Target polarization characteristic combined transmissivity estimation defogging method

Abstract

The invention provides a transmittance estimation defogging method in combination with target polarization characteristics, and belongs to the technical field of image processing. The method comprises the following steps: collecting polarization images of a shooting target at multiple angles; calculating the Stokes vector and the degree of polarization of the probe light based on the polarization image; segmenting a heavy fog area and a non-heavy fog area in combination with the light intensity information and the polarization information, and determining the maximum value of the target light, the minimum value of the target light and the light intensity value at the infinity; building a mathematical model of the detection light and the target light in combination with the transmissivity, and determining a mathematical model of the maximum transmissivity and the minimum transmissivity; determining the maximum transmissivity and the minimum transmissivity by combining the maximum value of the target light, the minimum value of the target light and the light intensity value at the infinity based on the mathematical models of the maximum transmissivity and the minimum transmissivity; and obtaining a target light image through an atmospheric scattering model by taking the maximum transmissivity and the minimum transmissivity as constraint conditions. The method is high in target recovery quality, is closer to a natural state, and does not generate obvious distortion.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a transmittance estimation defogging method combined with target polarization characteristics. BACKGROUND

[0002] In a clear sky, it is relatively simple to take a high-definition target image, but in a hazy medium such as heavy fog, the absorption, scattering and reflection characteristics of electromagnetic waves of different wavelengths are severely disturbed by the hazy medium, and the quality of the image will be greatly reduced.

[0003] In the prior art, in order to realize image defogging, many methods have been proposed. The method based on polarization imaging utilizes more light information to perform defogging processing on the image, and has the advantages of simple algorithm, strong defogging ability in heavy fog weather and the like. The polarization-based defogging scheme collects the difference imaging of two polarized images by rotating a linear polarizer to realize defogging.

[0004] However, the existing polarization defogging method takes the polarization information in the image as the polarization information of the air light, and performs polarization image defogging enhancement by continuously optimizing and solving the light intensity value at infinity. This results in limitations of the polarization defogging method in some scenarios. Because in practice, the polarization information is provided by the atmospheric light and the target light together.

[0005] Therefore, there is a need for a transmittance estimation defogging method combined with target polarization characteristics. SUMMARY

[0006] Therefore, the present application provides a transmittance estimation defogging method combined with target polarization characteristics, which determines more accurate light intensity values at infinity and constraint ranges of target light by segmenting heavy fog regions and non-heavy fog regions, obtains more accurate defogged images by combining an atmospheric scattering model.

[0007] To this end, the present application provides the following technical solutions: A transmittance estimation defogging method combined with target polarization characteristics, comprising: acquiring polarization images of the target at multiple angles; calculating Stokes vectors and degrees of polarization of the probe light based on the polarization images; segmenting heavy fog regions and non-heavy fog regions combined with light intensity information and polarization information, and determining maximum values of target light, minimum values of target light and light intensity values at infinity; constructing mathematical models of the probe light and the target light combined with transmittance, and determining mathematical models of maximum transmittance and minimum transmittance; determining the maximum transmittance and the minimum transmittance based on a mathematical model of the maximum transmittance and the minimum transmittance, the maximum value of the target light, the minimum value of the target light and the light intensity value at infinity; obtaining the target light image through the atmospheric scattering model with the maximum transmittance and the minimum transmittance as constraint conditions.

[0008] Further, the method further comprises performing gray scale stretching on the target light image to realize image enhancement.

[0009] Further, the mathematical model of the probe light and the target light comprises:

[0010] wherein, is the light intensity of the probe, is the light intensity value at infinity, is the light intensity value of the target light, is the transmittance, is the degree of polarization of the atmosphere, is the degree of polarization of the target, is the degree of polarization of the probe light.

[0011] Further, the method further comprises: regarding the region with the light intensity value greater than a preset threshold and the degree of polarization less than a preset threshold as the heavy fog region, and regarding the remaining part as the non-heavy fog region.

[0012] Further, the method further comprises: regarding the window with the highest average light intensity value in the heavy fog region as the light intensity value at infinity; regarding the maximum value of the average window light intensity in the non-heavy fog region as the maximum value of the target light; regarding the minimum value of the average window light intensity in the non-heavy fog region as the minimum value of the target light.

[0013] Further, the mathematical model of the maximum transmittance and the minimum transmittance comprises:

[0014]

[0015] wherein, is the minimum transmittance, is the maximum transmittance.

[0016] Further, the plurality of angle polarization images comprises: 0-degree, 45-degree, 135-degree and 90-degree polarization images.

[0017] The advantages and positive effects of the present application are as follows: The present method determines the light intensity value at infinity by dividing the heavy fog area and the non-heavy fog area, solves the problem that the traditional method fails to solve the sky area under the light fog condition, and leads to inaccurate image restoration.

[0018] The present method uses the maximum value and the minimum value of the average window light intensity of the non-heavy fog area as the maximum value and the minimum value of the target light for boundary constraint, so that the solved target light is more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0020] Figure 1 The logic diagram of the transmittance estimation defogging method combined with the target polarization characteristics in the embodiment 1 of the present application; Figure 2 The heavy fog light intensity diagram under different scenes in the embodiment 2 of the present application; Figure 3 The light intensity threshold diagram under different scenes in the embodiment 2 of the present application; Figure 4 The polarization threshold diagram under different scenes in the embodiment 2 of the present application; Figure 5 The threshold fusion diagram under different scenes in the embodiment 2 of the present application; Figure 6 The atmospheric window diagram under different scenes in the embodiment 2 of the present application; Figure 7 The transmittance estimation filter pseudo-color diagram under different scenes in the embodiment 2 of the present application; Figure 8 The defogging final diagram under different scenes in the embodiment 2 of the present application; Figure 9 The light intensity pseudo-color diagram under different scenes in the embodiment 2 of the present application; Figure 10 The defogging pseudo-color diagram under different scenes in the embodiment 2 of the present application; Figure 11 The flowchart of the transmittance estimation defogging method combined with the target polarization characteristics in the embodiment 2 of the present application. DETAILED DESCRIPTION

[0021] In the following, the technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work should belong to the protection scope of the present application.

[0022] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product, or device.

[0023] The present application provides a transmittance estimation defogging method combined with target polarization characteristics, determines the sky region through multi-scale fusion of polarization segmentation map and light intensity segmentation map. The maximum and minimum values of transmittance are estimated through the segmented heavy fog region and non-heavy fog region; the target light is solved through the boundary constraint of the target light boundary value of the non-heavy fog region.

[0024] Embodiment 1 In combination Figure 1 As shown in the figure, a transmittance estimation defogging method combined with target polarization characteristics comprises: S1, obtaining images with different polarization directions through a polarization camera Corresponding images ; The Stokes vector of the probe light is calculated through the images with different polarization directions, and the calculation formula is:

[0025] Among them, I0 represents the light intensity value corresponding to the 0-degree polarization direction, I45 represents the light intensity value corresponding to the 45-degree polarization direction, I90 represents the light intensity value corresponding to the 90-degree polarization direction, I135 represents the light intensity value corresponding to the 135-degree polarization direction; I represents the light intensity value, I0-90 represents the difference between the light intensity of the 0-degree component and the 90-degree component, represents the difference between the light intensity of the 45-degree component and the 135-degree component; represents the Stokes vector of the probe light.

[0026] S2, based on the Stokes vector of the probe light, the degree of polarization of the probe light is calculated, which is expressed by the formula:

[0027] wherein, represents the degree of polarization of the probe light.

[0028] S3, the target in the image usually has obvious polarization characteristics, which can accurately reflect the characteristic information of the target, and the sky area has very low polarization characteristics, which is usually much smaller than the polarization characteristics of the target.

[0029] Therefore, in heavy fog weather, the polarization characteristics obtained are considered to be provided by the target light and the atmospheric light together, and a mathematical model of the probe light and the target light is constructed by combining the atmospheric scattering model and the transmittance:

[0030] wherein, is the light intensity of the detector, is the light intensity value at infinity, is the light intensity value of the target light, is the transmittance, is the degree of polarization of the atmosphere, is the degree of polarization of the target, is the degree of polarization of the probe light.

[0031] S4, in the actual fog image, not all areas are covered by heavy fog, and the transmittance of the area covered by heavy fog is relatively low. In order to solve the target light to the greatest extent, the minimum transmittance usually needs to be estimated in the heavy fog area, and the maximum transmittance usually needs to be estimated in the non-heavy fog area.

[0032] Due to the influence of heavy fog, the polarization component of the probe light is usually smaller than that of the target, that is, and according to the similarity of the target in the image scene, the minimum value of the target light is obtained, and the minimum transmittance of the sky area is solved; And for the non-heavy fog area, the transmittance is relatively high, and the influence of the air light is small. In order to ensure that the target light is not distorted, the maximum transmittance is usually solved. Since in the case of no fog, , the maximum transmittance of the non-sky area is obtained by solving the maximum value of the target light.

[0033] Mathematical model of maximum transmittance and minimum transmittance:

[0034]

[0035] wherein, is the minimum transmittance, is the maximum transmittance.

[0036] S4, combine light intensity information and polarization information to segment out heavy fog area and non-heavy fog area, determine the light intensity value at infinity, the degree of polarization of the atmosphere, and the maximum value of the target light and the minimum value of the target light.

[0037] 1) A multi-scale threshold fusion method is used to determine the heavy fog area and the non-heavy fog area: threshold segmentation is performed on the light intensity scale and the polarization scale to obtain the heavy fog area and the non-heavy fog area. The heavy fog area is the region where the light intensity scale is a high-light region and the polarization scale is a dark region, and the remaining part is the non-heavy fog area, i.e. the region where the light intensity value is greater than a preset threshold and the degree of polarization is less than a preset threshold is the heavy fog area, and the remaining part is the non-heavy fog area.

[0038] 2) In the heavy fog area, the window with the highest average light intensity value is taken as the sky window, and the maximum light intensity value of the window is taken as the light intensity value at infinity , and the average polarization degree is taken as the polarization degree of the atmosphere .

[0039] 3) The key to solving the maximum transmittance and the minimum transmittance is to obtain the maximum value and the minimum value of the target light for boundary constraint. According to the structural similarity of the scene, the average window light intensity maximum value of the non-heavy fog area is taken as the maximum value of the target light , and the average window light intensity minimum value of the non-heavy fog area is taken as the minimum value of the target light for boundary constraint.

[0040] After determining the maximum value and the minimum value of the target light, the transmittance is estimated through the relationship, and the estimated transmittance and are obtained.

[0041] Due to the characteristics of the transmittance itself, it is within the range of 0 to 1. In order to further reduce the error, the estimated transmittance is subjected to mean filtering to obtain the transmittance, which is represented by the formula:

[0042] S5, based on the transmittance, the target light image is solved through the atmospheric scattering model, which is represented by the formula:

[0043] wherein, is the fog image captured by the detector; is the light intensity value at infinity; is the transmittance.

[0044] Embodiment 2 S1, using a tripod and a gimbal, combined with a polarization camera to build a shooting platform. Each polarization unit is 2*2 pixels, arranged in 0, 45, 90, 135 clockwise. They are arranged according to the repeated polarization unit. Four different component polarization images are obtained, the Stokes vector is calculated, and the polarization degree of each pixel is calculated.

[0045] S2, in order to ensure the accuracy of the defogging, the polarization degree of the sky area needs to be determined. The traditional method for solving the sky area is usually to select the area with the maximum light intensity as the sky area. This is because in heavy fog conditions, the air light is usually much larger than the target light. But this method is not accurate, when the light is light or the target light is greater than the air light, this method fails. In order to solve this situation, this method determines the sky area by fusing the light intensity threshold and the polarization threshold.

[0046] S3, the key to solving the maximum transmittance and the minimum transmittance is to obtain the maximum and minimum values of the target light to perform boundary constraint. According to the structural similarity of the scene, the maximum value and the minimum value of the average window light intensity of the non-fog area are selected as the maximum value and the minimum value of the target light to perform boundary constraint.

[0047] S4, after obtaining the maximum value and the minimum value of the target, the estimated transmittance is obtained by the transmittance formula. Because of the characteristics of the transmittance itself, it is within the range of 0 to 1. Therefore, in order to further reduce the error, the estimated transmittance is mean filtered to obtain the transmittance.

[0048] S5, using the atmospheric scattering model formula, the image of the target light is obtained.

[0049] In heavy fog conditions, the light intensity value of the target light is small, and the gray scale stretching is performed on the target light to improve the brightness, so that the image is clearer.

[0050] The image processing results in this embodiment are as follows Figures 2-10The light intensity map and the dehazed image are compared more intuitively by using a pseudo-color map. In the light intensity map, it can be seen that the atmospheric light interferes with the target light under heavy fog conditions. The contrast of the image is significantly improved, and the target light covered by the atmospheric light is well restored. In order to more objectively reflect the quality of the image, four image evaluation indexes, including information entropy, average gradient, image standard deviation, and SSIM (Structural Similarity Index Measure), are used to evaluate the quality of the image. The information entropy indicates the amount of information in the image. The larger the value, the more information the image contains. The average gradient reflects the rate of change of the small details of the image, and can reflect the clarity of the image to a certain extent; the image standard deviation reflects the dispersion degree of the image pixel value and the mean value, and the larger the standard deviation value, the clearer the image edge. SSIM evaluates the similarity of two images from three aspects of brightness, contrast, and structure. The algorithm in this paper maintains the maximum similarity while having a high information entropy and average gradient, which reflects the effectiveness of the dehazing algorithm. The numerical evaluation results are shown in Table 1. According to the results, the method is suitable for most scenes, and the target recovery quality is high, closer to the natural state, and does not produce obvious distortion.

[0051] Table 1

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for part or all of the technical features; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for dehazing by combining transmittance estimation with target polarization characteristics, characterized in that, include: Acquire polarization images of the target from multiple angles; The Stokes vector and degree of polarization of the probe light are calculated based on the polarization image. By combining light intensity information and polarization information, the foggy region and the non-foggy region are segmented, and the maximum value, minimum value and light intensity value at infinity of the target light are determined. Mathematical models of the probe light and the target light are constructed by combining transmittance, and mathematical models for the maximum and minimum transmittance are determined. Based on the mathematical model of maximum and minimum transmittance, the maximum and minimum transmittance are determined by combining the maximum value of the target light, the minimum value of the target light, and the light intensity value at infinity. Using maximum and minimum transmittance as constraints, the target light image is obtained through an atmospheric scattering model.

2. The method according to claim 1, characterized in that, It also includes grayscale stretching of the target light image to achieve image enhancement.

3. The method according to claim 1, characterized in that, The mathematical models for the probe light and the target light are as follows: in, It is the light intensity of the detector. It is the light intensity value at infinity. It is the light intensity value of the target light. It's transmittance. It is the degree of atmospheric polarization. It is the polarization degree of the target. It detects the degree of polarization of light.

4. The method according to claim 1, characterized in that, The process of segmenting foggy and non-foggy regions by combining light intensity and polarization information includes: The area with light intensity greater than a preset threshold and polarization less than a preset threshold is designated as a foggy area, and the remaining area is designated as a non-foggy area.

5. The method according to claim 4, characterized in that, Determining the maximum value, minimum value, and intensity value of the target light at infinity includes: The window with the highest average light intensity in the foggy area is taken as the light intensity value at infinity. The maximum average window light intensity in the non-foggy area is taken as the maximum value of the target light; The minimum average window light intensity in the non-foggy area is taken as the minimum target light intensity.

6. The method according to claim 3, characterized in that, Mathematical models for maximum and minimum transmittance include: in, Minimum transmittance, This represents the maximum transmittance.

7. The method according to claim 1, characterized in that, The polarization images at multiple angles include: Polarization images at 0 degrees, 45 degrees, 135 degrees, and 90 degrees.

Images