Infrared super lens image enhancement method based on dark channel prior

Abstract

The invention discloses an infrared super lens image enhancement method based on dark channel prior, and relates to the technical field of image enhancement. The method comprises the following steps: firstly, calculating a dark channel graph corresponding to an infrared super lens image; estimating the scattering light intensity of the super lens by using the dark channel map; establishing an image degradation model of the super lens; calculating the transmissivity of the super lens according to the model, the scattering light intensity and the dark channel graph; substituting the scattered light intensity and the transmissivity into the model to obtain an enhanced image; and finally, restoring the target brightness of the enhanced image in combination with the input image information to obtain a final output image. By using the method provided by the invention, the definition and contrast of the infrared super-lens image can be improved in real time in a single frame, and the method is easy to integrate in hardware equipment for deployment.

Description

Technical Field

[0001] This application belongs to the field of image processing and is an infrared superlens image enhancement method based on dark channel prior Background Art

[0002] A superlens is a lens made of a metasurface. The metasurface consists of a layer of sub-wavelength-scale nanostructure unit arrays that can precisely control the phase, amplitude, and polarization state of light. By designing the shape, size, and arrangement of these nanostructures, the superlens can achieve customized control of the light wavefront, thereby replacing traditional refractive lenses for optical functions such as focusing, imaging, and beam shaping.

[0003] As a new type of optical element, superlens is an ideal choice for reducing the weight and cost of infrared optical systems due to its ultra-thinness, lightness, high design flexibility, and convenience for large-scale mass production. However, its inherent dispersion problem will lead to chromatic aberration and low transmittance, which will significantly reduce the imaging quality, especially in terms of resolution, contrast, geometry and details. These problems limit the widespread use of superlenses in practical applications.

[0004] In the existing methods, some scholars have enhanced infrared super-lens images by training neural network models through deep learning, but such methods require a large number of paired data sets, have a long inference time, are not real-time, and are difficult to integrate and deploy into hardware devices. Therefore, the present invention proposes to use a dark channel prior-based method to enhance super-lens infrared images, effectively remove image blur, and increase the clarity and contrast of images in real time in a single frame. Summary of the invention

[0005] The present invention aims at the problem of blur noise in infrared super-lens images due to the dispersion and low transmittance of super-lens. First, the dark channel map of the input infrared super-lens image is calculated, and the scattered light intensity of the super-lens and the transmittance of the super-lens are estimated based on it. Then, the enhanced infrared super-lens image is inversely calculated using the super-lens image degradation model established by us, and a deblurred image with uniform target intensity is obtained. Finally, our brightness enhancement technology combined with the original image information is used to restore the brightness of the enhanced infrared super-lens image, thereby improving the clarity and contrast of the infrared super-lens image.

[0006] On the basis of the above, the present invention provides an infrared superlens image enhancement method based on dark channel prior, including the following technical solutions:

[0007] Step 1. Calculate the dark channel map of the input image;

[0008] Step 2. Estimate the scattered light intensity of the metalens;

[0009] Step 3. Establish an image degradation model of the superlens;

[0010] Step 4. Calculate the transmittance of the metalens;

[0011] Step 5. Enhance the image using the image degradation model of the superlens;

[0012] Step 6. Perform brightness restoration on the enhanced image to obtain the output image.

[0012] Preferably, the step 1 specifically includes:

[0013] As a preferred solution of the infrared superlens image enhancement method based on dark channel prior of the present invention, the calculation of the dark channel map of the input image includes selecting the minimum channel of the image and applying a minimum filter: The formula for calculating the dark channel map is: Among them, I dark (x) is the dark channel image; I c (y) is a channel of the input image; Ω(x) is the neighborhood of pixel x.

[0014] Preferably, the step 2 specifically includes:

[0015] As a preferred embodiment of the infrared superlens image enhancement method based on dark channel prior described in the present invention, the estimation of the superlens scattered light intensity includes finding the first n% of the brightest pixels in the dark channel image, finding these pixels with the highest brightness in the original image, and selecting the pixel with the highest brightness among these pixels as the scattered light intensity A.

[0020] Preferably, the step 3 specifically includes:

[0021] As a preferred solution of the infrared super-lens image enhancement method based on dark channel prior described in the present invention, the image degradation model of the super-lens is established, The image degradation model formula of the super lens is: I(x=J(x)t(x)+A(1-t(x))(1-J(x)) (2) Where I(x) is the normalized degenerate infrared superlens image; t(x) is the transmittance; J(x) is the normalized ideal infrared superlens image; A is the scattered light intensity.

[0016] Preferably, the step 4 specifically includes:

[0017] As a preferred scheme of the infrared superlens image enhancement method based on dark channel prior described in the present invention, the calculation of the transmittance of the superlens includes calculating the transmittance using the dark channel map and the estimated scattered light intensity, and applying guided filtering to optimize the transmittance using the original image as a guided map.

[0018] As a preferred solution of the infrared superlens image enhancement method based on dark channel prior described in the present invention, the calculation of transmittance using the dark channel map and the estimated scattered light intensity includes: The formula for calculating transmittance is: Where t(x) is the transmittance; I dark (x) is the calculated dark channel image; A is the estimated scattered light intensity; ω∈(0,1).

[0019] As a preferred solution of the infrared superlens image enhancement method based on dark channel prior of the present invention, the application of guided filtering to optimize the transmittance using the original image as a guided image includes calculating the local coefficients and using the guided filtering to optimize: The local coefficient calculation formula is as follows: Where t(x) is the rough transmittance map; I is the guide map; cov(I,p) is the covariance of I and p; is the variance of I; ∈ is a regularization parameter to avoid division by zero; and are the means of t(x) and I respectively; The guided filtering formula is as follows: in, is the optimized transmittance map; I is the guide map; a i and b i is the local linear coefficient; w is the local window, size is r×r; w is the size of the window (i.e. the number of pixels).

[0020] Preferably, the step 5 specifically includes:

[0021] As a preferred embodiment of the infrared superlens image enhancement method based on dark channel prior of the present invention, the image enhancement using the image degradation model of the superlens includes: Substitute it into the image degradation model formula of the superlens and inversely solve for the enhanced image: The inverse enhanced image calculation formula is: Where, J(x) is the enhanced image; For optimized transmittance; A is the scattered light intensity; I(x) is the normalized original image; t 0 is the lower bound of the transmittance.

[0022] Preferably, the step 6 specifically includes:

[0023] As a preferred solution of the infrared superlens image enhancement method based on dark channel prior described in the present invention, the brightness restoration of the enhanced image to obtain the output image includes a brightness enhancement technology combined with the original image information, The formula is: in, is the result after final brightness restoration; J(x) is the enhanced result of the modified model; α is the linear brightness restoration coefficient.

[0024] Beneficial effects of the present invention:

[0025] Enhance the clarity and contrast of infrared super-lens images: The present invention establishes a degradation model for super-lens images to address image blurring caused by super-lens dispersion and low transmittance. The infrared super-lens image is enhanced using dark channel priors and the established model, so that the enhanced infrared super-lens image has a balanced target intensity while removing blur, and the brightness enhancement technology combined with the original image information is used to increase the clarity and contrast of the infrared super-lens image after deblurring. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is an implementation flow chart of the present invention.

[0027] Figure 2 The figure is a result comparison diagram of an embodiment of the method of the present invention.

[0028] Figure 3 This is an experimental comparison diagram of the method of the present invention and other methods. DETAILED DESCRIPTION

[0028] The following is combined with Figure 1-2 This application is described in further detail.

[0029] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0030] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0031] The present application embodiment provides an infrared superlens image enhancement method based on dark channel prior, referring to Figure 1 , the method comprising:

[0032] Step 1: Calculate the dark channel map of the input infrared superlens image.

[0033] The input image is an infrared superlens image with only one channel, read as a grayscale image, and a minimum filter is applied: The formula for calculating the dark channel map is: Among them, I dark (x) is the dark channel image; I(y) is the grayscale image of the read infrared superlens image; Ω(x) is the neighborhood of pixel x, and in the embodiment of the present application, the size is 15×15.

[0034] Step 2: Estimate the intensity of scattered light from the metalens.

[0035] First, find the brightest 0.1% of pixels in the dark channel image, then find the pixels with the highest brightness in the original image, and select the pixel with the highest brightness among these pixels as the scattered light intensity A.

[0039] Step 3: Establish the image degradation model of the metalens.

[0040] We divide the image taken by the superlens into two parts, one is the scene light after the superlens attenuation, and the other is the scattered light generated by the dispersion of the superlens through the superlens: The image degradation model formula of its super lens is: I(x)=J(x)t(x)+A(1-t(x))(1-J(x)) (2) Where I(x) is the normalized degenerate infrared superlens image; t(x) is the transmittance; J(x) is the normalized ideal infrared superlens image; A is the scattered light intensity.

[0036] Step 4: Calculate the transmittance of the metalens.

[0037] First, use the calculated dark channel image I dark (x) and the estimated scattered light intensity A to calculate the transmittance graph t(x): The formula for calculating the transmittance map is: Where t(x) is the transmittance; I dark (x) is the calculated dark channel image; A is the estimated scattered light intensity; In this embodiment, ω is 0.95.

[0038] Then the guided filter is applied to optimize the calculated transmittance map t(x) using the original image I(x) as the guided image: First, calculate the local coefficient, the formula is: Where t(x) is the rough transmittance map; I is the guide map; cov(I,p) is the covariance of I and p; is the variance of I; ∈ is a regularization parameter to avoid division by zero; and are the means of t(x) and I respectively; Then use guided filtering to optimize transmittance, the formula is: in, is the optimized transmittance map; I is the guide map; a i and b i is the local linear coefficient; w is the local window, which is 60×60 in this embodiment; w is the size of the window (i.e. 3600).

[0039] Step 5: Enhance the image using the image degradation model of the superlens.

[0040] The estimated scattered light intensity A and the optimized transmittance Substitute it into the image degradation model of the superlens and inversely calculate the enhanced image: The calculation formula of the enhanced image calculated inversely is: Where, J(x) is the enhanced image; For optimized transmittance; A is the scattered light intensity; I(x) is the normalized original image; t 0 is the set lower limit of transmittance, which is 0.01 in this embodiment.

[0041] Step 5: Perform brightness restoration on the enhanced image to obtain the output image.

[0042] The enhanced image is subjected to a brightness enhancement technique combined with the original image information, and finally an enhanced infrared superlens image with restored brightness is obtained. The formula for brightness recovery is: in, is the result after final brightness restoration; J(x) is the enhanced result of the modified model; α is the linear brightness recovery coefficient, which is 7 in this embodiment.

[0043] Figure 2 The comparison between the results obtained in this embodiment and the original image is shown. The first line is the input original infrared super-lens image, and the second line is the enhanced infrared super-lens image. Through comparative experiments, the method in this paper is compared with histogram equalization (HE), multi-scale Retinex (MSR), and dark channel algorithm (DCP), and is quantitatively evaluated using evaluation indicators such as linear ambiguity index (LIF), logarithmic Michelson contrast (AME), enhanced mean edge energy (EMEE), gradient entropy (GE) and edge strength (ES). The results show that the method of the present invention has achieved significant improvements in data indicators and enhanced the contrast and clarity of infrared super-lens images.

[0044]

[0045] The above is only a preferred implementation of an infrared super-lens image enhancement method based on dark channel prior. The protection scope of an infrared super-lens image enhancement method based on dark channel prior is not limited to the above embodiment. All technical solutions under this idea belong to the protection scope of the present invention. It should be pointed out that for those skilled in the art, several improvements and changes without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An infrared superlens image enhancement method based on dark channel prior, characterized in that: The steps include: Step 1. Calculate the dark channel map of the input image; Step 2. Estimate the scattered light intensity of the metalens; Step 3. Establish an image degradation model of the superlens; Step 4. Calculate the transmittance of the metalens; Step 5. Enhance the image using the image degradation model of the superlens; Step 6. Perform brightness restoration on the enhanced image to obtain the output image.

2. The infrared superlens image enhancement method based on dark channel prior according to claim 1 is characterized in that: Calculating the dark channel map of the input image in step 1 includes selecting the minimum channel of the image and applying a minimum filter; The formula for calculating the dark channel map is: Among them, I dark (x) is the dark channel image; I c (y) is a channel of the input image; Ω(x) is the neighborhood of pixel x.

3. The infrared superlens image enhancement method based on dark channel prior according to claim 1 is characterized in that: The estimation of the scattered light intensity of the metalens in step 2 includes finding the brightest first n% of pixels in the dark channel image, finding the pixel with the highest brightness in the input image, and selecting the pixel with the highest brightness among the pixels as the scattered light intensity A.

4. The infrared superlens image enhancement method based on dark channel prior according to claim 1 is characterized in that: The image degradation model of the superlens is established in step 3, including: The image degradation model formula of the superlens is: I(x)=J(x)t(x)+A(1-t(x))(1-J(x)) (2) Where I(x) is the normalized degenerate infrared superlens image; t(x) is the transmittance of the metalens; J(x) is the normalized ideal infrared superlens image; A is the scattered light intensity.

5. The infrared superlens image enhancement method based on dark channel prior according to claim 1, characterized in that: The calculation of the transmittance of the metalens in step 4 includes calculating the transmittance of the metalens using the dark channel image and the estimated scattered light intensity, and applying guided filtering to optimize the transmittance using the original image as a guided image.

6. The infrared superlens image enhancement method based on dark channel prior according to claim 5 is characterized in that: The method of calculating the transmittance of the metalens using the dark channel image and the estimated scattered light intensity includes: The formula for calculating transmittance is: Where, t(x) is the transmittance; I dark (x) is the calculated dark channel image; A is the estimated scattered light intensity; ω∈(0,1).

7. The infrared superlens image enhancement method based on dark channel prior according to claim 5, characterized in that: The applying guided filtering to optimize the transmittance using the input image as a guide map includes calculating local coefficients and optimizing using guided filtering; The local coefficient calculation formula is as follows: Where t(x) is the rough transmittance map; I is the guide map; cov(I, p) is the covariance of and p; is the variance of ; ∈ is the regularization parameter to avoid division by zero; and are the means of t(x) and I respectively; The guided filtering formula is as follows: in, is the optimized transmittance map; I is the guide map; a i and b i is the local linear coefficient; w is the local window, the size of which is r×r; |w| is the size of the window, i.e. the number of pixels.

8. The infrared superlens image enhancement method based on dark channel prior according to claim 1, characterized in that: The step 5 of enhancing the image using the image degradation model of the metalens includes: Substitute it into the image degradation model formula of the superlens and inversely solve for the enhanced image: The enhanced image calculation formula is: Where, J(x) is the enhanced image; For optimized transmittance; A is the scattered light intensity; I(x) is the normalized original image; t0 is the set lower limit of transmittance.

9. The infrared superlens image enhancement method based on dark channel prior according to claim 1, characterized in that: The step 6 restores the brightness of the enhanced image to obtain the output image, including combining the brightness enhancement technology of the original image information, The formula is: in, is the result after final brightness restoration; J(x) is the enhanced result of the modified model; α is the linear brightness restoration coefficient.

Images