Microwave radiometer stripe elimination preprocessing method based on moving median smoothing

Abstract

The invention discloses a microwave radiometer stripe elimination preprocessing method based on moving median smoothing, and the method comprises the following steps: 1, carrying out the circular scanning through a microwave radiometer, and obtaining a heat source count value Cw, a cold source count value Cc, an observation target count value Ce, a heat source brightness temperature Tw, and a cold source brightness temperature Tc; step 2, performing noise smoothing on moving medians between scanning lines on the heat source observation count value Cw and the cold source count value Cc to obtain a smoothed heat source observation count value and a smoothed cold source count value; and step 3, performing two-point calibration by using the heat source brightness temperature Tw and the cold source brightness temperature Tc obtained in the step 1 and the smoothed heat source observation count value and the smoothed cold source count value obtained in the step 2 to obtain the stripe-eliminated observation target brightness temperature Te. The method provided by the invention can effectively solve the problem of influence of stripe noise on the data quality of the microwave radiometer in the existing stripe elimination method.

Description

Technical Field

[0001] This invention belongs to the field of space microwave remote sensing technology, specifically relating to a microwave radiometer anti-fringe preprocessing method based on moving median smoothing. Background Technology

[0002] Microwave radiometers receive microwave radiation heat from targets such as the atmosphere and ocean to observe parameters such as atmospheric temperature and humidity profiles and ocean salinity. They are core equipment in meteorology, agriculture, environment, and remote sensing. Preprocessing technology, as a bridge connecting raw observation data and subsequent applications, directly determines the quality of microwave radiometer data and the value of its final application.

[0003] Stripe noise is a common problem in microwave radiometer preprocessing studies. It manifests as signal deviation along the track direction, forming a visually apparent striped pattern. Stripe noise affects microwave radiometers throughout core processes such as data quality control and calibration, directly limiting the accuracy of subsequent quantitative inversion, and therefore requires close attention.

[0004] Existing methods for eliminating fringes can be divided into two categories. One category processes the brightness temperature after calibration, extracting fringes through principal component analysis and ensemble empirical mode decomposition. The other category processes the original count values ​​before calibration, using rectangular or triangular windows for smoothing, or spline fitting for count value processing. Rectangular and triangular windows are essentially weighted averages, and their results are significantly "pulled" by extreme outliers within the window. Spline fitting, on the other hand, fits the data using a polynomial curve, and outliers can cause the curve to deviate from the true trend. Moving median, however, is based on the sorting of data within the window, and extreme outliers have almost no impact on the result. Summary of the Invention

[0005] The purpose of this invention is to provide a preprocessing method for eliminating stripes in microwave radiometers based on moving median smoothing, so as to solve the problem of stripe noise affecting the data quality of microwave radiometers in existing stripe elimination methods.

[0006] To achieve the above objectives, the present invention employs the following technical solution:

[0007] A microwave radiometer antifringe preprocessing method based on moving median smoothing includes the following steps:

[0008] Step 1: Perform a circular scan using a microwave radiometer to obtain the heat source count value C. w Cold source count value C c Observation target count value C e Heat source brightness temperature T w and cold source brightness T c ;

[0009] Step 2, observe the heat source count value C.w and cold source count value C c The median noise of the scan lines was smoothed separately to obtain the smoothed heat source and cold source counts.

[0010] Step 3, using the heat source brightness temperature T obtained in Step 1 w Cold source brightness T c The smoothed heat source and cold source counts obtained in step 2 are calibrated at two points to obtain the observed target brightness temperature T after stripe removal. e .

[0011] Furthermore, in step 1, for the m-th scan line, the four heat source count values ​​obtained are C wm,1 C wm,2 C wm,3 C wm,4 The brightness temperatures of the four heat sources are T. wm,1 T wm,2 T wm,3 T wm,4 The count values ​​of the four cold sources are C cm,1 C cm,2 C cm,3 C cm,4 The brightness temperatures of the four cold sources are T. cm,1 T cm,2 T cm,3 T cm,4 Where 1≤m≤M, M is the total number of scan lines, and the count value of the nth observed target is C. em,n , where 1≤n≤N, and N is the total number of observed targets.

[0012] Furthermore, step 2 includes the following sub-steps:

[0013] Step 21: Perform median smoothing on the first heat source count value of all scan lines. For the first heat source count value C of the m-th scan line... wm,1 The smoothed heat source count value is The moving median sliding window size is k, where k is an odd number;

[0014] The smoothed heat source count value of the first heat source in the m-th scan line is obtained using the following procedure.

[0015] Step 211, define the original data points covered by the moving median sliding window as:

[0016]

[0017] Wherein, the left boundary L of the window m and right boundary R m The following conditions must be met:

[0018] (1) When m is far from the edge, i.e. when and hour,

[0019]

[0020] (2) When m is close to the edge, that is, when and hour,

[0021] Define the window to automatically shrink to avoid exceeding the bounds, that is:

[0022]

[0023] Step 212, process the data C within the window. wj,1 Sort in ascending order to get:

[0024]

[0025] Among them, Z wj,1-1 ≤Z wj,1-2 ≤…≤Z wj,1-p p = R m -L m +1 represents the actual number of data points in the current window, where p≤k;

[0026] Step 213: Calculate the smoothed heat source count value.

[0027] If p is odd p = 2 × t + 1, where t is a positive integer;

[0028] If p is even p = 2 × t;

[0029] Step 22, count the heat source values ​​C of the 2nd, 3rd, and 4th heat sources on the m-th scan line. wm,2 C wm,3 C wm,4 Following the same process as step 21, the smoothed heat source count values ​​are obtained respectively. The count values ​​C of the 1st, 2nd, 3rd, and 4th cold sources on the m-th scan line. cm,1 C cm,2 C cm,3 C cm,4 Following the same process as step 21, the smoothed cold source count value of the m-th scan line is obtained respectively.

[0030] Furthermore, in step 21, the value of k is 9.

[0031] Furthermore, step 3 includes the following sub-steps:

[0032] Step 31, calculate the average cold source count value of the m-th scan line.

[0033]

[0034] Step 32, calculate the average heat source count value of the m-th scan line.

[0035]

[0036] Step 33, calculate the scaling factor k of the m-th scan line. m b m :

[0037]

[0038] Step 34: Calculate the brightness temperature of the observed target using the calibration coefficient of the m-th scan line obtained in Step 33. This is the brightness temperature of the observed target after stripe removal corresponding to the m-th scan line.

[0039] T em,n =k m ×C em,n +b m

[0040] Among them, C em,n This is the count value of the nth observed target;

[0041] Step 35: Traverse all scan lines to obtain the brightness temperature of the observed target after stripe removal for each scan line.

[0042] The beneficial effects of this invention compared to the prior art are:

[0043] This invention processes the raw count values ​​before calibration, offering significant advantages in handling outlier counts with abrupt changes. Rectangular and triangular windows are essentially weighted averages, and their results are significantly "pulled" by extreme outliers within the window. Spline fitting fits data using polynomial curves, and outliers can cause the curve to deviate from the true trend. In contrast, this invention uses a moving median based on the sorting of data within the window, so extreme outliers have almost no impact on the results. This effectively solves the problem of stripe noise affecting the data quality of microwave radiometers in existing stripe removal methods. Attached Figure Description

[0044] Figure 1 This is a flowchart of the microwave radiometer anti-fringe preprocessing technique based on moving median smoothing according to the present invention.

[0045] Figure 2 This is a schematic diagram of the count value processing.

[0046] Figure 3 The above figure shows a comparison of heat source count values ​​after different smoothing methods, and the bottom figure shows a local magnification of scan lines 300-400, which is a comparison of the embodiments of the present invention.

[0047] Figure 4 The results of stripe removal in an embodiment of the present invention were obtained through a vacuum calibration experiment, wherein (a) is the brightness temperature of the target scene before stripe removal, (b) is the brightness temperature of the target scene after stripe removal, and (c) is the stripe that has been removed. Detailed Implementation

[0048] The present invention provides a microwave radiometer anti-fringe preprocessing method based on moving median smoothing, comprising the following steps:

[0049] Step 1: Perform a circular scan using a microwave radiometer to obtain the heat source count value C. w Cold source count value C c Observation target count value C e Heat source brightness temperature T w and cold source brightness T c For the m-th scan line, the four heat source count values ​​obtained are C wm,1 C wm,2 C wm,3 C wm,4 The brightness temperatures of the four heat sources are T. wm,1 T wm,2 T wm,3 T wm,4 The count values ​​of the four cold sources are C cm,1 C cm,2 C cm,3 C cm,4 The brightness temperatures of the four cold sources are T. cm,1 T cm,2 T cm,3 T cm,4 Where 1≤m≤M, M is the total number of scan lines, and the count value of the nth observed target is C. em,n , where 1≤n≤N, and N is the total number of observed targets;

[0050] Step 2, observe the heat source count value C. w and cold source count value C c The median noise of the scan lines was smoothed separately to obtain the smoothed heat source and cold source counts.

[0051] like Figure 2 As shown, step 2 includes the following sub-steps:

[0052] Step 21: Perform median smoothing on the first heat source count value of all scan lines, where the first heat source count value C of the m-th scan line is... wm,1 The smoothed heat source count value is The moving median sliding window size is k, where k is an odd number;

[0053] The smoothed heat source count value of the first heat source in the m-th scan line is obtained using the following procedure.

[0054] Step 211, define the original data points covered by the moving median sliding window as:

[0055]

[0056] Wherein, the left boundary L of the window m and right boundary R m The following conditions must be met:

[0057] (1) When m is far from the edge, i.e. when and hour,

[0058]

[0059] (2) When m is close to the edge, that is, when and hour,

[0060] Define the window to automatically shrink to avoid exceeding the bounds, that is:

[0061]

[0062] Step 212, process the data C within the window. wj,1 Sort in ascending order to get:

[0063] Z wj,1 =[Z wj,1-1 Z wj,1-2 ,…,Z wj,1-p (4);

[0064] Among them, Z wj,1-1 ≤Z wj,1-2 ≤…≤Z wj,1-p p = R m -L m +1 represents the actual number of data points in the current window, where p≤k;

[0065] Step 213: Calculate the smoothed heat source count value.

[0066] If p is odd p = 2 × t + 1, where t is a positive integer;

[0067] If p is even p = 2 × t;

[0068] Step 22, count the heat source values ​​C of the 2nd, 3rd, and 4th heat sources on the m-th scan line. wm,2 C wm,3 C wm,4 Following the same process as step 21, the smoothed heat source count values ​​are obtained respectively. The count values ​​C of the 1st, 2nd, 3rd, and 4th cold sources on the m-th scan line. cm,1 C cm,2 C cm,3 C cm,4 Following the same process as step 21, the smoothed cold source count value of the m-th scan line is obtained respectively.

[0069] Step 3, using the heat source brightness temperature T obtained in Step 1 w Cold source brightness T c The smoothed heat source and cold source counts obtained in step 2 are calibrated at two points to obtain the observed target brightness temperature T after stripe removal. e .

[0070] Step 3 includes the following sub-steps:

[0071] Step 31, calculate the average cold source count value of the m-th scan line.

[0072]

[0073] Step 32, calculate the average heat source count value of the m-th scan line.

[0074]

[0075] Step 33, calculate the scaling factor k of the m-th scan line. m b m :

[0076]

[0077] Step 34: Calculate the brightness temperature of the observed target using the calibration coefficient of the m-th scan line obtained in Step 33. This is the brightness temperature of the observed target after stripe removal corresponding to the m-th scan line.

[0078] T em,n =k m ×C em,n +b m (9).

[0079] Among them, Cem,n The count value for the nth observed target (obtained from step 1).

[0080] Step 35: Traverse all scan lines to obtain the brightness temperature of the observed target after stripe removal for each scan line.

[0081] Figure 3 The left figure shows a comparison of heat source count values ​​after different smoothing methods. The black line in the figure represents unsmoothed data, the blue line represents triangular weighted moving average, the red line represents cubic spline smoothing, and the green line represents the moving median smoothing of the present invention. The right figure is a local magnification of the scan line 300-400.

[0082] Figure 4 The results of stripe removal in an embodiment of the present invention were obtained through a vacuum calibration experiment. From bottom to top, the results are: (a) the brightness temperature of the target scene before stripe removal, (b) the brightness temperature of the target scene after stripe removal, and (c) the removed stripes. (c) demonstrates the effectiveness of the method of the present invention; the extracted signal deviation along the track direction forms the visual stripe pattern.

[0083] pass Figure 3 A comparison of the smoothing effects shows that the original unsmoothed data (black) exhibits significant spikes and outlier interference, making the triangular weighted moving average easily skewed by outliers, resulting in significant curve fluctuations. In contrast, the moving median smoothing (green) of this invention effectively resists the influence of outliers and maintains the stability of the data trend. Combined with... Figure 4 After processing by the method of this invention, the stripe noise of the target scene brightness temperature is significantly removed (this is verified by comparing the brightness temperature before and after elimination and extracting the stripe noise separately).

[0084] In summary, the method of the present invention not only has a strong smoothing ability against outlier interference, but also effectively suppresses stripe noise in microwave radiometer data, significantly reducing the impact of stripes on data quality and improving the stability and reliability of the data.

[0085] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A microwave radiometer anti-fringe preprocessing method based on moving median smoothing, characterized in that, Includes the following steps: Step 1: Use a microwave radiometer to perform a circular scan and obtain the heat source count value C. w Cold source count value C c Observation target count value C e Heat source brightness temperature T w and cold source brightness T c ; Step 2, observe the heat source count value C. w and cold source count value C c The median noise of the scan lines was smoothed separately to obtain the smoothed heat source and cold source counts. Step 3, using the heat source brightness temperature T obtained in Step 1 w Cold source brightness T c The smoothed heat source and cold source counts obtained in step 2 are calibrated at two points to obtain the observed target brightness temperature T after stripe removal. e .

2. The microwave radiometer anti-fringe preprocessing method based on moving median smoothing as described in claim 1, characterized in that, In step 1, for the m-th scan line, the four heat source count values ​​obtained are C wm,1 C wm,2 C wm,3 C wm,4 The brightness temperatures of the four heat sources are T. wm,1 T wm,2 T wm,3 T wm,4 The count values ​​of the four cold sources are C cm,1 C cm,2 C cm,3 C cm,4 The brightness temperatures of the four cold sources are T. cm,1 T cm,2 T cm,3 T cm,4 Where 1≤m≤M, M is the total number of scan lines, and the count value of the nth observed target is C. em,n , where 1≤n≤N, and N is the total number of observed targets.

3. The microwave radiometer anti-fringe preprocessing method based on moving median smoothing as described in claim 2, characterized in that, Step 2 includes the following sub-steps: Step 21: Perform median smoothing on the first heat source count value of all scan lines. For the first heat source count value C of the m-th scan line... wm,1 The smoothed heat source count value is The moving median sliding window size is k, where k is an odd number; The smoothed heat source count value of the first heat source in the m-th scan line is obtained using the following procedure. Step 211, define the original data points covered by the moving median sliding window as: Wherein, the left boundary L of the window m and right boundary R m The following conditions must be met: (1) When m is far from the edge, i.e. when and hour, (2) When m is close to the edge, that is, when and hour, Define the window to automatically shrink to avoid exceeding the bounds, that is: Step 212, process the data C within the window. wj,1 Sort in ascending order to get: WITH wj,1 =[Z wj,1-1 ,WITH wj,1-2 ,…,WITH wj,1-p ]; Among them, Z wj,1-1 ≤Z wj,1-2 ≤…≤Z wj,1-p p = R m -L m +1 represents the actual number of data points in the current window, where p≤k; Step 213: Calculate the smoothed heat source count value. If p is odd Where t is a positive integer; If p is even Step 22, count the heat source values ​​C of the 2nd, 3rd, and 4th heat sources on the m-th scan line. wm,2 C wm,3 C wm,4 Following the same process as step 21, the smoothed heat source count values ​​are obtained respectively. The count values ​​C of the 1st, 2nd, 3rd, and 4th cold sources on the m-th scan line. cm,1 C cm,2 C cm,3 C cm,4 Following the same process as step 21, the smoothed cold source count value of the m-th scan line is obtained respectively.

4. The microwave radiometer anti-fringe preprocessing method based on moving median smoothing as described in claim 3, characterized in that, In step 21, the value of k is 9.

5. The microwave radiometer anti-fringe preprocessing method based on moving median smoothing as described in claim 3, characterized in that, Step 3 includes the following sub-steps: Step 31, calculate the average cold source count value of the m-th scan line. Step 32, calculate the average heat source count value of the m-th scan line. Step 33, calculate the scaling factor k for the m-th scan line. m b m : Step 34: Calculate the brightness temperature of the observed target using the calibration coefficient of the m-th scan line obtained in Step 33. This is the brightness temperature of the observed target after stripe removal corresponding to the m-th scan line. T em,n =k m ×C em,n +b m Among them, C em,n This is the count value of the nth observed target; Step 35: Traverse all scan lines to obtain the brightness temperature of the observed target after stripe removal for each scan line.

Images