Underwater polarization imaging method based on Stokes vector

Abstract

The invention discloses an underwater polarization imaging method based on a Stokes vector, and the method comprises the following steps: 1, obtaining a plurality of polarization images with different polarization directions, so as to obtain the Stokes vector of each polarization image; step 2, based on the Stokes vector of each polarization image, obtaining a Stokes vector model of back scattering light through frequency domain filtering; 3, according to the Stokes vector model of the backscattered light, obtaining a backscattered light image of any polarization angle, solving a maximum backscattered light image in combination with the low-rank characteristic of the backscattered light image, correlating a minimum backscattered light image by using the degree of polarization, and finally fusing the minimum backscattered light image into an underwater imaging physical model to establish a restored image inversion model; and 4, according to the restored image inversion model, obtaining a final imaging result.

Description

Technical Field

[0001] The invention belongs to the technical field of underwater polarization imaging, and in particular relates to an underwater polarization imaging method based on Stokes vector. Background Art

[0002] Underwater optical imaging technology is widely used in fields such as marine resource exploration and marine rescue. However, due to the scattering and absorption effects of suspended particles in the water on light, underwater images usually show blurred details and poor contrast, which seriously affects visual perception. The fundamental reason for the degradation of underwater image quality is the interference of backscattered light. Based on the phenomenon that backscattered light has partial polarization characteristics, underwater polarization imaging technology has been proven to be an effective means to improve the quality of degraded images.

[0003] However, most existing underwater polarization imaging methods assume that backscattered light is uniformly distributed and cannot accurately estimate the backscattered light, resulting in a limited degree of descattering in the final imaging. At the same time, most existing underwater polarization imaging methods are limited by the background area and the human-computer interaction process. The above problems greatly limit the on-site application and application scope of underwater polarization imaging technology. Summary of the invention

[0004] Purpose of the invention: In order to solve the problems of limited descattering degree and small application scope of existing underwater polarization imaging methods, the present invention proposes an underwater polarization imaging method based on Stokes vector. Through Stokes vector analysis and the correlation between polarization angle and polarization degree, the estimation of backscattered light and restoration of the target are realized. The method of the present invention not only overcomes the difficulty that traditional methods are limited by background areas, but also can automatically realize underwater image restoration of targets with different polarization characteristics in the absence of background areas.

[0005] Technical solution: An underwater polarization imaging method based on Stokes vector, comprising the following steps:

[0006] Step 1: Acquire multiple polarization images with different polarization directions to obtain the Stokes vector of each polarization image;

[0007] Step 2: Based on the Stokes vector of each polarization image, obtain the Stokes vector model of the backscattered light through frequency domain filtering;

[0008] Step 3: Based on the Stokes vector model of backscattered light, the backscattered light image of any polarization angle is obtained. The maximum backscattered light image is obtained by combining the low-rank characteristics of the backscattered light, and the minimum backscattered light image is associated with the polarization degree. Finally, it is integrated into the classic underwater imaging physical model to establish the restored image inversion model;

[0009] Step 4: Obtain the final imaging result based on the restored image inversion model.

[0010] Furthermore, in step 1, the step of acquiring a plurality of polarization images with different polarization directions to obtain the Stokes vector of each polarization image includes:

[0011] Acquire a polarization image I(0) with a polarization direction of 0°, a polarization image I(45) with a polarization direction of 45°, a polarization image I(90) with a polarization direction of 90°, and a polarization image I(135) with a polarization direction of 135°;

[0012] According to the following formula, the Stokes vector of each polarization image is obtained:

[0013] S0=I(0)+I(90)

[0014] S1=I(0)-I(90)

[0015] S2=I(45)-I(135)

[0016] Among them, S0 is the total light intensity; S1 is the light intensity difference between 0° and 90°, and S2 is the light intensity difference between 45° and 135°.

[0017] Furthermore, in step 2, the Stokes vector based on each polarization image is filtered in the frequency domain to obtain a Stokes vector model of the backscattered light, which is expressed as:

[0018]

[0019] Among them, S0 B is the total intensity of backscattered light, S1 B is the light intensity difference between 0° and 90°, S2 B is the light intensity difference at 45° and 135°, F and F -1 are Fourier transform and inverse Fourier transform respectively, H(u,v) represents the distance between two points in the spectrum, D 0 is the cut-off frequency.

[0020] Further, in step 3, the backscattered light image of any polarization angle is obtained according to the Stokes vector model of the backscattered light, which is expressed as:

[0021]

[0022] B=B max +B min

[0023] Where θ is the polarization angle of the backscattered light, B is the backscattered light image, and B max is the maximum backscattered light image, B min This is the minimum backscattered light image.

[0024] Furthermore, the low-rank characteristic of the backscattered light is combined to obtain the maximum backscattered light image, which is expressed as:

[0025] B max =argmin{rank(B(θ))}

[0026]

[0027]

[0028] Among them, rank(·) is the rank operator, AOP B is the polarization angle of the backscattered light.

[0029] Furthermore, the minimum backscattered light image is associated with the polarization degree, which is expressed as:

[0030]

[0031] Where Pscat is the polarization degree of the backscattered light, expressed as:

[0032]

[0033] Furthermore, in step 3, the integration into the classical underwater imaging physical model to establish a restored image inversion model specifically includes:

[0034] The classical underwater imaging physical model is expressed as:

[0035] I=D+B=L·t+A ∞ (1-t)

[0036] Among them, I is the total light intensity of the underwater image, D is the target reflected light, B is the backscattered light, L is the restored image, t is the medium transmittance, A ∞ is the backscattered light intensity at infinity;

[0037] The obtained backscattered light image is integrated into the classical underwater imaging physical model to obtain the restored image inversion model, which is expressed as:

[0038]

[0039] Furthermore, in step 4, the final imaging result is obtained according to the restored image inversion model, and the specific operations include:

[0040] Taking maximizing image contrast as the objective function, the optimal solution of cutoff frequency that satisfies the maximum image contrast is obtained;

[0041] The total light intensity S0, the backscattered light B, and the backscattered light intensity A at infinity∞ The optimal solution of the cutoff frequency is substituted into the restored image inversion model to obtain the final imaging result.

[0042] Further, the image contrast is expressed as:

[0043]

[0044] Where N is the total number of pixels of the restored image L. is the average grayscale value of the restored image L, and L(x,y) is the grayscale value corresponding to the pixel point (x,y);

[0045] The optimal solution of the cut-off frequency is expressed as:

[0046] (D 0 ) optimal =argmax{Contrast(L)}

[0047] Among them, Contrast(L) is the contrast of the restored image.

[0048] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0049] (1) The method of the present invention not only overcomes the limitation of the traditional method on the background area, but also can handle the polarization changes of complex target surfaces, greatly improving the descattering effect and the feasibility of field application, and promoting the application and development of underwater polarization imaging technology;

[0050] (2) The method of the present invention can directly calculate the uniform distribution of backscattered light, avoiding the unreasonable assumption about the uniformity of backscattered light intensity in traditional underwater polarization imaging methods, and realize automatic restoration of underwater images of targets with different polarization characteristics without the need for human-computer interaction and without being restricted by the image background area, thus promoting the application and development of underwater polarization imaging technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flow chart of an underwater polarization imaging method based on Stokes vector proposed by the present invention;

[0052] Figure 2 A model diagram of a polarization imaging system in a simulated underwater environment proposed by the present invention;

[0053] Figure 3 The effect diagram of applying the method of the present invention is shown in FIG. Figure 3 (a) is the total light intensity image underwater; Figure 3 (b) is a diagram showing the effect of the actual application of the present invention. DETAILED DESCRIPTION

[0054] The technical solution of the present invention is now further described in conjunction with the accompanying drawings and embodiments.

[0055] like Figure 1 As shown, this embodiment discloses an underwater polarization imaging method based on Stokes vector, which mainly includes the following steps:

[0056] Step 1: Acquire multiple underwater images with different polarization directions, and establish the Stokes vector model of backscattered light through frequency domain filtering.

[0057] Figure 2 The polarization imaging system in a simulated underwater environment is shown. The LED white light source 3 emits a starting light beam, which passes through a first polarizer 4 with a horizontal polarization direction, a transparent water tank 5 and turbid water 8, and irradiates a target 7 and suspended particles 6. In this embodiment, the turbid water 8 is made by adding 15 ml of milk to clean water. The target 7 is composed of plastic material and metal material. In this embodiment, a plastic Rubik's Cube with metal coins attached is used. The metal coins with high polarization characteristics and the plastic Rubik's Cube with low polarization characteristics together form a non-uniform polarization characteristic target. The target 7 reflects the starting light beam to obtain target reflected light, and the suspended particles 6 scatter the starting light beam to obtain backscattered light. The target reflected light and the backscattered light are irradiated onto the CCD detector 1 through the second polarizer 2.

[0058] The total light intensity of the underwater image received by the CCD detector 1 has the following relationship with the target reflected light and backscattered light, which is the classic underwater imaging physical model:

[0059] I=D+B=L·t+A ∞ (1-t)

[0060] Among them, I is the total light intensity of the underwater image, D is the target reflected light, B is the backscattered light, L is the restored image, t is the medium transmittance, A ∞ is the backscattered light intensity at infinity.

[0061] The polarization direction of the second polarizer 2 is rotated to 0° to obtain the corresponding polarization image I(0); the polarization direction of the second polarizer 2 is rotated to 45° to obtain the corresponding polarization image I(45); the polarization direction of the second polarizer 2 is rotated to 90° to obtain the corresponding polarization image I(90); the polarization direction of the second polarizer 2 is rotated to 135° to obtain the corresponding polarization image I(135).

[0062] Substitute the collected polarization image into the following formula to obtain the Stokes vector of the polarization image:

[0063] S0=I(0)+I(90)

[0064] S1=I(0)-I(90)

[0065] S2=I(45)-I(135)

[0066] Among them, S0 is the total light intensity of the underwater image; S1 is the light intensity difference between 0° and 90°; S2 is the light intensity difference between 45° and 135°;

[0067] Through frequency domain filtering, the Stokes vector model of backscattered light is established, namely:

[0068]

[0069] Among them, S0 B is the total intensity of backscattered light, S1 B is the light intensity difference between 0° and 90°; S2 B is the light intensity difference at 45° and 135°, F and F -1 are Fourier transform and inverse Fourier transform respectively, H(u,v) represents the distance between two points in the spectrum, D 0 is the cut-off frequency.

[0070] Step 2: Calculate the polarization angle image through the Stokes vector model, combine the low-rank characteristics to find the maximum backscattered light image, and use the polarization degree to correlate the minimum backscattered light image, and finally integrate it into the classic underwater imaging physical model to establish the restored image inversion model. The specific operations include:

[0071] According to the Stokes vectors of the three backscattered lights, the backscattered light image at any polarization angle is obtained, namely:

[0072]

[0073] B=B max +B min

[0074] Where θ is the polarization angle of the backscattered light, B is the backscattered light image, and B max , B min They are the maximum and minimum backscattered light images respectively.

[0075] According to the low-rank characteristics of the backscattered light, the maximum backscattered light image B is obtained max ,Right now:

[0076] B max =argmin{rank(B(θ))}

[0077]

[0078] Among them, rank(·) is the rank operator, AOP B is the polarization angle of the backscattered light.

[0079] According to the polarization degree of the backscattered light, the minimum backscattered light image B is associated min ,Right now:

[0080]

[0081] Wherein, Pscat is the polarization degree of the backscattered light.

[0082] According to the definition of polarization angle and polarization degree, it can be expressed by the following formula:

[0083]

[0084]

[0085] The above parameters are combined with the classic underwater imaging physical model to establish the restored image inversion model:

[0086]

[0087] Among them, L is the restored image, A ∞ is the backscattered light intensity at infinity, which can be set to the average value of the first 0.1% pixels with the largest grayscale value in the total light intensity image. ∞ The result is 0.7987.

[0088] Step 3: Perform inversion based on the image contrast and the restored image inversion model to obtain the final imaging result. The specific operations include:

[0089] Independent variable cutoff frequency D 0 The selection of is related to the quality of the restored image. The method for obtaining it is: taking the image contrast as the objective function, obtain the optimal solution that satisfies the maximum image contrast, that is:

[0090] (D 0 ) optimal =argmax{Contrast(L)}

[0091] Among them, Contrast(L) is the contrast of the restored image.

[0092] Specifically, the image contrast is obtained by the following formula:

[0093]

[0094] Where N is the total number of pixels of the restored image L, is the average grayscale value of the image, and L(x,y) is the grayscale value corresponding to the pixel point (x,y).

[0095] Specifically, this embodiment uses a genetic algorithm to find the optimal solution, sets the population size to 50, the maximum number of genetic iterations to 100, the crossover probability to 0.5, the mutation probability to 0.01, and finally gives the solution value (D 0 ) optimal =2.

[0096] The above calculations can be used to obtain the total light intensity S0 of the underwater image, the backscattered light B, and the backscattered light intensity A at infinity. ∞ And the optimal solution (D 0 ) optimal Substitute the restored image into the inversion model to obtain the final imaging result.

[0097] To verify the effectiveness of the method in this embodiment, the target image without background area was obtained by using image segmentation technology to obtain the total light intensity image in turbid water. Figure 3 In (a), the image is not very clear and is covered by a layer of "fog" as can be seen by the naked eye; however, after being processed by the method of this embodiment, the image clarity and visibility are significantly improved, as shown in FIG. Figure 3 As shown in (b) in the figure, the method of this embodiment is not limited by the background area and human-computer interaction, and can achieve underwater image restoration of targets with non-uniform polarization characteristics.

[0098] In order to quantitatively evaluate the image quality, in addition to the image contrast already cited, standard deviation, information entropy, and average gradient are also used to evaluate the quality of the restored image. The larger the value, the higher the image quality. The results are shown in Table 1.

[0099] Table 1 uses image contrast, standard deviation, information entropy, and average gradient to evaluate the quality of the restored image

[0100]

[0101]

[0102] It can be seen from Table 1 that the restored image obtained by the method of this embodiment has significantly improved various objective evaluation indicators compared with the original total light intensity image, which verifies that the method of this embodiment has a superior descattering effect.

Claims

1. An underwater polarization imaging method based on Stokes vector, characterized in that: The following steps are involved: Step 1: Acquire multiple polarization images with different polarization directions to obtain the Stokes vector of each polarization image; Step 2: Based on the Stokes vector of each polarization image, obtain the Stokes vector model of the backscattered light through frequency domain filtering; Step 3: Based on the Stokes vector model of backscattered light, the backscattered light image of any polarization angle is obtained. The maximum backscattered light image is obtained by combining the low-rank characteristics of the backscattered light, and the minimum backscattered light image is associated with the polarization degree. Finally, it is integrated into the classic underwater imaging physical model to establish the restored image inversion model; Step 4: Obtain the final imaging result based on the restored image inversion model.

2. The underwater polarization imaging method based on Stokes vector according to claim 1, characterized in that: In step 1, the acquisition of multiple polarization images with different polarization directions to obtain the Stokes vector of each polarization image includes: Acquire a polarization image I(0) with a polarization direction of 0°, a polarization image I(45) with a polarization direction of 45°, a polarization image I(90) with a polarization direction of 90°, and a polarization image I(135) with a polarization direction of 135°; According to the following formula, the Stokes vector of each polarization image is obtained: S0=I(0)+I(90) S1=I(0)-I(90) S2=I(45)-I(135) Among them, S0 is the total light intensity; S1 is the light intensity difference between 0° and 90°, and S2 is the light intensity difference between 45° and 135°.

3. The underwater polarization imaging method based on Stokes vector according to claim 2, characterized in that: In step 2, the Stokes vector based on each polarization image is filtered in the frequency domain to obtain the Stokes vector model of the backscattered light, which is expressed as: Among them, S0 B is the total intensity of backscattered light, S1 B is the light intensity difference between 0° and 90°, S2 B is the light intensity difference at 45° and 135°, F and F -1 They are Fourier transform and inverse Fourier transform respectively, H(u,v) represents the distance between two points in the spectrum, and D0 is the cutoff frequency.

4. The underwater polarization imaging method based on Stokes vector according to claim 3, characterized in that: In step 3, the backscattered light image of any polarization angle is obtained according to the Stokes vector model of the backscattered light, which is expressed as: Where θ is the polarization angle of the backscattered light, B is the backscattered light image, and B max is the maximum backscattered light image, B min This is the minimum backscattered light image.

5. The underwater polarization imaging method based on Stokes vector according to claim 4, characterized in that: The maximum backscattered light image is obtained by combining the low-rank characteristics of the backscattered light, which is expressed as: B max =argmin{rank(B(θ))} Among them, rank(·) is the rank operator, AOP B is the polarization angle of the backscattered light.

6. The underwater polarization imaging method based on Stokes vector according to claim 5, characterized in that: The minimum backscattered light image is associated with the polarization degree, which is expressed as: Where Pscat is the polarization degree of the backscattered light, expressed as:

7. The underwater polarization imaging method based on Stokes vector according to claim 6, characterized in that: In step 3, the integration into the classical underwater imaging physical model to establish a restored image inversion model specifically includes: The classical underwater imaging physical model is expressed as: I=D+B=L·t+A ∞ (1-t) Among them, I is the total light intensity of the underwater image, D is the target reflected light, B is the backscattered light, L is the restored image, t is the medium transmittance, A ∞ is the backscattered light intensity at infinity; The obtained backscattered light image is integrated into the classical underwater imaging physical model to obtain the restored image inversion model, which is expressed as:

8. The underwater polarization imaging method based on Stokes vector according to claim 7, characterized in that: In step 4, the final imaging result is obtained according to the restored image inversion model, and the specific operations include: Taking maximizing image contrast as the objective function, the optimal solution of cutoff frequency that satisfies the maximum image contrast is obtained; The total light intensity S0, the backscattered light B, and the backscattered light intensity A at infinity ∞ The optimal solution of the cutoff frequency is substituted into the restored image inversion model to obtain the final imaging result.

9. The underwater polarization imaging method based on Stokes vector according to claim 8, characterized in that: The image contrast is expressed as: Where N is the total number of pixels of the restored image L. is the average grayscale value of the restored image L, and L(x,y) is the grayscale value corresponding to the pixel point (x,y); The optimal solution of the cut-off frequency is expressed as: (D0) optimal =argmax{Contrast(L)} Among them, Contrast(L) is the contrast of the restored image.

Images