Interference image pollutant identification method based on parasitic image interference elimination

Abstract

The invention discloses an interference image pollutant identification method based on parasitic image interference elimination, and solves the problem that pollutant identification is inaccurate due to the fact that pollutant characterization is not obvious and pollutant identification is easy to be interfered by a parasitic image in a pollutant identification process of an interference image. The method comprises the following steps of: firstly, performing polynomial fitting on an interference dimension of an interference image to correct an image signal, and calculating correction coefficients of the interference dimension and a breadth dimension; and eliminating parasitic image interference in the image through relative radiation correction processing. And decomposing the image into different frequency components by using Haar wavelet transform, obtaining a vertical edge information image, calculating feature intensity, screening out an area with relatively high feature intensity, and identifying the area as a pollutant. And finally, mapping the position of the pollutant to an original interference image through coordinate restoration. According to the method, the pollutants in the interference image can be quickly and accurately identified, the quality evaluation of the interference image can be improved, and a reliable basis can be provided for subsequent data analysis and decision making.

Description

Technical Field

[0001] The invention relates to an image recognition method, and in particular to an interference image pollutant recognition method based on parasitic image interference elimination. Background Art

[0002] With the continuous development of large-aperture static interferometric spectral imaging technology, interferometric imaging plays an increasingly important role in scientific research and industrial applications. However, due to the influence of factors such as environmental noise and optical system errors, pollutants will appear in some areas of the interference image and the pollutant representation is not obvious. This phenomenon often leads to a significant decrease in imaging quality, especially in high spectral resolution application scenarios. In addition, Figure 1 and Figure 2 As shown, the parasitic image of the interference image appears as unnecessary artifacts or noise in the image, which can be easily confused with the characteristics of the pollutants and reduce the accuracy of recognition.

[0003] Therefore, eliminating the interference caused by parasitic images and then quickly and accurately identifying pollutants will not only help improve the quality assessment of interference images, but also provide a reliable basis for subsequent data analysis and decision-making. However, existing pollutant identification methods are unable to accurately identify them, which has hindered the assessment of interference image quality. Summary of the invention

[0004] In order to solve the technical problem that pollutant characterization is not obvious and is easily interfered by parasitic images during interference image identification of pollutants, resulting in inaccurate pollutant identification, the present invention provides an interference image pollutant identification method based on parasitic image interference elimination.

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

[0006] A method for identifying pollutants in interference images based on parasitic image interference elimination is characterized in that it includes the following steps:

[0007] Step 1, polynomial fitting is performed on the interference image along the interference dimension;

[0008] For the interference image I (H,W) Perform polynomial fitting along the interference dimension H to obtain the fitted interference image I′ (H,W) , where the interference dimension H represents the number of rows; the width dimension W represents the number of columns;

[0009] Step 2: Calculate the interference image I′ (H,W) The interference dimension correction coefficient k H ;

[0010] Step 3: Calculate the interference image I′ (H,W) The width dimension correction coefficient k W ;

[0011] Step 4: Correct the coefficient k by interferometer H and width dimension correction factor k W For the interference image I (H,W) Perform relative radiation correction to obtain the interference image I″ which eliminates the parasitic image interference;

[0012] Step 5: Use Haar wavelet to perform wavelet transform on the interference image I″ after eliminating parasitic image interference, decompose the interference image I″ into different frequency components, and obtain the vertical edge information image cV;

[0013] Step 6, calculating the amplitude of each pixel in the vertical edge information image cV, thereby obtaining the characteristic intensity of each pixel in the vertical edge information image cV; sorting the characteristic intensity of each column of pixels, and selecting the tth quantile after sorting as the characteristic intensity benchmark, 60%≤t≤80%;

[0014] Step 7, processing the characteristic intensity of each pixel in the vertical edge information image cV column by column, and identifying the pixel whose characteristic intensity is greater than a preset multiple of the characteristic intensity benchmark as an initial pollutant;

[0015] Step 8: Restore the coordinates of the initial pollutant to the interference image I (H,W) In the coordinates of , the coordinates of four pixels are obtained, and the pixel with the smallest pixel value is selected to identify it as the interference image I (H,W) final pollutant; complete interference image pollutant identification.

[0016] Furthermore, step 1 is specifically as follows:

[0017] For the interference image I (H,W) Perform polynomial fitting along the interference dimension H to obtain the fitted interference image I′ (H,W) , the interference dimension H represents the number of rows; the width dimension W represents the number of columns; the fitting formula is as follows:

[0018] I′ j =f n (I j );

[0019] Among them, I j The interference image I (H,W) The jth interference curve of the graph, j = 1, 2, ... or W; f n () represents an n-order polynomial function, n≥1; I′ j Represents the interference image I′ after polynomial fitting (H,W) The j-th interference curve of .

[0020] Furthermore, step 2 specifically includes:

[0021] 2.1. Interference image I′ (H,W) The average calculation is performed along the width dimension W to obtain the mean vector m=(m 1 ,m 2 ,…,m i ,…,m H ) T ;in, p (i,j) The interference image I′ (H,W) The pixel value of the i-th row and j-th column in ;

[0022] 2.2. Average the mean vector m to get the global mean m all :

[0023]

[0024] 2.3. Calculation of the interference dimension correction coefficient k using the relative radiation correction method H :

[0025]

[0026] Furthermore, step 3 specifically includes:

[0027] 3.1. Interference image I′ (H,W) The average calculation is performed along the interference dimension H, and the mean vector n = (n 1 ,n 2 ,…,n j ,…,n W ) T ;in,

[0028] 3.2. Calculation of the width dimension correction coefficient k using the relative radiation correction method W :

[0029]

[0030] Furthermore, step 4 is specifically as follows:

[0031] For the interference image I (H,W) Relative radiation correction is performed to obtain the interference image I″ which eliminates the parasitic image interference:

[0032] I″=I·k H ·k W .

[0033] Furthermore, step 6 is specifically as follows:

[0034] Calculate the amplitude cV′ of each pixel in the vertical edge information image cV (x,y) =|cV (x,y)|, to obtain the feature intensity of each pixel in the vertical edge information image cV; where (x, y) is the coordinate of the pixel; sort the feature intensity of each column of pixels in ascending order, and select the tth quantile after sorting as the feature intensity benchmark cV′ y , 60≤t≤80.

[0035] Furthermore, step 7 is specifically as follows:

[0036] Each pixel in the vertical edge information image cV is processed column by column, and the feature intensity greater than Q·cV′ is y The point (x pol ,y pol ) is identified as a pollutant; wherein Q is a preset multiple and Q≥1.

[0037] Furthermore, step 8 is specifically as follows:

[0038] The initial pollutant coordinates (x pol ,y pol ) is restored to the original interference image I (H,W) The coordinates (x ori ,y ori )middle:

[0039]

[0040] Interference image I′ (H,W) The coordinates (x ori ,y ori ) contains 4 pixels, namely Select the corresponding pixel point with the smallest pixel value and identify it as the interference image I (H,W) The final pollutant is detected and the interference image pollutant identification is completed.

[0041] Furthermore, in step 1, n=4.

[0042] Further, in step 7, Q=4;

[0043] In step 8, t=70%.

[0044] Beneficial effects of the present invention:

[0045] 1. The present invention provides an interference image pollutant identification method based on parasitic image interference elimination, which eliminates the interference of parasitic images, avoids misjudgment caused by confusion between parasitic images and pollutant features, and significantly improves the accuracy of pollutant identification.

[0046] 2. The interference image pollutant identification method based on parasitic image interference elimination provided by the present invention is suitable for various types of large-aperture static interference spectral imaging systems, can handle pollutant identification problems in different scenarios, and provides more reliable technical support for hyperspectral imaging in scientific research and industrial applications, and has broad application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Schematic diagram of the parasitic image area of ​​the interference image. The framed area in (a) is the parasitic image area, and the framed area in (b) is the area where the parasitic image of the interference image is easily confused with the contaminant characteristics;

[0048] Figure 2 It is a schematic diagram of some pollutants in the interference image;

[0049] Figure 3 It is a flow chart of an embodiment of an interference image pollutant identification method based on parasitic image interference elimination of the present invention;

[0050] Figure 4 is a schematic diagram of eliminating parasitic image interference by an interference image in an embodiment of the present invention;

[0051] Figure 5 is a schematic diagram of identifying pollutants using a vertical edge information image in an embodiment of the present invention;

[0052] Figure 6 Schematic diagram of identifying pollutants using interference images in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] The embodiment of the present invention provides a method for identifying pollutants in interference images based on eliminating parasitic image interference, which can accurately identify pollutants on the basis of eliminating parasitic image interference and improve the quality assessment of interference images. Figure 3 As shown, the specific steps include:

[0055] Step 1, polynomial fitting is performed on the interference image along the interference dimension;

[0056] For the interference image I (H,W) Perform polynomial fitting along the interference dimension H to obtain the fitted interference image I′ (H,W) , the interference dimension H represents the number of rows; the width dimension W represents the number of columns; the fitting formula is as follows:

[0057] I′ j =f n (I j );

[0058] Among them, I j The interference image I (H,W) The jth interference curve of the graph, j = 1, 2, ... or W; f n () represents an n-order polynomial function, n≥1; I′ j Represents the interference image I′ after polynomial fitting (H,W) The j-th interference curve of ; in this embodiment, H=256, W=1024, n=4.

[0059] Step 2: Calculate the interference dimension correction coefficient k H ;

[0060] 2.1. Interference image I′ (H,W) The average calculation is performed along the width dimension W to obtain the mean vector m=(m 1 ,m 2 ,…,m i ,…,m H ) T ;in, p (i,j) The interference image I′ (H,W) The pixel value of the i-th row and j-th column in ;

[0061] 2.2. Average the mean vector m to get the global mean m all :

[0062]

[0063] 2.3. Calculation of the interference dimension correction coefficient k using the relative radiation correction method H :

[0064]

[0065] Step 3: Calculate the width dimension correction coefficient k W ;

[0066] 3.1. Interference image I′ (H,W) The average calculation is performed along the interference dimension H, and the mean vector n = (n 1 ,n 2 ,…,n j ,…,n W ) T ;in,

[0067] 3.2. Calculation of the width dimension correction coefficient k using the relative radiation correction method W :

[0068]

[0069] Step 4, eliminating the interference of parasitic images of the interference image;

[0070] like Figure 4 As shown, for the interference image I (H,W) Relative radiation correction is performed to obtain the interference image I″ which eliminates the parasitic image interference:

[0071] I″=I·K H ·k W ;

[0072] Step 5: wavelet transform processing;

[0073] The interference image I″ is subjected to wavelet transform by using Haar wavelet to eliminate the parasitic interference, and the interference image I″ is decomposed into different frequency components to obtain the basic information image cA, the horizontal edge information image cH, the vertical edge information image cV and the diagonal edge information image cD; and the rows and columns of the basic information image cA, the horizontal edge information image cH, the vertical edge information image cV and the diagonal edge information image cD are all the interference image I″.

[0074] Step 6: Calculate feature strength;

[0075] Calculate the amplitude cV′ of each pixel in the vertical edge information image cV (x,y) =|cV (x,y) |, to obtain the characteristic intensity of each pixel in the vertical edge information image cV; where (x, y) is the coordinate of the pixel; and sort the characteristic intensity of each column of pixels in ascending order, and select the tth quantile after sorting as the characteristic intensity benchmark cV′ y , 60≤t≤80; in this embodiment, t=70%;

[0076] Step 7: Pollutant identification;

[0077] Each pixel in the vertical edge information image cV is processed column by column, and the feature intensity greater than Q·cV′ is y The point (x pol ,y pol ) identified as initial contaminants, such as Figure 5 As shown; wherein, Q is a preset multiple and Q≥1; in this embodiment, Q=4 is preferred;

[0078] Step 8: coordinate restoration;

[0079] The initial pollutant coordinates (x pol ,ypol ) is restored to the original interference image I (H,W) The coordinates (x ori ,y ori )middle:

[0080]

[0081] Interference image I′ (H,W) The coordinates (x ori ,y ori ) contains 4 pixels, namely

[0082] like Figure 6 As shown, the corresponding pixel point with the smallest pixel value is selected and identified as the interference image I (H,W) final pollutant.

[0083] Complete interference image pollutant identification.

[0084] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for identifying pollutants in interference images based on parasitic image interference elimination, characterized in that: The following steps are involved: Step 1, polynomial fitting is performed on the interference image along the interference dimension; For the interference image I (H,W) Perform polynomial fitting along the interference dimension H to obtain the fitted interference image I′ (H,W) , where the interference dimension H represents the number of rows; the width dimension W represents the number of columns; Step 2: Calculate the interference image I′ (H,W) The interference dimension correction coefficient k H ; Step 3: Calculate the interference image I′ (H,W) The width dimension correction coefficient k W ; Step 4: Correct the coefficient k by interferometer H and width dimension correction factor k W For the interference image I (H,W) Perform relative radiation correction to obtain the interference image I″ which eliminates the parasitic image interference; Step 5: Use Haar wavelet to perform wavelet transform on the interference image I″ after eliminating parasitic image interference, decompose the interference image I″ into different frequency components, and obtain the vertical edge information image cV; Step 6, calculating the amplitude of each pixel in the vertical edge information image cV, thereby obtaining the characteristic intensity of each pixel in the vertical edge information image cV; sorting the characteristic intensity of each column of pixels, and selecting the tth quantile after sorting as the characteristic intensity benchmark, 60%≤t≤80%; Step 7, processing the characteristic intensity of each pixel in the vertical edge information image cV column by column, and identifying the pixel whose characteristic intensity is greater than a preset multiple of the characteristic intensity benchmark as an initial pollutant; Step 8: Restore the coordinates of the initial pollutant to the interference image I (H,W) In the coordinates of , the coordinates of four pixels are obtained, and the pixel with the smallest pixel value is selected to identify it as the interference image I (H,W) final pollutant; complete interference image pollutant identification.

2. The interference image pollutant identification method based on parasitic image interference elimination according to claim 1 is characterized in that: Step 1 is as follows: For the interference image I (H,W) Perform polynomial fitting along the interference dimension H to obtain the fitted interference image I′ (H,W) , the interference dimension H represents the number of rows; the width dimension W represents the number of columns; the fitting formula is as follows: I′ j =f n (I j ); Among them, I j The interference image I (H,W) The jth interference curve of the graph, j = 1, 2, ... or W; f n () represents an n-order polynomial function, n≥1; I′ j Represents the interference image I′ after polynomial fitting (H,W) The j-th interference curve of .

3. The interference image pollutant identification method based on parasitic image interference elimination according to claim 2 is characterized in that: Step 2 specifically includes: 2.

1. Interference image I′ (H,W) The average calculation is performed along the width dimension W to obtain the mean vector m = (m1, m2, ..., m i ,…,m H ) T ;in, p (i,j) The interference image I′ (H,W) The pixel value of the i-th row and j-th column in ; 2.

2. Average the mean vector m to get the global mean m all : 2.

3. Calculation of the interference dimension correction coefficient k using the relative radiation correction method H :

4. The interference image pollutant identification method based on parasitic image interference elimination according to claim 3 is characterized in that: Step 3 specifically includes: 3.

1. Interference image I′ (H,W) The average calculation is performed along the interference dimension H, and the mean vector n=(n1,n2,…,n j ,…,n W ) T ;in, 3.

2. Calculation of the width dimension correction coefficient k using the relative radiation correction method W :

5. The interference image pollutant identification method based on parasitic image interference elimination according to claim 4 is characterized in that: Step 4 is as follows: For the interference image I (H,W) Relative radiation correction is performed to obtain the interference image I″ which eliminates the parasitic image interference: I″=I·k H ·k W 。 6. The interference image contaminant identification method based on parasitic image interference elimination according to claim 5 is characterized in that: Step 6 is as follows: Calculate the amplitude cV′ of each pixel in the vertical edge information image cV (x,y) =|cV (x,y) |, to obtain the characteristic intensity of each pixel in the vertical edge information image cV; Where (x, y) is the coordinate of the pixel point; sort the feature intensity of each column of pixels from small to large, and select the tth quantile after sorting as the feature intensity benchmark cV′ y , 60≤t≤80.

7. The interference image pollutant identification method based on parasitic image interference elimination according to claim 6 is characterized in that: Step 7 is as follows: Each pixel in the vertical edge information image cV is processed column by column, and the feature intensity greater than Q·cV′ is y The point (x pol ,y pol ) is identified as a pollutant; wherein Q is a preset multiple and Q≥1.

8. The interference image pollutant identification method based on parasitic image interference elimination according to claim 7 is characterized in that: Step 8 is as follows: The initial pollutant coordinates (x pol ,y pol ) is restored to the original interference image I (H,W) The coordinates (x ori ,y ori )middle: Interference image I′ (H,W) The coordinates (x ori ,y ori ) contains 4 pixels, namely and Select the corresponding pixel point with the smallest pixel value and identify it as the interference image I (H,W) The final pollutant is detected and the interference image pollutant identification is completed.

9. The interference image contaminant identification method based on parasitic image interference elimination according to any one of claims 2 to 8, characterized in that: In step 1, n=4.

10. The interference image pollutant identification method based on parasitic image interference elimination according to claim 8, characterized in that: In step 7, Q=4; In step 8, t=70%.

Images