CN-115690604-B - Stripe noise removing method based on mean value compensation

CN115690604BCN 115690604 BCN115690604 BCN 115690604BCN-115690604-B

Abstract

Aiming at the problem that stripe noise is easy to generate at the joint of land and water areas, the invention provides a stripe noise removing method based on mean value compensation, which comprises the steps of firstly dividing an image with stripe noise according to water and land, then dividing the extracted water into two types, wherein one type is all water areas S 1 , the other type is water areas S2 penetrating through all columns, and acquiring a starting row r 1 of the S2 area and an ending row r 2 of the S 2 area; and calculating a compensation value by utilizing the S 2 area, and finally, performing pixel-by-pixel compensation on the original stripe noise image in step 4 to obtain a denoised image Y. The invention can effectively remove the stripe noise in the image, avoid generating new stripe noise after removing the existing stripe noise, and can furthest reserve the interrelation of gray information among different detectors.

Inventors

YANG BO
GENG ZEMIN
LI XIN

Assignees

武汉大学

Dates

Publication Date: 20260508
Application Date: 20221111

Claims (3)

1. The method for removing the stripe noise based on the mean value compensation is characterized by comprising the following steps of: Step 1, dividing an image with stripe noise according to a water body and land; Step 2, dividing the water body extracted in the step 1 into two types, wherein one type is all water body areas The second type is a water body area penetrating through all columns Acquisition of Start row of region A kind of electronic device End row of region ; Step3, utilize The area calculation compensation value specifically comprises the following steps: Step 31, using area Calculating a reference column Col as a reference area, the reference column being composed of The average value of all gray values of each row of the region is formed, and the reference column Col has the pixel number RefNum of +1; The calculation formula of the reference column in step 31 is as follows: (1) In the middle of Representing the first reference column Elements ColNum are The number of pixels per line in the region, colNum, corresponds to the number of pixels per line in the whole image, Is the reference area Row of lines Pixel values of the columns; Step 32, calculating The average value of all pixel values in the region is used as a reference average value, and under the condition that the precision needs to be improved, the method uses And The regions jointly calculate a reference mean value; Calculation in step 32 The average value of all pixel values in the region is taken as a reference average value, and the calculation formula is as follows: (2) in the formula, Representing the average of all pixel values of the reference column, The number of elements that the reference column has is represented, For reference column A pixel value; Step 33, calculating based on the reference mean value The difference value between the gray value average value and the reference average value of all pixels in each column of the region; The calculation formula of the difference in step 33 is as follows; (3) In the middle of Is S2 region Row of lines The gray value of the column, Is the first The mean value of the column compensates for the value, Representing the number of elements that the reference column has; And 4, performing pixel-by-pixel compensation on the original stripe noise image to obtain a denoised image Y.
2. The method for removing stripe noise based on mean value compensation as claimed in claim 1, wherein in step 1, automatic segmentation is realized by using a semantic segmentation method, or manual intervention is performed to roughly segment an image, and the image is divided into a water body S and a land area L.
3. The method for removing banding noise based on mean value compensation as set forth in claim 1, wherein the calculation formula of the image Y after denoising in step 4 is as follows: in order to correct the pixel values of the image, In order to correct the pixel values of the pre-image, For the compensation value calculated in step 3, 、 Representing rows and columns.

Description

Stripe noise removing method based on mean value compensation Technical Field The invention belongs to the field of remote sensing image processing, and particularly relates to a method for removing stripe noise of a remote sensing image with a large amount of water body areas. Background In the radiation correction process of Beijing resource satellite application center high-resolution and high-resolution model images, it is found that even after radiation correction is performed in a correction coefficient mode, on a remote sensing image with a large-area water body, the response degree between detectors is different due to the mutation of radiation difference between the water and the land, so that stripe noise is easy to generate in a land and water junction area. These banding noise have a significant impact on the normal analytical use of subsequent images after distribution to units. Therefore, it is necessary to remove the banding noise of such images. At present, the method for removing the stripe noise is divided into two types, the first type is to convert to a frequency domain through a Fourier transform method and the like, and the stripe noise with periodic property is removed through setting a cut-off frequency. The second category is to compare classical methods such as standard moment matching and histogram matching by a statistical method, wherein the principles of the methods are that a certain normal detector is selected as a standard, and other sensors use the standard as a reference to reduce the mean value and the standard deviation to the reference sensor, so that the strip noise is removed. Experiments show that after the strip noise at the land-water junction is processed, the moment matching method can generate new strip noise in other land areas, so that subsequent processing and use are more difficult. Disclosure of Invention The invention solves the technical problem of providing a stripe noise removing method based on mean value compensation, thereby realizing the effective removal of stripe noise at the land-water junction and maximally ensuring that the image gray information of a non-stripe noise area is not influenced. The technical scheme for realizing the purpose of the invention is that the method for removing the stripe noise based on the mean value compensation comprises the following steps: And step 1, dividing the image with the stripe noise according to the water body and the land. Step 2, the water body extracted in step 1 is divided into two types, one type is all water body areas S1, the second type is a water body area S2 penetrating through all columns, a starting row r 1 of an S 2 area and an ending row r 2 of an S 2 area are obtained. And 3, calculating a compensation value by using the S2 region, wherein the method specifically comprises the following steps of: Step 31, calculating a reference column Col by taking the region S 2 as a reference region, wherein the reference column consists of the average value of all gray values of each row of the region S 2, and the number RefNum of pixels in the reference column Col is r 2-r1 +1; Step 32, calculating the average value of all pixel values in the S 2 area as a reference average value, and calculating the reference average value by utilizing the S 1 area and the S 2 area together under the condition that the accuracy needs to be improved; Step 33, calculating the difference between the gray value average value of all pixels in each column of the S 2 area and the reference average value by taking the reference average value as a reference; and 4, compensating the whole scene image according to the compensation value calculated in the step 3 to obtain an image with stripe noise removed. Furthermore, in the step 1, automatic segmentation is realized by using a semantic segmentation method, or manual intervention is performed to roughly segment the image, and the image is divided into a water body S and a land area L. Further, the calculation formula of the reference column in step 31 is as follows: Where Col i denotes the i-th element of the reference column, colNum is the number of pixels in each row of the S 2 region, colNum is the same as the number of pixels in each row of the whole image, and X i,j is the pixel value of the i-row and j-column of the reference region. Further, in step 32, the average value of all pixel values in the area S 2 is calculated as the reference average value, and the calculation formula is: Where Refx represents the average value of all pixel values in the reference column, refNum represents the number of elements in the reference column, and Col i is the i-th pixel value in the reference column. Further, the calculation formula of the difference in step 33 is as follows; wherein X i,j is the gray value of row j of the S2 region, dif j is the mean value compensation value of the j-th column, and ColNum is the number of pixels in each column of the reference region S 2. Further, t