Underwater polarization imaging method based on combination of frequency domain processing and global estimation
Through frequency domain processing and global estimation technology, the problem of backscattered light interference in underwater polarization imaging is solved, and efficient image recovery is achieved. It is especially suitable for static and moving targets in complex scattering environments, improving image quality and processing automation.
Patent Information
- Application Number
- CN202411110461.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-08-14
AI Technical Summary
Existing underwater polarization imaging technology is difficult to effectively remove backscattered light in complex scattered environments, resulting in a decline in image quality, especially in the case of moving targets, and often requires human-computer interaction and background prior information, which affects processing speed and accuracy. sex.
Using a method based on frequency domain processing combined with global estimation, the orthogonal polarization image is obtained through histogram stretching and high-pass filtering preprocessing, the polarization degree of backscattered light is accurately estimated using frequency domain low-pass filter and optimal threshold factor, and The descattering effect was evaluated by augmentation metrics, and image restoration was finally performed in combination with the physical model.
Without human-computer interaction, the backscattered light is effectively removed, which improves the contrast and recovery quality of underwater images, especially in complex scattering environments with significant restoration effects.
Smart Images

Figure CN119027347A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image enhancement, and in particular to an underwater polarization imaging method based on frequency domain processing combined with global estimation. Background Art
[0002] Affected by the water itself and turbid media such as suspended particles and soluble organic matter, the light containing the target information is absorbed and strongly scattered, resulting in increased noise and a significant decrease in contrast in underwater images. Under the scattering effect, the interference of backscattered light is the main reason for the degradation of image quality. Therefore, overcoming the influence of backscattered light on underwater optical imaging is a work worth studying. Since backscattered light is partially polarized light, the underwater polarization imaging method is proposed as an underwater optical imaging technology that effectively suppresses backscattered light.
[0003] Previous polarization imaging descattering methods usually require prior information of the background area and rarely consider the impact of light field inhomogeneity on image restoration, which not only reduces the processing speed of imaging, but also introduces errors in image restoration, especially for moving targets in complex scattering environments. Therefore, we propose an underwater polarization imaging method based on frequency domain processing combined with global estimation to solve the above problems. Summary of the invention
[0004] 1. Technical issues to be resolved
[0005] In view of the deficiencies in the prior art, the present invention provides an underwater polarization imaging method based on frequency domain processing combined with global estimation, which solves the problems raised in the above background technology.
[0006] (II) Technical solution
[0007] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:
[0008] An underwater polarization imaging method based on frequency domain processing combined with global estimation includes the following steps:
[0009] Step 1: Acquire multiple underwater images with different polarization directions, and establish an underwater polarization imaging model based on them, and further obtain the polarization degree of target reflected light and the polarization degree of backscattered light;
[0010] Step 2: First, the orthogonal polarization images are preprocessed by histogram stretching and high-pass filtering to reduce the influence of low-frequency scattered light, and the orthogonal polarization image pairs are more accurately obtained through polarization degree correlation;
[0011] Step 3: Process the intensity image through a frequency domain low-pass filter and combine it with the optimal threshold factor to accurately estimate the polarization degree of the backscattered light. At the same time, use the enhancement metric to evaluate the descattering effect to optimize the estimation of the polarization degree of the target light. Finally, restore the image in combination with the physical model.
[0012] Furthermore, in the step 1, the polarization degree of the target reflected light and the polarization degree of the backscattered light range from 0 to 1.
[0013] Furthermore, in the step 2, the image acquisition method for performing histogram stretching and high-pass filtering preprocessing on the orthogonal polarization image is as follows: ⊥ (x, y) is histogram stretched, and its histogram is stretched I' ⊥ (x,y) performs frequency domain preprocessing operations, namely:
[0014]
[0015] Among them I || (x,y) and I ⊥ (x, y) represents an orthogonal polarization image pair;
[0016] After stretching I' ⊥ (x, y) is transformed to the frequency domain through Fourier transformation and Gaussian high-pass filtering is performed to filter out the backscattered light and obtain the high-frequency part where the target signal light is relatively concentrated; after Gaussian high-pass filtering in the frequency domain, the corrected minimum light intensity map is obtained. for:
[0017]
[0018] in, and represent Fourier transform and inverse Fourier transform respectively, H HP is the transfer function of the Gaussian high-pass filter.
[0019] Furthermore, in the step 2, the method for more accurately obtaining the orthogonal polarization image pair by using the polarization degree correlation is: according to Polarization degree preservation and correlation calculation are performed to obtain the maximum light intensity map after preprocessing ,Right now:
[0020]
[0021] This ensures that the degree of polarization remains consistent before and after stretching and filtering.
[0022] Furthermore, in the step 3, the intensity image is processed by a frequency domain low-pass filter, and the method for obtaining the polarization degree of the backscattered light is accurately estimated by combining the optimal threshold factor: first, the intensity image is transformed into the frequency domain by a Fourier function, and then multiplied by the transfer function of the filter, and the result of the multiplication is returned to the spatial domain image by an inverse Fourier transform;
[0023]
[0024] In order to improve the image contrast, in the low contrast area, the threshold factor γ is introduced to optimize the accuracy of the estimation filter. The difference between the overall intensity B of the backscattered light and the original image intensity is defined as ΔI. The accuracy of the frequency domain low-pass filter is improved by setting the threshold factor γ:
[0025] ΔI(x,y)=|B(x,y)-I(x,y)|
[0026]
[0027] in is the optimized backscattered light intensity estimation;
[0028] The DOP (degree of polarization) of the backscattered light can be estimated as:
[0029]
[0030] in is the DOP estimate of the backscattered light, is the polarization degree estimate of the target light.
[0031] Furthermore, in step 3, the enhancement metric is used to evaluate the descattering effect to optimize the estimation of the polarization degree of the target light. Finally, the acquisition method for image restoration combined with the physical model is: for P target We use the enhanced metric (EME) to evaluate the descattering effect; at the same time, we search for P in [0,1] with a step size of 0.01. target The optimal value of and the value of σ with a step size of 0.01 in [0,0.3] are:
[0032]
[0033] Among them, optional means optimization, the image is sorted by ordinal number (k 1 ,k 2 ) is divided into m×n regions, and are the maximum and minimum grayscale values of each region respectively; here we divide the image into 80×80 regions to calculate the EME value, so k 1 and k 2The value of is 1 to 80;
[0034] The final target signal light expression is:
[0035]
[0036] Get clear recovery results.
[0037] (III) Beneficial effects
[0038] Compared with the prior art, the present invention provides an underwater polarization imaging method based on frequency domain processing combined with global estimation, which has the following beneficial effects:
[0039] The present invention first performs histogram stretching and high-pass filtering preprocessing on the orthogonal polarization image to reduce the influence of low-frequency scattered light, and more accurately obtains the orthogonal polarization image pair through polarization degree correlation. Subsequently, the intensity image is processed by a frequency domain low-pass filter, combined with the optimal threshold factor, to accurately estimate the polarization degree of the backscattered light, and the enhanced metric (EME) is used to evaluate the descattering effect to optimize the estimated polarization degree of the target light. Finally, the image is restored in combination with the physical model. No human-computer interaction is required, and the disadvantage that traditional methods are difficult to restore moving targets is effectively solved. Experimental studies were conducted on static and moving targets in turbid water, and satisfactory image restoration quality was achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 It is the overall flow chart of the present invention;
[0041] Figure 2 This is a model diagram of a polarization imaging system simulating an underwater environment according to the present invention;
[0042] Figure 3 (a) is the total light intensity image; (b) and (c) are images restored by traditional methods; and (d) is the effect diagram of the actual application of the present invention. DETAILED DESCRIPTION
[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.
[0044] Example
[0045] like Figure 1-3 As shown, an underwater polarization imaging method based on frequency domain processing combined with global estimation is proposed in one embodiment of the present invention, comprising:
[0046] Step 1: Acquire multiple underwater images with different polarization directions, and thereby establish a target reflected light intensity model and a backscattered light intensity model;
[0047] like Figure 2 The figure shows the polarization imaging system in the simulated underwater environment of the present invention. A collimated LED light source with a central wavelength of 625nm is selected to cooperate with a linear polarizer to realize active illumination in the horizontal direction. The initial light beam passes through the first polarizer 2 with the polarization direction set to the horizontal direction, the transparent water tank 5 and the turbid water 6 to irradiate the target object 4 and the suspended particles 3, wherein the turbid water 6 is made by adding 20ml of milk to the clean water, and the target object 4 is a plastic toy. The target object 4 reflects the initial light beam to obtain the target reflected light, and the suspended particles 3 scatter the initial light beam to obtain the backscattered light; the target reflected light and the backscattered light are irradiated to the CCD detector 8 through the second polarizer 7.
[0048] The polarization direction of the second polarizer 7 is rotated to 0° to obtain the polarization image I0; the polarization direction of the second polarizer 7 is rotated to 45° to obtain the polarization image I45; the polarization direction of the second polarizer 7 is rotated to 90° to obtain the polarization image I90; the polarization direction of the second polarizer 7 is rotated to 135° to obtain the polarization image I135. The Stokes vector is calculated using the four polarization images:
[0049]
[0050] Among them, I represents the total light intensity of the scene; Q is the intensity difference between the horizontal and vertical directions; U is the intensity difference between the 45° and 135° directions.
[0051] Calculate the overall degree of polarization:
[0052]
[0053] Where P represents the overall polarization degree.
[0054] Obtain two orthogonal underwater polarization images
[0055]
[0056] Among them, I || is the polarization image containing the maximum visible backscattered light intensity, recorded as the brightest image; I ⊥ It is the polarization image that contains the minimum visible backscattered light intensity orthogonal to its polarization direction and is recorded as the darkest image.
[0057] I=T+B=I || +I ⊥
[0058] Among them, I represents the total light intensity of the scene irradiation received by the detector, which can be decomposed in mutually orthogonal directions; similarly, T and B can also be decomposed; T represents the target reflected light, which appears as a clear image of the target in the underwater image, that is, the desired image; B represents the backscattered light, which is the main reason for the poor quality of the total light intensity I of the scene irradiation received by the detector.
[0059] Brightest Image I || With the darkest image I ⊥ They can all be composed of their respective target reflected light and backscattered light. Specifically, the following relationship can be established:
[0060] I ⊥ =T ⊥ +B ⊥
[0061] I || =T || +B ||
[0062] Among them, T || With B || are the intensity of the brightest target reflected light and the intensity of the brightest backscattered light, T ⊥ With B ⊥ are the intensity of the darkest target reflected light and the intensity of the darkest backscattered light, respectively.
[0063] The polarization degree of the target reflected light is calculated based on the intensity of the brightest target reflected light and the darkest target reflected light, that is:
[0064]
[0065] Among them, P tar Indicates the polarization degree of the target reflected light.
[0066] The polarization degree of backscattered light is calculated based on the intensity of the brightest backscattered light and the intensity of the darkest backscattered light, that is:
[0067]
[0068] Among them, P sca Represents the polarization degree of backscattered light.
[0069] According to the brightest image, the darkest image, the polarization degree of the target reflected light and the polarization degree of the backscattered light, the target reflected light intensity model is established:
[0070]
[0071] According to the brightest image, the darkest image, the polarization degree of the target reflected light and the polarization degree of the backscattered light, a backscattered light intensity model is established:
[0072]
[0073] Step 2: First, the orthogonal polarization images are preprocessed by histogram stretching and high-pass filtering to reduce the influence of low-frequency scattered light, and the orthogonal polarization image pairs are more accurately obtained through polarization degree correlation;
[0074] According to the polarization imaging model in the classic underwater scattering environment, the total light intensity I(x, y) received by the camera during underwater imaging includes two parts: the target object signal light T(x, y) and the backscattered light B(x, y):
[0075] I(x,y)=T(x,y)+B(x,y)
[0076] =L(x,y)·t(x,y)+A ∞ [1-t(x,y)]
[0077] Where L(x,y) is the unattenuated target signal light, that is, the intensity of the direct light, t(x,y) is the transmission coefficient of the transmission medium, and A ∞ is the backscattered light at infinite distance.
[0078] Underwater polarization imaging technology is based on orthogonal polarization images. || (x,y) and I ⊥ (x, y) is obtained, and combined with the above physical model to restore the clear image of the target object. At this time, I(x, y) is the light intensity signal received by the camera when no linear polarizer is placed:
[0079] I(x,y)=I || (x,y)+I ⊥ (x,y)
[0080] =[T || (x,y)+B || (x,y)]+[T ⊥ (x,y)+B ⊥ (x,y)]
[0081]
[0082] Where P(x,y) is the degree of polarization.
[0083] For imaging in turbid media, the grayscale range of the image is compressed into a narrow band. The compression of the grayscale range can be alleviated by performing a histogram stretching method.
[0084]
[0085] It should be noted that there is a polarization relationship between the two orthogonal polarization images, so independently stretching the grayscale range of the two orthogonal polarization images may break this polarization relationship. In order to maintain the polarization relationship, we need to first process one of the orthogonal images using the histogram stretching method, and then process the other orthogonal image using the DOP correlation.
[0086] In water, due to scattering and absorption, the two orthogonal polarization images I || (x,y) and I ⊥ (x, y) are both difficult to accurately reflect the light intensity and polarization information of the original target object. If these two orthogonal polarization images are used for subsequent polarization image restoration, it is difficult to obtain high-quality imaging effects. In fact, there are differences in the spectrum of the polarization image between the reflected light of the target object and the backscattered light of the scattering medium. The backscattered light of the relative scattering medium information is mainly concentrated in the low-frequency components of the spectrum; while the target signal light information is mainly concentrated in the high-frequency components of the spectrum. Therefore, the stretched orthogonal polarization image can be preprocessed by designing a frequency domain high-pass filter to extract the high-frequency components to separate the target signal light from the backscattered light.
[0087] Among them, due to the minimum light intensity diagram I perpendicular to the polarization direction of the active illumination line ⊥ The scattered light in (x, y) is initially suppressed, making the target information more obvious. Therefore, I' is selected after stretching its histogram. ⊥ (x,y) is preprocessed in the frequency domain. ⊥ (x, y) is transformed to the frequency domain through Fourier transformation and Gaussian high-pass filtering is performed to filter out the backscattered light and obtain the high-frequency part where the target signal light is relatively concentrated. The corrected minimum light intensity map is obtained after Gaussian high-pass filtering in the frequency domain. for:
[0088]
[0089] in, and represent Fourier transform and inverse Fourier transform respectively, H HP is the transfer function of the Gaussian high-pass filter.
[0090] In order to maintain the original polarization relationship, the maximum light intensity diagram The same histogram stretching and high-pass filtering cannot be performed directly. Instead, according to Polarization degree preservation and correlation calculation are performed to obtain the maximum light intensity map after preprocessing :
[0091]
[0092] The function P(·) ensures that the degree of polarization remains consistent before and after stretching and filtering.
[0093] After frequency domain filtering preprocessing, the orthogonal polarization image pair and It contains the high-frequency part of the original image spectrum, that is, it filters out a part of the backscattered light to highlight more target object information.
[0094] At the same time, considering the low-frequency characteristics of backscattered light, we perform frequency domain low-pass filtering on the entire image to accurately estimate the entire intensity B, rather than selecting the targetless background area for calculation. The process is to first transform the intensity image to the frequency domain through the Fourier function, then multiply it with the filter transfer function, and then return the result of the multiplication to the spatial domain image through the inverse Fourier transform.
[0095]
[0096] in and represents Fourier transform and inverse Fourier transform. The transfer function of the low-pass filter is:
[0097]
[0098] Where D(u,v) is the distance from the data point to the center of the frequency domain, D 0 is the cutoff frequency, and λ is the filter order. In this paper, we set the cutoff frequency to D 0 =50Hz, and the filter order is set to λ=2.
[0099] In order to further improve the image contrast, especially in low-contrast areas, a threshold factor γ is introduced to optimize the accuracy of the estimated filter. The difference between the overall intensity B of the backscattered light and the original image intensity is defined as ΔI. The accuracy of the frequency domain low-pass filter is improved by setting the threshold factor γ:
[0100] ΔI(x,y)=|B(x,y)-I(x,y)|
[0101]
[0102] in is the optimized backscattered light intensity estimation.
[0103] The backscattered DOP estimate can be expressed as:
[0104]
[0105] For P targetWe use the enhanced measure (EME) to evaluate the descattering effect. The higher the EME value, the better the image quality. As a non-reference evaluation metric, EME can accurately and quantitatively evaluate the image quality in most cases. Therefore, we simultaneously search for P in [0,1] with a step size of 0.01. target The optimal value of and the value of σ with a step size of 0.01 in [0,0.3] are:
[0106]
[0107] Among them, the images are numbered (k 1 ,k 2 ) is divided into m×n regions, and Here we divide the image into 80×80 regions to calculate the EME value, so k 1 and k 2 The value is 1 to 80.
[0108] Will Substituting in, the final target signal light expression can be obtained as:
[0109]
[0110] To verify the effectiveness of the present invention, the restored images are compared with those of two traditional underwater polarization imaging methods that require manual interception of background non-target areas in the image to estimate parameters. Figure 3 (a) is the total light intensity image, and the image is not clear enough to be seen by the naked eye. Figure 3 (b) and (c) are images restored by traditional methods. Although the visibility of traditional methods has been partially improved, it has not achieved satisfactory results. This is because the method of estimating parameters by manually intercepting the background area without target in the image is not stable and reliable, and the evaluation index is single during the algorithm imaging process, so the obtained results may not be optimal. The method of the present invention has significantly stronger visibility, such as Figure 3 (d) This indicates that the present invention estimates relevant parameters more accurately and can be processed automatically without human-computer interaction, which is a more efficient method.
[0111] In order to quantitatively evaluate the image quality, in addition to the contrast and sharpness already cited, the image enhancement evaluation index (EME) is also used to evaluate the quality of the restored image. The larger the value, the higher the image quality. The results are shown in the following table:
[0112] Total light intensity diagram CLAHE Schechner The present invention Contrast 7.7949 42.9413 95.9831 173.4208 Sharpness 15.0025 14.4855 20 26.1771 EME 2.0635 2.4719 3.1335 5.2084
[0113] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. An underwater polarization imaging method based on frequency domain processing combined with global estimation, Features: The following steps are involved: Step 1: Acquire multiple underwater images with different polarization directions, and establish an underwater polarization imaging model based on them, and further obtain the polarization degree of target reflected light and the polarization degree of backscattered light; Step 2: First, the orthogonal polarization images are preprocessed by histogram stretching and high-pass filtering to reduce the influence of low-frequency scattered light, and the orthogonal polarization image pairs are more accurately obtained through polarization degree correlation; Step 3: Process the intensity image through a frequency domain low-pass filter and combine it with the optimal threshold factor to accurately estimate the polarization degree of the backscattered light. At the same time, use the enhancement metric to evaluate the descattering effect to optimize the estimation of the polarization degree of the target light. Finally, restore the image in combination with the physical model.
2. The underwater polarization imaging method based on frequency domain processing combined with global estimation according to claim 1, Features: In the step 1, the polarization degree of the target reflected light and the polarization degree of the backscattered light range from 0 to 1.
3. The underwater polarization imaging method based on frequency domain processing combined with global estimation according to claim 1, Features: In the step 2, the image acquisition method for performing histogram stretching and high-pass filtering preprocessing on the orthogonal polarization image is as follows: Perform histogram stretching and compare the histogram after stretching Perform frequency domain preprocessing operations, namely: Where I(x,y) and represents an orthogonal polarization image pair; After stretching After Fourier transformation to frequency domain and Gaussian high-pass filtering, the backscattered light is filtered out to obtain the high-frequency part where the target signal light is relatively concentrated; after Gaussian high-pass filtering in the frequency domain, the corrected minimum light intensity map is obtained. for: in, and represent Fourier transform and inverse Fourier transform respectively, H HP is the transfer function of the Gaussian high-pass filter.
4. The underwater polarization imaging method based on frequency domain processing combined with global estimation according to claim 1, Features: In the step 2, the method for more accurately obtaining the orthogonal polarization image pair by using the polarization degree correlation is: according to Polarization degree preservation and correlation calculation are performed to obtain the maximum light intensity map after preprocessing Right now: This ensures that the degree of polarization remains consistent before and after stretching and filtering.
5. The underwater polarization imaging method based on frequency domain processing combined with global estimation according to claim 1, Features: In the step 3, the intensity image is processed by a frequency domain low-pass filter, and the method for accurately estimating the polarization degree of the backscattered light in combination with the optimal threshold factor is as follows: first, the intensity image is transformed into the frequency domain by a Fourier function, and then multiplied by the transfer function of the filter, and the result of the multiplication is returned to the spatial domain image by an inverse Fourier transform; In order to improve the image contrast, in the low contrast area, the threshold factor γ is introduced to optimize the accuracy of the estimation filter. The difference between the overall intensity B of the backscattered light and the original image intensity is defined as ΔI. The accuracy of the frequency domain low-pass filter is improved by setting the threshold factor γ: ΔI(x,y)=|B(x,y)-I(x,y)| in is the optimized backscattered light intensity estimation; The DOP (degree of polarization) of the backscattered light can be estimated as: in is the DOP estimate of the backscattered light, is the polarization degree estimate of the target light.
6. The underwater polarization imaging method based on frequency domain processing combined with global estimation according to claim 1, Features: In step 3, the enhancement metric is used to evaluate the descattering effect to optimize the estimation of the polarization degree of the target light. Finally, the image restoration method combined with the physical model is as follows: target We use the enhanced metric (EME) to evaluate the descattering effect; at the same time, we search for P in [0,1] with a step size of 0.
01. target The optimal value of and the value of σ with a step size of 0.01 in [0,0.3] are: Among them, optional means optimization, the image is sorted by ordinal number (k 1 ,k 2 ) is divided into m×n regions, and are the maximum and minimum grayscale values of each region respectively; here we divide the image into 80×80 regions to calculate the EME value, so k 1 and k 2 The value of is 1 to 80; The final target signal light expression is: Get clear recovery results.
Citation Information
Patent Citations
Objective evaluation method and system for quality of infrared polarization image
CN107424146A
Ultrafast single-pixel polarization imaging system
CN115979967A
Underwater polarization imaging method for recovering non-uniform polarization characteristic target
CN116245758A
Underwater polarization visual detection system and imaging method based on Raspberry Pi 3B +
CN117014579A
Underwater polarization imaging method based on target light polarization degree limiting condition
CN118333905A
Cited By
De-scattering imaging method based on polarization common-mode dual suppression and related device
CN119595554A
Image enhancement method and system based on adaptive filtering and polarization weighted fusion
CN119919327A
Image enhancement method and system based on adaptive filtering and polarization weighted fusion
CN119919327B
Polarization reconstruction method suitable for multi-turbidity underwater scene
CN120031766A
Underwater image restoration method based on polarization imaging and low-frequency filtering
CN120355631A
