling at alternate points, retaining only one pixel (maximum value) in each 4*4 pixel block, so that the width and height of the image are reduced to 1 / 4 of the original. This method can further reduce the amount of data and speed up processing;

[0023] Step (3), sliding window: Apply a 256x256 sliding window to the downsampled image. This window moves from left to right and from top to bottom on the image, one block at a time, so that the entire image is processed evenly;

[0024] Step (4), threshold determination: first calculate the mean and standard deviation of the image, then for each 256x256 window, calculate the mean and standard deviation of the pixels in the window, and calculate the relationship between the mean and standard deviation of the window and the mean and standard deviation of the image to determine which area the window belongs to (large bright area, partial bright area, small bright area). Then, traverse each 4x4 small area in the window to determine whether the average value of each small area is greater than the average value of the window plus n times the standard deviation. If the condition is met, the pixels in the small area are marked as strong scattering points, where strong scattering points are marked as 1 and the rest are marked as 0. Based on the above operations, a binary image is obtained.

[0025] Step (4.1), determine the window attributes: determine which area the 256*256 window belongs to (large bright area, partial bright area, small bright area) through the relationship between the 256*256 window and the image mean standard deviation. If the mean of the 256*256 window is greater than 1.2 times the image mean or the standard deviation of the 256*256 window is greater than 1.2 times the image standard deviation, then the 256*256 window is determined to be a large bright area. If the mean of the 256*256 window is greater than 0.7 times the image mean or the standard deviation of the 256*256 window is greater than 0.7 times the image standard deviation, then the 256*256 window is determined to be a partial bright area. If none of the above conditions are met, the 256*256 window is determined to be a small bright area.

[0026]

[0027] Step (4.2), determine the pixel attributes: for the 256*256 window, traverse each 4x4 small area in the window, and determine whether the mean of the 4x4 small area is greater than or equal to the 256*256 window mean + N times the standard deviation. If it is satisfied, the 4x4 small area is marked as 1 (strong scattering point area), otherwise it is marked as 0 (non-strong scattering point area), where n is set to different values ​​according to different scenarios. The specific judgment method is as follows:

[0028]

[0029] Step (5), dilation and erosion operation: Based on the binary image obtained in step (4), dilation and erosion operation is performed. First, a 7*7 kernel is used to perform a dilation operation to merge adjacent foreground pixels into a larger connected area, and then a 3*3 kernel is used to perform an erosion operation to remove noise in the image;

[0030] Step (6), mapping the original image: perform a logical AND operation on the image after dilation and erosion and the original image, and only retain the pixels that are also marked as strong scattering points in the marked image. Then, according to the position information of the window, the positions of these strong scattering points are mapped back to the original image, so that the strong scattering points are marked in the original image.

[0031] Application Examples

[0032] 1. Experimental conditions:

[0033] The hardware platform of the experiment is: the processor is Intel(R) Core(TM) i7-1165G7, the main frequency is 2.8GHz, and the memory is 16GB

[0034] The software platform of the experiment is: Win 10.

[0035] 2. Experimental content and results analysis:

[0036] Strong scattering points will cause the scattering coefficient of some locations in the imaging area to be too large, and the wind speed inversion result will also be too large, resulting in an increase in the inversion error. Figure 2 Figure 3 The results before and after image removal are used as a comparison to illustrate the improvement of the inversion effect by removal. According to experimental statistics, using the disclosed method to remove strong scattering points from SAR images, the wind speed error is reduced by 0.4RMSE and the wind direction error is reduced by 0.2RMSE.

[0037] The above technical scheme is only an exemplary embodiment of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the application methods and principles disclosed in the present invention, and it is not limited to the method described in the above specific embodiment of the present invention. Therefore, the method described above is only preferred and does not have a restrictive meaning.

Claims

1. A method for removing strong scattering points from SAR images based on image threshold segmentation, comprising the following steps: S1, image quantization: convert the original SAR image into an 8-bit image and enhance the contrast between the strong scattered points and the sea background; S2, image sampling: downsampling the quantized image to reduce the number of pixels; S3, sliding window processing: use a sliding window on the downsampled image and perform uniform sliding window processing on the entire image; S4, threshold judgment: calculate the average value and standard deviation of the pixels in the window, traverse each small area, judge whether its average value is greater than the window average value plus N times the standard deviation, mark the strong scattering points, and obtain the binary image of the strong scattering points; S5, dilation and corrosion operation: performing dilation and corrosion operation based on the binary image obtained in step S4 to remove noise in the image; S6, mapping the original image: performing a logical AND operation on the image obtained in step S5 and the original image, and marking the strong scattering points in the original image.

2. The method according to claim 1, characterized in that The step S1 specifically includes: The original imaging data is quantized from 32-bit floating point to 8-bit, and data stretching enhancement processing is added to enhance the contrast between strong scattered points and the sea background.

3. The method according to claim 1 or 2, characterized in that: The step S2 specifically includes: By sampling alternate points, only one pixel is retained in each 4*4 pixel block, so that the width and height of the image are reduced to 1 / 4 of the original.

4. The method according to claim 1 or 2, characterized in that: The step S3 specifically includes: A 256x256 sliding window is applied to the downsampled image. This window moves from left to right and from top to bottom across the image, one block at a time, so that the entire image is processed evenly.

5. The method according to claim 1 or 2, characterized in that: The step S4 specifically includes: S41, calculating the mean and standard deviation of the image, then for each 256x256 window, calculating the mean and standard deviation of the pixels in the window, and determining which region the window belongs to based on the relationship between the mean and standard deviation of the window and the mean and standard deviation of the image, where the types of regions include: large bright region, partial bright region, and small bright region; S42, traverse each 4x4 small area in the window, and determine whether the average value of each small area is greater than the average value of the 256*256 window plus n times the standard deviation; If the condition is met, the pixels in the small area are marked as strong scattering points, where the strong scattering points are marked as 1 and the other points are marked as 0. Based on the above operations, a binary image is obtained.

6. The method according to claim 5, characterized in that In step S41, the step of determining which region the window belongs to based on the relationship between the mean and standard deviation of the window and the mean and standard deviation of the image specifically includes: If the mean of the 256*256 window is greater than 1.2 times the image mean, or the standard deviation of the 256*256 window is greater than 1.2 times the image standard deviation, the 256*256 window is determined to be a large bright area; If the mean of the 256*256 window is greater than 0.7 times the image mean, or the standard deviation of the 256*256 window is greater than 0.7 times the image standard deviation, the 256*256 window is determined to be a partially bright area; If none of the above conditions are met, the 256*256 window is judged as a small bright area.

7. The method according to claim 5, characterized in that In step S42, the specific determination method is as follows: Among them, Mean 区域 is the average value of the 4*4 small area, Mean 滑窗 and Std 滑窗 They are the mean and standard deviation of the 256*256 window respectively.

8. The method according to claim 1, characterized in that The step S5 specifically includes: First, a 7*7 kernel is used to perform a dilation operation to merge adjacent foreground pixels into a larger connected area, and then a 3*3 kernel is used to perform an erosion operation to remove noise from the image.

9. The method according to claim 1, characterized in that: The step S6 specifically includes: Perform a logical AND operation on the original image after dilation and erosion, and only keep the pixels that are also marked as strong scattering points in the marked image; Then, according to the position information of the sliding window, the positions of these strong scattering points are mapped back to the original image, thereby marking the strong scattering points in the original image.

Images