Underwater polarization difference fusion imaging method based on angle optimization

Abstract

The invention discloses an underwater polarization difference fusion imaging method based on angle optimization. Belongs to the technical field of underwater polarization, and particularly relates to the technical field of underwater polarization image enhancement. The method comprises the following steps: firstly, analyzing a relationship between an optimal weight coefficient and an image enhancement measure (EME) value based on Stokes vector difference, and calculating a background light polarization angle based on the optimal weight coefficient; secondly, obtaining polarization images in two optimal orthogonal directions by combining a polaroid Mueller matrix, and carrying out differential processing; and finally, selecting a proper weight factor, and carrying out weighted fusion on the obtained polarization difference image and the intensity image to obtain a final clear image.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and specifically to an underwater polarization differential fusion imaging method based on angle optimization. Background Art

[0002] In underwater imaging, due to the influence of light scattering, absorption, and background light, it is usually a great challenge to obtain clear underwater images. Traditional underwater image processing methods often rely on the contrast adjustment of a single image and cannot effectively remove the interference of complex background light or scattered light. As an imaging method that can effectively describe the propagation direction and polarization state of light waves, polarization images can provide a new perspective for underwater image enhancement. However, how to optimize the removal of background light in underwater polarization images and improve the clarity of target images remains an important technical problem. Therefore, we propose an underwater polarization differential fusion imaging method based on angle optimization to solve the above problems. Summary of the Invention

[0003] (I) Technical Problems to be Solved

[0004] In view of the deficiencies of the prior art, the present invention provides an underwater polarization differential fusion imaging method based on angle optimization, which solves the problems raised in the above background art.

[0005] (II) Technical Solutions

[0006] The present invention specifically adopts the following technical solutions to achieve the above object:

[0007] An underwater polarization differential fusion imaging method based on angle optimization, comprising the following steps:

[0008] Step 1: Construct an underwater polarization imaging system to obtain images at different polarization angles.

[0009] Step 2: Based on Stokes vector differential analysis, analyze the relationship between the optimal weight coefficient and the image enhancement measure (EME) value, and calculate the background light polarization angle based on the optimal weight coefficient.

[0010] Step 3: Combine the Mueller matrix of the polarizer to obtain polarization images in two best orthogonal directions and perform differential processing.

[0011] Step 4: Select an appropriate weight factor, and perform weighted fusion on the obtained polarization differential image and the intensity image to obtain the final clear image.

[0012] Further, in the above Step 1, an underwater polarization imaging system is constructed to obtain images at different polarization angles. Usually, different polarization angles are used to image the target area to capture the scattered light and transmitted light at different angles.

[0013] Further, in the second step, the relationship between the optimal weight coefficient and the image enhancement measure (EME) value is analyzed based on Stokes vector difference: First, the Stokes vector is composed of four key parameters S0, S1, S2, and S3, and is thus represented as a standard 4×1 column vector:

[0014]

[0015] where I 0° , I 45° , I 90° , I 135° each component represents the distribution of light intensity in each direction, and the two components I L and I R correspond to the light intensities of left-handed circularly polarized light and right-handed circularly polarized light in the light wave, respectively.

[0016] Secondly, the calculation formula for the weight coefficient is:

[0017]

[0018] where is the polarization angle of the background light.

[0019] where I M is the backscattered light, S1 and S2 are components in the Stokes vector, and θ A is the polarization angle of the background light.

[0020] Finally, the maximum value of EME of the output I M is set as the optimal judgment index, and based on this, the optimal value search is performed for the weight coefficient . Calculate EME:

[0021]

[0022] where x and y are the coordinate values of the pixels, and the principle is to divide the image into k1×k2 small regions (l and k are row and column numbers).

[0023] Further, in the second step, the polarization angle of the background light is calculated based on the optimal weight coefficient: The range of the weight coefficient is (0, 1), and the step size is 0.01. The EME value is used to evaluate I M . When the EME value is the largest, the weight coefficient is optimal at this time, that is:

[0024]

[0025] (θ || , θ ⊥ , ψ) = argmax[f EME (I M )] (7)

[0026] Among them, θ || is the polarization angle in the parallel direction, and θ ⊥ is the polarization angle in the vertical direction.

[0027] After obtaining the optimal weight coefficient, the two best mutually orthogonal polarization azimuth angles are inversely calculated using the weight coefficient.

[0028] 5. Further, in step three, the Mueller matrix of the polarizer is combined to obtain polarization images in two best orthogonal directions, and differential processing is performed: The polarization images I / / and I ⊥ are respectively expressed as:

[0029]

[0030] Among them, θ || is the polarization angle in the parallel direction, and θ ⊥ is the polarization angle in the vertical direction.

[0031] Then, the two best orthogonal images are differentiated, and the differential image can be obtained as:

[0032] I pd = S1sin2θ A - S2cos2θ A (10)

[0033] Further, in step four, a suitable weight factor is selected, and the obtained polarization differential image and intensity image are weighted and fused to obtain the final clear image: A weighted fusion method is adopted, where the fusion weight factor is determined according to the different polarization characteristics of the target object and the concentration of the turbid medium. The fused image can be expressed as:

[0034] I f (x,y) = a(x,y)I p (x,y) + b(x,y)I l (x,y) (11)

[0035] Among them, I p(x,y) is the optimized polarization image, I l(x,y) is the obtained intensity image, a (x,y) is the weighting factor of the polarization image, and b (x,y) is the weighting factor of the intensity image.

[0036] (III) Beneficial effects

[0037] Compared with the prior art, the present invention provides an underwater polarization differential fusion imaging method based on angle optimization, having the following beneficial effects:

[0038] (1) By analyzing the relationship between the optimal weight coefficient and the image enhancement measure (EME) value through Stokes vector difference, the polarization angle of the background light can be accurately calculated, thereby effectively optimizing the image quality.

[0039] (2) Using the Mueller matrix of the polarizer to obtain the polarization image in the best orthogonal direction, through differential processing, unnecessary light scattering and reflection noise in underwater imaging are effectively eliminated, and the contrast and clarity of the image are improved.

[0040] (3) By selecting appropriate weight factors to weightedly fuse the polarization difference image and the intensity image, the details and clarity of the underwater image can be further enhanced, and each important feature of the image is optimized, making the final output image more visible, reducing the interference in the underwater environment, improving the recognition effect of the target, and ensuring the stability of the image quality.

[0041] The method described in the present invention can be applied in the field of underwater image enhancement technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is the general flow chart of the present invention;

[0043] Figure 2 is the model diagram of the polarization imaging system in the simulated underwater environment of the present invention;

[0044] Figure 3 In (a) is the original image; (b) is the restored image by the traditional polarization difference method; (c) is the restored image by the fast target enhancement method based on polarization; (d) is the restored image by the method described in the present invention DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0046] Embodiment

[0047] As shown in Figure 1 and 2 and 3, a method for underwater polarization difference fusion imaging based on angle optimization proposed in an embodiment of the present invention includes:

[0048] Step 1: Construct an underwater polarization imaging system to obtain images with different polarization angles;

[0049] As shown in Figure 2The polarization imaging system in the simulated underwater environment of the present invention is shown. A collimated LED light source 1 with a central wavelength of 625 nm is selected and combined with a linear polarizer 2, 7 to enable the active illumination starting beam in the horizontal direction to pass through the first polarizer 2 with a polarization direction set to the horizontal direction, a transparent water tank 5, and turbid water 6 to irradiate a target object 4 and suspended particles 3. Among them, the turbid water 6 is made by adding 20 ml of milk to clear water, and the target object 4 is a coin. The target object 4 reflects the starting beam to obtain target reflected light, and the suspended particles 3 scatter the starting beam to obtain backscattered light; the target reflected light and the backscattered light are irradiated onto a CCD detector 8 through a second polarizer 7.

[0050] Rotate the polarization direction of the second polarizer 7 to 0°, and obtain a polarization image I 0° ; Rotate the polarization direction of the second polarizer 7 to 45°, and obtain a polarization image I 45° ; Rotate the polarization direction of the second polarizer 7 to 90°, and obtain a polarization image I 90° ; Rotate the polarization direction of the second polarizer 7 to 135°, and obtain a polarization image I 135° . Calculate the Stokes vector using four polarization images:

[0051]

[0052] Among them, S0 is the total intensity of the light field; S1 is the intensity difference between the horizontal and vertical directions; S2 is the intensity difference between the 45° and 135° directions.

[0053] Based on the Stokes vector, the degree of polarization P can be calculated, that is, the proportion of the energy of polarized light in the total energy of the beam, which can be expressed as:

[0054]

[0055] Describing the vibration direction of the beam vector as the polarization angle, it can be expressed as:

[0056]

[0057] Step 2: Based on the Stokes vector difference analysis, analyze the relationship between the optimal weight coefficient and the image enhancement measure (EME) value, and calculate the background light polarization angle based on the optimal weight coefficient;

[0058] In the Stokes vector, since S0 contains both the target signal light and the scattered light, while S1 and S2 are hardly affected by the target signal light, it can be known that the polarization angle θ can provide more accurate information about the scattered light. Compared with the traditional polarization difference method, the method proposed in this paper does not require rotating the angle of the polarizer, but combines the image enhancement measure (EME) value. When the EME value is the largest, the polarization azimuth angle θ at this time can be obtained A ,

[0059] To eliminate the scattered light in the θ A polarization direction, the polarization angle in the parallel direction is obtained according to the polarization difference principle as:

[0060]

[0061] The polarization angle in the vertical direction is:

[0062]

[0063] Since the polarization angle definition formula can be transformed as:

[0064]

[0065] where I M is the backscattered light, S1 and S2 are components in the Stokes vector, θ A is the background light polarization angle, is the weight coefficient.

[0066] To determine the exact value of the most weighted coefficient, it is necessary to quantitatively describe the output interface of I M This invention selects the image enhancement measure EME value as the quantization standard. EME is an important index to describe the change of image sharpness, and its mathematical expression is:

[0067]

[0068] where x and y are the coordinate values of pixels. The principle is to divide the image into k1×k2 small regions (l and k are row and column numbers). Calculate the logarithmic mean of the maximum and minimum gray levels in the small region, which represents the change degree of the local domain gray level of the image. The stronger the local gray level change, the stronger the details shown in the image.

[0069] First, bring the Stokes vector images S1 and S2 into the model, and then set the maximum EME of the output result I M as the optimal judgment index, and search for the optimal value of the weight coefficient In the case of linearly polarized light incidence, the range of the weight coefficient is (0,1), so set the step size to 0.01 to search for the weight coefficient when the highest EME value is obtained.

[0070] After completing the traversal process, the optimal weight coefficient can be determined. Subsequently, according to the calculation formula of, take the inverse of the optimal weight coefficient to obtain the background light polarization angle θ A .

[0071] Step 3: Combine the Mueller matrix of the polarizer to obtain the polarization images in two best orthogonal directions and perform differential processing.

[0072] The Mueller matrix is a common method used to describe the polarization information of an object. When the angle between the central axis of the polarizer and the horizontal direction is α, the Mueller matrix corresponding to the polarizer can be expressed as:

[0073]

[0074] Using the Stokes vector of the light beam and the Mueller matrix of the polarizer, the Stokes vector of the light beam after passing through the polarizer can be obtained:

[0075] S out = M·S in (11)

[0076] where S out is the output Stokes component, and S in is the input Stokes component.

[0077] Under the condition of knowing the Stokes vector of the light and the Mueller matrix of the polarizer, the Stokes vector of the light at any projection axis angle can be deduced, and thus the intensity of the light after passing through the polarizer can be calculated:

[0078]

[0079] Furthermore, the polarization images I / / and I ⊥ of two optimal mutually orthogonal polarization directions can be obtained and are respectively expressed as:

[0080]

[0081] Then, according to the traditional polarization difference principle, the two optimal orthogonal images are differentiated to obtain the differential image as:

[0082] I pd = I || - I ⊥ (15)

[0083] Further simplification gives the optimized differential image as:

[0084] I pd = S1sin2θ A - S2cos2θ A (16)

[0085] Step 4: Select appropriate weight factors and perform weighted fusion on the obtained polarization difference image and intensity image to obtain the final clear image.

[0086] When an object is immersed in water, the scattering of light by water molecules causes a decrease in the contrast and clarity of the underwater image. Underwater polarization difference imaging technology can effectively distinguish the target reflected light from the water body scattered light by utilizing the differences in the polarization characteristics of the reflected light and the scattered light. However, the differences in the surface roughness of different underwater targets lead to different degrees of depolarization of the reflected light, thus affecting the effect of polarization difference imaging. Traditional polarization difference imaging not only removes unpolarized light but also eliminates the reflected light with severe depolarization, resulting in lower brightness of these targets in the image. The intensity image can make up for this deficiency, and the two complement each other.

[0087] By combining the image obtained through polarization difference imaging and the intensity image, a clearer final image can be obtained under turbid water conditions. In the present invention, a weighted fusion method is adopted, where the weight factor for fusion is determined according to the different polarization characteristics of the target object and the concentration of the turbid medium. The fused image can be expressed as:

[0088] I f (x,y) = a(x,y)I p (x,y) + b(x,y)I l (x,y) (17)

[0089] where, I p(x,y) is the optimized polarization image, I l(x,y) is the obtained intensity image, a (x,y) is the weighting factor for the polarization image, and b (x,y) is the weighting factor for the intensity image.

[0090] The values of the weight coefficients a and b directly affect the pixel gray values of the fused image. The intensity of the original underwater image can be divided into three parts based on the degree of polarization: the direct reflected light from high-polarization smooth targets, the direct reflected light from low-polarization rough targets, and the medium-polarization scattered light from the water body.

[0091] For the weight factor a, I p Since it contains the polarization intensity part in the original image, when a certain pixel in the original image has a high degree of polarization, it can be considered that it contains a higher reflected light from the target, so its value is correspondingly larger.

[0092] For the weight factor b, since the intensity image itself contains high gray values and the overall contrast of the image is low, in order to ensure the overall brightness level of the fused image without affecting the contrast, b should be selected as a larger value.

[0093] To meet the above conditions, the values of the weight factors can be expressed as:

[0094]

[0095] where, Ip(x,y) is the optimized polarization image, I l(x,y) is the acquired intensity image, a (x,y) is the weighting factor of the polarization image, b (x,y) is the weighting factor of the intensity image.

[0096] To verify the effectiveness of the present invention, a comparison is made with the traditional polarization difference method and the fast target enhancement method based on polarization. Among them Figure 3 (a) is the original image, which is not very clear to the naked eye. Figure 3 (b) and (c) Although the visibility has been partially improved, it has not reached a satisfactory level. And Figure 3 (d) is the method using the present invention, and the visibility has been significantly improved, and the image quality is also significantly better, approaching a clear image, and the details of the object are easier to distinguish. In particular, the flower pattern on the coin is clearer. This further shows that the present invention estimates the relevant parameters more accurately and has a better effect of removing scattered light, and is a more efficient method.

[0097] To quantitatively evaluate the image quality, the contrast, information entropy, and standard deviation of each image are calculated, and the results are as follows:

[0098] Original image Traditional polarization difference Fast target enhancement based on polarization The present invention Information entropy 5.1804 5.3545 5.7694 6.5221 Standard deviation 9.2059 10.2592 13.3785 23.3665 Contrast 56.1358 94.2564 134.2571 170.1563

[0099] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. 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 differential fusion imaging method based on angle optimization, characterized in that: It includes the following steps: Step 1: Construct an underwater polarization imaging system to obtain images at different polarization angles. Step 2: Based on Stokes vector difference, analyze the relationship between the optimal weight coefficient and the Image Enhancement Measure (EME) value, and calculate the background light polarization angle based on the optimal weight coefficient. Step 3: Combine the Mueller matrix of the polarizer to obtain polarization images in two optimal orthogonal directions and perform differential processing. Step 4: Select an appropriate weight factor, and weighted fuse the obtained polarization difference image and intensity image to obtain the final clear image.

2. The underwater polarization differential fusion imaging method based on angle optimization according to claim 1, wherein: In Step 1, an underwater polarization imaging system is constructed to obtain images at different polarization angles. Usually, different polarization angles are used to image the target area to capture the scattered light and transmitted light at different angles.

3. A method for underwater polarization differential fusion imaging based on angle optimization according to claim 1, characterized in that: In Step 2, based on Stokes vector difference, analyze the relationship between the optimal weight coefficient and the Image Enhancement Measure (EME) value: First, the Stokes vector consists of four key parameters S0, S1, S2, and S3, and is thus represented as a standard 4×1 column vector: Among them, I 0° , I 45° , I 90° , I 135° Each component represents the distribution of light intensity in each direction. I L and I R The two components respectively correspond to the light intensities of left-handed circularly polarized light and right-handed circularly polarized light in the light wave. Secondly, the weight coefficient The calculation formula is as follows: where I M is the backscattered light, S1 and S2 are components in the Stokes vector, and θ A is the polarization angle of the background light. Finally, set the maximum value of the EME of the output I M as the optimal judgment index, and search for the optimal value of the weight coefficient accordingly. Calculate the EME: where x and y are the coordinate values of the pixels. The principle is to divide the image into k1×k2 small regions (l and k are row and column numbers).

4. A method for underwater polarization differential fusion imaging based on angle optimization according to claim 1, characterized in that: In the second step, calculate the background light polarization angle based on the optimal weight coefficient: the range of the weight coefficient is (0, 1), the step size is 0.01, and the EME value is used to evaluate I M , when the EME value is the largest, the weight coefficient is optimal at this time, that is: (θ || , θ ⊥ , ψ) = argmax[f EME (I M )] (6) where θ || is the polarization angle in the parallel direction, and θ ⊥ is the polarization angle in the perpendicular direction. After obtaining the optimal weight coefficient, use the weight coefficient to invert the two best mutually orthogonal polarization azimuth angles.

5. A method for underwater polarization differential fusion imaging based on angle optimization according to claim 1, characterized in that: In step 3, the polarization images in two optimal orthogonal directions are obtained by combining the Mueller matrix of the polarizer, and differential processing is performed: the polarization images I / / and I ⊥ are respectively expressed as: where θ || is the polarization angle in the parallel direction, and θ ⊥ is the polarization angle in the perpendicular direction. Then, perform a difference on the two best orthogonal images to obtain the difference image as: I pd = S1sin2θ A - S2cos2θ A (9) 6. A method for underwater polarization differential fusion imaging based on angle optimization according to claim 1, characterized in that: In Step 4, select an appropriate weight factor, and weighted fuse the obtained polarization difference image and intensity image to obtain the final clear image: A weighted fusion method is adopted, where the fusion weight factor is determined according to the different polarization characteristics of the target object and the concentration of the turbid medium. The fused image can be expressed as: I f (x,y) = a(x,y)I p (x,y) + b(x,y)I l (x,y) (10) Among them, I p(x,y) is the optimized polarization image, and I l(x,y) is the acquired intensity image, a (x,y) is the weighting factor of the polarization image, and b (x,y) is the weighting factor of the intensity image.

Images