Infrared high dynamic range image compression and detail enhancement method

Abstract

The invention discloses an infrared high dynamic range image compression and detail enhancement method, which comprises the following steps of: respectively sending an input infrared high dynamic range image into a contrast enhancement branch and a detail enhancement branch for processing, carrying out weighted fusion on obtained data, and outputting a 8-bit infrared image; wherein the contrast enhancement branch is sequentially subjected to mean filtering, dynamic range mapping and contrast histogram equalization limitation, and the detail enhancement branch is sequentially subjected to Laplace filtering, dynamic range mapping, return-to-zero corrosion and stripe noise suppression. According to the method, the details of the high-dynamic-range compressed image are enhanced, the algorithm execution efficiency is high, and the method is more suitable for being applied to a real-time scene.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, and in particular to an infrared high dynamic range image compression and detail enhancement method. Background Art

[0002] High-bit-width infrared detectors have the characteristics of high precision and high sensitivity, so infrared high dynamic range images have become an important development direction of infrared imaging technology. However, display devices can usually only display 8-bit wide image data, and high-bit data is not conducive to subsequent rapid processing, so the research on infrared high dynamic range image compression is becoming more and more important.

[0003] Existing high dynamic range compression and enhancement methods generally use multi-scale image pyramids and complex nonlinear filtering operators, which have problems such as loss of weak texture features and low execution efficiency.

[0004] Therefore, it is of practical value to adopt infrared high dynamic range image compression and detail enhancement methods with less detail loss and high execution efficiency. Summary of the invention

[0005] Aiming at the problems of detail loss and low execution efficiency of high dynamic range image compression and detail enhancement algorithms, the present invention designs an infrared high dynamic range image compression and detail enhancement method.

[0006] The technical solution adopted by the present invention to solve its technical problem is: an infrared high dynamic range image compression and detail enhancement method, the steps are: the input infrared high dynamic range image is sent to a contrast enhancement branch and a detail enhancement branch for processing respectively, the obtained data are weightedly fused, and an 8-bit infrared image is output; the contrast enhancement branch sequentially performs mean filtering, dynamic range mapping and histogram equalization enhancement with limited contrast; the detail enhancement branch sequentially performs Laplace filtering, dynamic range mapping, zeroing corrosion and stripe noise suppression.

[0007] Furthermore, the contrast enhancement branch is implemented by filtering formula The input image is subjected to a 3×3 mean filter, where x and y are pixel coordinates, I′(x, y) represents the pixel value after the mean filter, and Ω represents the neighborhood containing the pixel’s own coordinates (x, y).

[0008] Furthermore, in the contrast enhancement branch, the formula Perform dynamic range mapping on the mean filtered image, where g is the pixel value, Mean is the mean of the image after mean filtering, and Std is the standard deviation of the image after mean filtering.

[0009] Furthermore, in the contrast enhancement branch, the image after dynamic range mapping is subjected to histogram equalization with limited contrast through the following steps:

[0010] (1) Divide the image into 64 regions with 8 rows and 8 columns;

[0011] (2) Calculate the histogram of each region;

[0012] (3) According to the threshold set in advance, the statistical number of single gray levels exceeding the threshold is evenly distributed to all gray levels, and distributed step by step from low to high;

[0013] (4) Calculate the cumulative probability histogram;

[0014] (5) For each pixel in the image, the enhanced grayscale value result is obtained using the bilinear interpolation method based on the cumulative probability histogram of the four adjacent regions.

[0015] Furthermore, the detail enhancement branch is implemented by formula Perform a 3×3 Laplace filter on the input image, where x and y are pixel coordinates, I′(x,y) represents the pixel value after filtering, and Ω i Represents the neighborhood of coordinates (x,y) that does not include the pixel itself.

[0016] Furthermore, the detail enhancement branch is implemented by formula Perform dynamic range mapping on the image after Laplace filtering, where g is the pixel value, Mean is the mean of the image after Laplace filtering, and Std is the standard deviation of the image after Laplace filtering.

[0017] Furthermore, in the detail enhancement branch, the image after dynamic range mapping is eroded by 3×3 zeroing by the following formula:

[0018]

[0019] Where min() and max() represent the minimum and maximum values ​​respectively.

[0020] Furthermore, in the detail enhancement branch, stripe noise is suppressed on the image after zeroing erosion, that is, the pixel mean of each row is first calculated, and then the pixel mean of the row is subtracted from each pixel value of the row.

[0021] Furthermore, through formula I ans (x,y)=I c (x,y)+0.5*I f (x, y) performs pixel-by-pixel weighted fusion on the images obtained by the contrast enhancement branch and the detail enhancement branch, and outputs the fused image, where I ans represents the output image, I c represents the contrast enhanced branch image, I f Represents detail enhanced branch image.

[0022] The beneficial effects of the present invention are as follows: the method of the present invention uses the Laplace operator to extract the image detail layer, processes the original high dynamic range image and the detail layer in parallel and compresses the dynamic range, and finally fuses the compression results, so that the output 8-bit infrared image is rich in details and has a strong contrast, the algorithm execution efficiency is high and the display quality of the compressed image is improved; compared with the existing method, the method enhances the details of the high dynamic range compressed image, and the algorithm execution efficiency is high, and it is more suitable for application in real-time scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is the algorithm flow chart of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0025] The technical problem to be solved by the present invention is to design an infrared high dynamic range image compression and detail enhancement method to solve the problems of large detail loss in infrared high dynamic range image compression and low algorithm execution efficiency.

[0026] Reference Figure 1 As shown, the present invention discloses an infrared high dynamic range image compression and detail enhancement method, the steps are: the input infrared high dynamic range image is sent to the contrast enhancement branch and the detail enhancement branch for processing respectively, the obtained data is weightedly fused, and an 8-bit infrared image is output.

[0027] The contrast enhancement branch performs mean filtering, dynamic range mapping and limited contrast histogram equalization in sequence.

[0028] In the contrast enhancement branch, the input image is subjected to a 3×3 mean filter, and the filtering formula is: Where x and y are the pixel coordinates, I′(x,y) represents the pixel value after mean filtering, and Ω represents the neighborhood containing the pixel’s own coordinates (x,y).

[0029] In the contrast enhancement branch, the formula Perform dynamic range mapping on the mean filtered image, where g is the pixel value, Mean is the mean of the image after mean filtering, and Std is the standard deviation of the image after mean filtering.

[0030] In the contrast enhancement branch, the image after dynamic range mapping is enhanced by histogram equalization with limited contrast. The specific process is as follows.

[0031] (1) The image is divided into 64 regions with 8 rows and 8 columns.

[0032] (2) Calculate the histogram of each region.

[0033] (3) According to the threshold set in advance, the statistical number of single gray levels exceeding the threshold is evenly distributed to all gray levels, and distributed step by step from low to high.

[0034] (4) Calculate the cumulative probability histogram.

[0035] (5) For each pixel in the image, the enhanced grayscale value result is obtained using the bilinear interpolation method based on the cumulative probability histogram of the four adjacent regions.

[0036] The detail enhancement branch performs Laplace filtering, dynamic range mapping, zero erosion and stripe noise suppression in sequence.

[0037] In the detail enhancement branch, a 3×3 Laplace filter is performed on the input image, and the filtering formula is: Where x and y are pixel coordinates, I′(x,y) represents the pixel value after filtering, and Ω i Represents the neighborhood of coordinates (x,y) that does not include the pixel itself.

[0038] In the detail enhancement branch, the dynamic range mapping is performed on the Laplace filtered image, and the calculation formula is: Where g is the pixel value, Mean is the mean of the image after Laplace filtering, and Std is the standard deviation of the image after Laplace filtering.

[0039] In the detail enhancement branch, the image after dynamic range mapping is eroded by 3×3 zeroing using the following formula:

[0040]

[0041] Where min() and max() represent the minimum and maximum values ​​respectively.

[0042] In the detail enhancement branch, the stripe noise is suppressed on the image after zeroing corrosion, that is, the pixel mean of each row is first calculated, and then each pixel value in the row is subtracted from the pixel mean of the row.

[0043] By formula I ans (x,y)=I c (x,y)+0.5*I f (x, y) performs pixel-by-pixel weighted fusion on the images obtained by the contrast enhancement branch and the detail enhancement branch, and outputs the fused image, where I ans represents the output image, I c represents the contrast enhanced branch image, I f Represents detail enhanced branch image.

[0044] The above embodiments are only illustrative of the principles and effects of the present invention, as well as some embodiments of its application. For those skilled in the art, several modifications and improvements may be made without departing from the creative concept of the present invention, and all of these belong to the protection scope of the present invention.

Claims

1. An infrared high dynamic range image compression and detail enhancement method, characterized in that: Steps The input infrared high dynamic range image is sent to the contrast enhancement branch and detail enhancement branch for processing respectively, and the obtained data is weighted fused to output an 8-bit infrared image. The contrast enhancement branch sequentially performs mean filtering, dynamic range mapping and histogram equalization enhancement for limiting contrast; The detail enhancement branch sequentially performs Laplace filtering, dynamic range mapping, zero erosion and stripe noise suppression.

2. The infrared high dynamic range image compression and detail enhancement method according to claim 1, characterized in that: In the contrast enhancement branch, the formula The input image is subjected to a 3×3 mean filter, where x and y are pixel coordinates, I′(x, y) represents the pixel value after the mean filter, and Ω represents the neighborhood containing the pixel’s own coordinates (x, y).

3. The infrared high dynamic range image compression and detail enhancement method according to claim 2, characterized in that: In the contrast enhancement branch, the formula Perform dynamic range mapping on the mean filtered image, where g is the pixel value, Mean is the mean of the image after mean filtering, and Std is the standard deviation of the image after mean filtering.

4. The infrared high dynamic range image compression and detail enhancement method according to claim 3, characterized in that: In the contrast enhancement branch, the image after dynamic range mapping is subjected to histogram equalization with limited contrast through the following steps: (1) Divide the image into 64 regions with 8 rows and 8 columns; (2) Calculate the histogram of each region; (3) According to the threshold set in advance, the statistical number of single gray levels exceeding the threshold is evenly distributed to all gray levels, and distributed step by step from low to high; (4) Calculate the cumulative probability histogram; (5) For each pixel in the image, the enhanced grayscale value result is obtained using the bilinear interpolation method based on the cumulative probability histogram of the four adjacent regions.

5. The infrared high dynamic range image compression and detail enhancement method according to claim 1, 2, 3 or 4, characterized in that: The detail enhancement branch is implemented by formula Perform a 3×3 Laplace filter on the input image, where x and t are pixel coordinates, I′(x, y) represents the pixel value after filtering, and Ω i Represents the neighborhood of coordinates (x,y) that does not include the pixel itself.

6. The infrared high dynamic range image compression and detail enhancement method according to claim 5, characterized in that: The detail enhancement branch is implemented by formula Perform dynamic range mapping on the image after Laplace filtering, where g is the pixel value, Mean is the mean of the image after Laplace filtering, and Std is the standard deviation of the image after Laplace filtering.

7. The infrared high dynamic range image compression and detail enhancement method according to claim 6, characterized in that: In the detail enhancement branch, the image after dynamic range mapping is eroded by 3×3 zeroing by the following formula: Where min() and max() represent the minimum and maximum values ​​respectively.

8. The infrared high dynamic range image compression and detail enhancement method according to claim 7, characterized in that: In the detail enhancement branch, stripe noise is suppressed on the image after zeroing erosion, that is, the pixel mean of each row is first calculated, and then the pixel mean of the row is subtracted from each pixel value of the row.

9. The infrared high dynamic range image compression and detail enhancement method according to claim 8, characterized in that: By formula I ans (x,y)=I c (x,y)+0.5*I f (x, y) performs pixel-by-pixel weighted fusion on the images obtained by the contrast enhancement branch and the detail enhancement branch, and outputs the fused image, where I ans Represents the output image, I c represents the contrast enhanced branch image, I f Represents detail enhanced branch image.

Images