An underwater image restoration method based on polarization imaging and low-frequency filtering

Abstract

The present invention discloses an underwater image restoration method based on polarization imaging and low-frequency filtering, comprising the following steps: S1. Using an underwater polarization imaging experimental platform to collect polarization images of water bodies at three different angles of turbidity and calculate the Stokes vector of the underwater target scene; S2. Calculating the polarization degree and polarization angle image of the underwater target scene based on the Stokes vector, and establishing a target light model based on the relationship between the polarization information; S3. Based on the low-frequency characteristics of the background light, extracting the low-frequency background light component of the turbid polarization image at each angle and calculating the corresponding background light Stokes vector accordingly; S4. Substituting the processed parameters into the imaging model and combining them with enhanced measurement evaluation indicators to adaptively determine the optimal filter cutoff frequency, thus achieving high-quality underwater image restoration. The present invention utilizes this method to effectively restore texture features and edge details in more target areas.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an underwater image restoration method based on polarization imaging and low-frequency filtering. Background Art

[0002] Advances in underwater detection technology have made visual perception a core pillar of deep-sea exploration, ecological monitoring, and military security. However, water absorption, scattering, and light attenuation lead to widespread image distortion, reduced contrast, and blurred details, severely limiting target recognition and operational efficiency. Addressing this issue of image quality degradation is a core challenge and key direction for improving marine exploration capabilities and promoting the intelligent development of underwater equipment.

[0003] Underwater polarization imaging technology effectively suppresses interference from water-scattered light by extracting the polarization characteristics of the target and background, significantly improving underwater image contrast and target recognition capabilities. It has become an effective means of addressing underwater visual degradation. Current research focuses on the extraction and modeling of degree of polarization (DoP) information, achieving scatter suppression by establishing Stokes vector calculation models or developing polarization differential imaging algorithms. However, these methods generally suffer from insufficient utilization of angle of polarization (AoP) information, resulting in insufficient exploration of the background light AoP distribution characteristics and limited imaging stability in turbid water or complex lighting conditions. Therefore, how to effectively explore the role of AoP information in water scatter suppression and systematically integrate it into image restoration models has become a core issue that needs to be overcome to improve the quality of underwater polarization imaging. Summary of the Invention

[0004] The purpose of the present invention is to provide an underwater image restoration method based on polarization imaging and low-frequency filtering to solve the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides an underwater image restoration method based on polarization imaging and low-frequency filtering, comprising the following steps:

[0006] S1. Use the underwater polarization imaging experimental platform to collect polarization images of three angles of different turbid water bodies and calculate the Stokes vector of the underwater target scene;

[0007] S2. Calculate the polarization degree and polarization angle image of the underwater target scene based on the Stokes vector, and establish a target light model based on the right-angle triangle relationship of the polarization information;

[0008] S3. Based on the low-frequency characteristics of the background light, a frequency-domain Gaussian low-pass filter is used to extract the low-frequency background light components of the turbid polarization image at each angle, and the corresponding background light Stokes vector is calculated accordingly;

[0009] S4. By substituting the processed parameters into the target light model and combining the enhanced measurement evaluation indicators, the optimal cutoff frequency of the filter is adaptively determined to achieve high-quality underwater image restoration.

[0010] Preferably, the Stokes vector in S1 、 and The calculation formula is as follows:

[0011] ;

[0012] in, 、 and Polarization images of water bodies with different turbidity at three angles.

[0013] Preferably, by Stokes vector 、 and Calculate the polarization information of the underwater scene:

[0014] ;

[0015] in, Expressed as the degree of polarization of light in the image, Indicates the polarization direction of polarized light in the image, represents the degree of polarization, Represents the polarization angle.

[0016] Preferably, the S2 includes:

[0017] S21. Combined with the polarization information analysis, the right triangle relationship of the polarization information is as follows: the hypotenuse corresponds to the intensity of the polarized light part in the image , the two right-angled sides correspond to and Component, where the intensity of the polarized light part of the image and The angle formed by the components is ;

[0018] S22. According to the properties of right triangle, we can get the Height of side for:

[0019] ;

[0020] S23, based on polarization information and perpendicular to Derivation of the edge height to obtain background light The mathematical expression is:

[0021] ;

[0022] in, for The polarization is high, for The degree of polarization, for The polarization angle;

[0023] S24. Underwater images Considered as a clear target image S and background light that blurs the image Add together and we get:

[0024] ;

[0025] S25, Combination The calculation formula is derived to construct an underwater target light model to obtain a clear target image S Solution:

[0026] .

[0027] Preferably, the S3 includes:

[0028] S31, based on the low-frequency characteristics of background light, through the frequency domain Gaussian low-pass filter, from the input image 、 and Extract the corresponding low-frequency background light component ;

[0029] S32, based on the extracted background light image 、 and Derivation of the Stokes vector components of the background light 、 and ;

[0030] S33, through polarization information and perpendicular to The simultaneous operation of the formula for the height of the edge determines all unknown background light parameters 、 and Numerical solution of ;

[0031] S34, substituting the calculated background light Stokes vector into the target light model, performing parameter estimation and model inversion, and obtaining a preliminary restored image;

[0032] S35. Use the enhanced measurement evaluation value as the image quality evaluation standard, calculate the enhanced measurement evaluation values ​​at different cutoff frequencies to quantitatively evaluate the quality of the restored image, and select the parameter combination that maximizes the enhanced measurement evaluation value:

[0033] ;

[0034] in, Indicates the cutoff frequency when the enhanced measurement evaluation value is maximum;

[0035] S36, cut-off frequency Perform update iterations.

[0036] Preferably, the low-frequency background light component for:

[0037] ;

[0038] in, Indicates three different angles. Take 0, 60 and 120, F represents the Fourier transform, is the inverse Fourier transform, , Represented as a Gaussian low-pass filter in the frequency domain;

[0039] ;

[0040] in, represents the frequency domain coordinates, is the cutoff frequency of the filter.

[0041] Preferably, the step S36 specifically includes: repeating steps S31 to S35, and recording the enhanced measurement evaluation value and the corresponding cutoff frequency of each group of restoration results.

[0042] Preferably, the S4 includes:

[0043] S41, complete cutoff frequency After the update iteration, save the EME values ​​of all restored images and the corresponding cutoff frequencies;

[0044] S42, extract the cutoff frequency when the EME value is maximum , use the cutoff frequency to perform steps S31-S34 to obtain the optimal restoration result.

[0045] Therefore, the present invention adopts the above-mentioned underwater image restoration method based on polarization imaging and low-frequency filtering, which has the following beneficial effects:

[0046] (1) This paper constructs a new target light model based on the basic theory of underwater polarization imaging technology. By utilizing the low-frequency distribution characteristics of background light, adaptive frequency-domain Gaussian low-pass filtering is used to achieve background light separation and model parameter estimation, and finally the underwater polarization image is restored through parameter inversion.

[0047] (2) The present invention can effectively restore the texture features and edge details of more target areas while maintaining the stability of the overall image structure; this detail restoration capability has practical value in improving the task performance of machine vision application scenarios such as underwater target recognition and robot path planning.

[0048] (3) To improve the utilization of polarization information, the HoP parameter is introduced based on the right triangle relationship between the Stokes vector, DoP and AoP, and the background light expression of multi-polarization information is obtained; combined with the underwater imaging model, the solution of the target light is obtained.

[0049] (4) In order to accurately estimate the parameters in the target light model, the EME value is used to implement an adaptive frequency-domain Gaussian low-pass filtering method to extract the background light components of the polarization images at different angles for calculating the unknown parameters in the model.

[0050] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of an underwater image restoration method based on polarization imaging and low-frequency filtering according to an embodiment of the present invention;

[0052] Figure 2 This is a multi-angle underwater polarization diagram according to an embodiment of the present invention;

[0053] Figure 3 This is an image of DoP and AoP according to an embodiment of the present invention;

[0054] Figure 4 Schematic diagram of the right triangle relationship of polarization information according to an embodiment of the present invention;

[0055] Figure 5 This is a flowchart of step S4 of an embodiment of the present invention;

[0056] Figure 6 Comparison of the restoration results of each method; (a) represents the underwater image , (b) represents the CLAHE method, (c) represents the UDCP method, (d) represents the UDPLC method, and (e) represents the method of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0058] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0059] Example

[0060] like Figure 1 As shown, the present invention provides an underwater image restoration method based on polarization imaging and low-frequency filtering, comprising the following steps:

[0061] S1. Acquisition of underwater turbidity polarization images: Specifically, the underwater polarization imaging experimental platform is used to collect polarization images of three different turbidity water bodies, which are represented as 、 and ,like Figure 2 As shown, the Stokes vector of the underwater target scene is calculated 、 and as follows:

[0062] ;

[0063] The polarization information of the underwater scene is calculated using the Stokes vector:

[0064] ;

[0065] in, Expressed as the degree of polarization of light in the image, that is, amplitude information, Indicates the polarization direction of polarized light in the image, that is, phase information.

[0066] S2. AoP information underwater imaging modeling, specifically: refer to Figure 3 , the degree of polarization DoP and angle of polarization AoP images of the underwater target scene are calculated according to the Stokes vector, and the target light model is established according to the right triangle relationship of the polarization information.

[0067] In this embodiment, step S2 includes:

[0068] S21. Combining the formula analysis of polarization information, the right triangle relationship of polarization information can be corresponded to Figure 4 The right triangle structure shown: the hypotenuse corresponds to the intensity of the polarized light part in the image , the two right-angled sides correspond to and Quantity, of which and The angle formed by the components is .

[0069] S22. According to the properties of right triangle, we can get the Height of side for:

[0070] ;

[0071] S23, by introducing intermediate variables , here called the height of polarization (HoP) of the image, can be used to 、 and light intensity A direct correlation is established to supplement the imaging model (i.e., target light model) with richer polarization feature information. The derivation of the formula for the height of the edge can be used to obtain the background light of The mathematical expression is:

[0072] ;

[0073] in, for The polarization is high HoP, for DoP, for AoP.

[0074] S24. Underwater images Can be regarded as a clear target image S and background light that blurs the image Add together and we get:

[0075] ;

[0076] S25, Combination The calculation formula can be used to construct an underwater target light model and finally obtain a clear target image. S Solution:

[0077] ;

[0078] Just need to estimate 、 and The above formula can be used to restore underwater polarization images.

[0079] S3, adaptive frequency domain Gaussian low-pass filtering, specifically: based on the low-frequency characteristics of the background light, the frequency domain Gaussian low-pass filter is used to extract the low-frequency background light components of the turbid polarization image at each angle, respectively expressed as 、 and , and calculate the corresponding background light Stokes vector accordingly.

[0080] In this embodiment, step S3 includes:

[0081] S31, based on the low-frequency characteristics of background light, the frequency domain Gaussian low-pass filtering technology is used to 、 and Extract the corresponding low-frequency background light component :

[0082] ;

[0083] in, Indicates three different angles. Take 0, 60 and 120, F represents the Fourier transform, is the inverse Fourier transform, , Represented as a Gaussian low-pass filter in the frequency domain.

[0084] ;

[0085] in, represents the frequency domain coordinates, is the cutoff frequency of the filter.

[0086] S32, based on the extracted background light image 、 and The Stokes vector components of the background light can be further derived 、 and Input image 、 and After the frequency domain Gaussian low-pass filter, we get 、 and ,Will 、 and Substituting into the Stokes vector calculation formula, we get 、 and .

[0087] S33, through polarization information and perpendicular to The simultaneous operation of the formula for the height of the edge determines all unknown background light parameters 、 and Numerical solution of .

[0088] S34, substitute the calculated background light Stokes vector into the target light model, that is, substitute into the clear target image S In the formula, parameter estimation and model inversion are performed to obtain a preliminary restored image, that is, a clear target image S .

[0089] S35. In practical applications, The selection of directly affects the accuracy of background light separation. σ The enhanced measurement evaluation (EME) value is used as the image quality evaluation standard to automatically optimize the value. The restored image quality is quantitatively evaluated by calculating the EME value at different cutoff frequencies, and the parameter combination that maximizes the index is selected:

[0090] ;

[0091] in, Indicates the cutoff frequency at which the enhancement measurement evaluation value is maximum.

[0092] S36, cut-off frequency Perform update iteration, specifically: repeat steps S31-S35, and record the EME value and corresponding cutoff frequency of each set of restoration results.

[0093] S4, underwater polarization image restoration, specifically: by processing the parameters (i.e., background light parameters 、 and ) is substituted into the target light model, and the optimal cutoff frequency of the filter is adaptively determined in combination with the enhanced measurement evaluation index, ultimately achieving high-quality underwater image restoration. Figure 5 .

[0094] In this embodiment, step S4 includes:

[0095] S41, complete cutoff frequency After the update iteration, the EME values ​​of all restored images and the corresponding cutoff frequencies are saved.

[0096] S42, extract the cutoff frequency when the EME value is maximum σ , use the cutoff frequency to perform steps S31-S34 to obtain the optimal restoration result.

[0097] To comprehensively evaluate the performance of our method, polarization images of targets in two different turbid water environments were selected. Based on evaluation indicators such as SSIM, PSNR, PCQI, LIPIPS, UIQM and UCIQE, comparative experiments were conducted between our method (OUR) and three processing methods: CLAHE, UDCP and UDPLC. Figure 6 A visual comparison of the processing effects of different methods is shown, and Table 1 provides the quantitative analysis results of the experimental data. The specific values ​​​​reflect the objective performance differences of each algorithm.

[0098] Table 1 Quantitative evaluation results

[0099]

[0100] Therefore, the present invention adopts the above-mentioned underwater image restoration method based on polarization imaging and low-frequency filtering. This method optimizes the imaging model by integrating AoP information and combines the parameter adaptability of frequency domain filtering to significantly improve the contrast and clarity of the restored image, thereby improving the overall performance of other underwater vision tasks.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An underwater image restoration method based on polarization imaging and low-frequency filtering, characterized in that: The following steps are involved: S1. Use the underwater polarization imaging experimental platform to collect polarization images of three angles of different turbid water bodies and calculate the Stokes vector of the underwater target scene; S2. Calculate the polarization degree and polarization angle image of the underwater target scene based on the Stokes vector, and establish a target light model based on the right-angle triangle relationship of the polarization information; S3. Based on the low-frequency characteristics of the background light, a frequency-domain Gaussian low-pass filter is used to extract the low-frequency background light components of the turbid polarization image at each angle, and the corresponding background light Stokes vector is calculated accordingly; S4. Substituting the processed parameters into the target light model and combining them with the enhanced measurement evaluation index to adaptively determine the optimal cutoff frequency of the filter, high-quality underwater image restoration is achieved; the processed parameters are background light parameters 、 and ; The S2 includes: S21. Combined with the polarization information analysis, the right triangle relationship of the polarization information is as follows: the hypotenuse corresponds to the intensity of the polarized light part in the image , the two right-angled sides correspond to and Component, where the intensity of the polarized light part of the image and The angle formed by the components is ; S22. According to the properties of right triangle, we can get the Height of side for: ； S23, based on polarization information and the intensity of the polarized light portion perpendicular to the image Derivation of the edge height to obtain background light The mathematical expression is: ； in, for The polarization is high, for The degree of polarization, for The polarization angle; S24. Underwater images Considered as a clear target image S and background light that blurs the image Add together and we get: ； S25, Combination The calculation formula is derived to construct an underwater target light model to obtain a clear target image S Solution: 。 2. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 1, characterized in that: The Stokes vector in S1 、 and The calculation formula is as follows: ； in, 、 and Polarization images of water bodies with different turbidity at three angles.

3. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 2, characterized in that: Vector by Stokes 、 and Calculate the polarization information of the underwater scene: ； in, Expressed as the degree of polarization of light in the image, Indicates the polarization direction of polarized light in the image, represents the degree of polarization, Represents the polarization angle.

4. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 3, characterized in that: The S3 includes: S31, based on the low-frequency characteristics of background light, through the frequency domain Gaussian low-pass filter, from the input image 、 and Extract the corresponding low-frequency background light component ; S32, based on the extracted background light image 、 and Derivation of the Stokes vector components of the background light 、 and ; S33, through polarization information and perpendicular to The simultaneous operation of the formula for the height of the edge determines all unknown background light parameters 、 and Numerical solution of ; S34, substituting the calculated background light Stokes vector into the target light model, performing parameter estimation and model inversion, and obtaining a preliminary restored image; S35. Use the enhanced measurement evaluation value as the image quality evaluation standard, calculate the enhanced measurement evaluation values ​​at different cutoff frequencies to quantitatively evaluate the quality of the restored image, and select the parameter combination that maximizes the enhanced measurement evaluation value: ； in, Indicates the cutoff frequency when the enhanced measurement evaluation value is maximum; S36, cut-off frequency Perform update iterations.

5. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 4, characterized in that: The low-frequency background light component for: ； in, Indicates three different angles. Take 0, 60 and 120, F represents the Fourier transform, is the inverse Fourier transform, , Represented as a Gaussian low-pass filter in the frequency domain; ； in, represents the frequency domain coordinates, is the cutoff frequency of the filter.

6. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 5, characterized in that: The S36 specifically includes: repeating steps S31 to S35, and recording the enhanced measurement evaluation value and the corresponding cutoff frequency of each group of restoration results.

7. The underwater image restoration method based on polarization imaging and low-frequency filtering according to claim 6, characterized in that: The S4 includes: S41, complete cutoff frequency After the update iteration, save the EME values ​​of all restored images and the corresponding cutoff frequencies; S42, extract the cutoff frequency when the EME value is maximum , use the cutoff frequency to perform steps S31-S34 to obtain the optimal restoration result.

Images