CN114937067A - Image registration method of sub-aperture polarization camera - Google Patents

Abstract

An image registration method of a split-aperture polarization camera comprises the steps of obtaining a polarization subimage, roughly registering the polarization subimage, extracting characteristic points of the polarization subimage, determining the number of points in the polarization subimage, obtaining the optimal registration coefficient of the polarization subimage and finishing the registration of the polarization subimage. According to the invention, the polarization sub-images are roughly registered, so that the calculated amount when the characteristic points of the polarization sub-images are extracted is reduced, the optimal registration coefficient when the number of the points in the polarization sub-images is the largest is determined by traversing the characteristic points, the high-precision, high-speed and full-automatic registration between the polarization sub-images is completed, the technical problem of position deviation between the sub-aperture polarization sub-images is solved, the image quality when the sub-aperture polarization camera is used for polarization imaging is improved, and the method can be popularized and used in the technical field of polarization imaging of the sub-aperture polarization camera.

Description

Image Registration Method for Split Aperture Polarization Cameras

technical field

The invention belongs to the technical field of image registration, and particularly relates to an image registration method of a split-aperture polarization camera.

Background technique

Polarization imaging technology can highlight the details of scene targets from complex backgrounds, and can significantly improve image quality under severe weather conditions. In order to meet different needs, various types of polarization imaging systems have been developed. Compared with the time division polarization imaging system, the amplitude division polarization imaging system and the focal plane polarization imaging system, the aperture division polarization imaging system has the advantages of compact structure, easy implementation and low cost, and can simultaneously obtain the full polarization information of the scene. However, even if the aperture polarization camera can achieve a better use state after artificial adjustment and assembly, there is still an inevitable positional deviation between the acquired polarized sub-images.

In the field of polarization imaging technology, the technical problem that needs to be solved in the wide application of the split aperture polarization camera is to provide an image registration method of the split aperture polarization camera with high precision, high speed and full automation.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is to overcome the above-mentioned shortcomings of the prior art, and to provide an image registration method for a split-aperture polarization camera with high precision, high speed and full automaticity.

The technical solution adopted to solve the above technical problems is composed of the following steps:

(1) Acquisition of polarized sub-images

Shoot the target with a split-aperture polarization camera, obtain the original frame image, cut the original frame image into four polarization sub-images, select any one of the four polarization sub-images as the reference polarization sub-image, and the remaining three polarization sub-images as the polarized sub-image to be registered.

(2) Coarse registration of polarized sub-images

Determine the Fourier transform F(ω _x ,ω _y ) of the reference polarizer image according to equation (1):

F(ω _x ,ω _y )=F{f(x,y)} (1)

Among them, f(x, y) is the reference polarized sub-image of the coordinate (x, y), and F{} represents the Fourier transform.

Determine the Fourier transform G(ω _x ,ω _y ) of the polarized sub-image to be registered according to formula (2):

G(ω _x ,ω _y (=F{g(x,y)} (2)

Among them, g(x, y) is the polarized sub-image to be registered with coordinates (x, y).

Determine the impulse function δ(x-dx, y-dy) according to formula (3):

Among them, dx is the displacement of the polarized sub-image to be registered in the horizontal direction, dy is the displacement of the polarized sub-image to be registered in the vertical direction, F ^-1 {} is the inverse Fourier transform, F ^* (ω _x , ω _y ) is the complex conjugate of F(ω _x ,ω _y ).

Determine the coordinates (x', y') of the roughly registered polarized sub-image according to formula (4):

Among them, (x ₀ , y ₀ ) are the coordinates of the polarized sub-image to be registered.

Determine the rough registration polarized sub-image I _rough (x ₀ , y ₀ ) according to formula (5):

I _rough (x ₀ ,y ₀ )=I(x′,y′) (5)

Among them, I(x', y') is the intensity of the polarized sub-image to be registered at the coordinates (x', y').

(3) Extract the feature points of polarized sub-images

Extract the feature points of the reference polarization sub-image and the coarse registration polarization sub-image respectively. For any feature point in the reference polarization sub-image, select the two feature points with the smallest and second smallest Euclidean distances from the coarse registration polarization sub-image, According to the ratio of the smallest Euclidean distance to the second smallest Euclidean distance is not greater than the set threshold condition, the selected feature point in the reference polarized sub-image and the feature point with the smallest Euclidean distance in the coarsely registered polarized sub-image form a feature point pair, other features The method for judging that points form feature point pairs is the same as above.

(4) Determine the number of points in the polarizer image

Determine the affine transformation matrix A between the roughly registered polarization sub-image and the reference polarization sub-image according to formula (6):

和

分别为参考偏振子图像和粗配准偏振子图像的第一个特征点对的坐标；

和

分别为参考偏振子图像和粗配准偏振子图像的第二个特征点对的坐标；

和

分别为参考偏振子图像和粗配准偏振子图像的第三个特征点对的坐标。in,

and

are the coordinates of the first feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the second feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the third feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image, respectively.

Determine the transformation coordinates (x ^cal , y ^cal ) of the remaining feature points of the roughly registered polarized sub-image according to formula (7):

Among them, (x ^rough , y ^rough ) are the coordinates of the remaining feature points in the roughly registered polarized sub-image.

The absolute value of the difference between the feature point transformation coordinates of the roughly registered polarized sub-image and the corresponding feature point coordinates of the reference polarized sub-image is less than 0.5, and the feature point is recorded as an interior point to determine the number of interior points in the polarized sub-image.

(5) Obtain the optimal registration coefficient of the polarized sub-image

Select any remaining three feature point pairs in the feature point pair, and repeat step (4) until all feature point pairs are traversed.

Determine the optimal registration coefficient A' according to formula (8):

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第一个特征点对的坐标；

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第二个特征点对的坐标；

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第三个特征点对的坐标。in,

and

are the coordinates of the first feature point pair of the reference polarized sub-image and the roughly registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the second feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the third feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest, respectively.

(6) Complete polarization image registration

Determine the coordinates (x ^reg , y ^reg ) of the polarized sub-image after registration according to formula (9):

According to formula (10), the registered polarization image I _reg (x′, y′) is obtained:

I _reg (x', y')=I _rough (x ^reg , y ^reg ) (10) where I _rough (x ^reg , y ^reg ) is the coarsely registered polarizer image at coordinates (x ^reg , y ^reg ) Strength of.

In the step (1) of the present invention for acquiring the polarized sub-image, the acquiring the original frame image includes a frame of divided aperture polarization including a 0° linearly polarized image, a 45° linearly polarized image, a 90° linearly polarized image, and a circularly polarized image. Image; the original frame image is cut into four polarized sub-images including 0° linearly polarized sub-image, 45° linearly polarized sub-image, 90° linearly polarized sub-image, and circularly polarized sub-image with the same size.

In the step of (3) extracting feature points of polarized sub-images of the present invention, the method for extracting the feature points of the reference polarized sub-image and the roughly registered polarized sub-images respectively is a scale-invariant feature transformation method or an accelerated robust feature method .

In the step of (3) extracting feature points of polarized sub-images of the present invention, the ratio of the minimum Euclidean distance to the next-minimum Euclidean distance is not greater than the set threshold condition, and the threshold value is 0.5-0.8.

The invention adopts the coarse registration of the polarized sub-image, which reduces the amount of calculation when extracting the feature points of the polarized sub-image, and determines the optimal registration coefficient when the number of points in the polarized sub-image is the largest by traversing the feature points to complete the polarized sub-image. The high-precision, fast and fully automatic registration between the two cameras solves the technical problem of positional deviation between the sub-images of the sub-aperture polarization, and improves the image quality of the polarization imaging using the sub-aperture polarization camera, which can be used for the polarization of the sub-aperture. The camera is imaging polarized.

Description of drawings

FIG. 1 is a flow chart of Embodiment 1 of the present invention.

FIG. 2 is a 0° linearly polarized sub-image of Example 1 of the present invention.

FIG. 3 is a registration result of a 45° linearly polarized sub-image in Example 1 of the present invention.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and examples, but the present invention is not limited to the following embodiments.

Example 1

The image registration method of the sub-aperture polarization camera of the present embodiment consists of the following steps (see FIG. 1 ):

(1) Acquisition of polarized sub-images

Shoot the target with a split-aperture polarization camera, obtain the original frame image, cut the original frame image into four polarization sub-images, select any one of the four polarization sub-images as the reference polarization sub-image, and the remaining three polarization sub-images as the polarized sub-image to be registered.

The acquisition of the original frame image in this embodiment includes a frame of aperture-divided polarized image including a 0° linearly polarized image, a 45° linearly polarized image, a 90° linearly polarized image, and a circularly polarized image; the original frame image is cut into four images. The polarized sub-images include 0° linearly polarized sub-images, 45° linearly polarized sub-images, 90° linearly polarized sub-images, and circularly polarized sub-images with the same size. The 0° linearly polarized sub-image serves as the reference polarized sub-image (see Figure 2).

(2) Coarse registration of polarized sub-images

Determine the Fourier transform F(ω _x ,ω _y ) of the reference polarizer image according to equation (1):

F(ω _x ,ω _y )=F{f(x,y)} (1)

Among them, f(x, y) is the reference polarized sub-image of the coordinate (x, y), and F{} represents the Fourier transform.

Determine the Fourier transform G(ω _x ,ω _y ) of the polarized sub-image to be registered according to formula (2):

G(ω _x ,ω _y )=F{g(x,t)} (2)

Among them, g(x, y) is the polarized sub-image to be registered with coordinates (x, y).

Determine the impulse function δ(x-dx, y-dy) according to formula (3):

Among them, dx is the displacement of the polarized sub-image to be registered in the horizontal direction, dy is the displacement of the polarized sub-image to be registered in the vertical direction, F ^-1 {} is the inverse Fourier transform, F ^* (ω _x , ω _y ) is the complex conjugate of F(ω _x ,ω _y ).

Determine the coordinates (x', y') of the roughly registered polarized sub-image according to formula (4):

Among them, (x ₀ , y ₀ ) are the coordinates of the polarized sub-image to be registered.

Determine the rough registration polarized sub-image I _rough (x ₀ , y ₀ ) according to formula (5):

I _rough (x ₀ ,y ₀ )=I(x′,y′) (5)

Among them, I(x', y') is the intensity of the polarized sub-image to be registered at the coordinates (x', y').

(3) Extract the feature points of polarized sub-images

The feature points of the reference polarizer image and the roughly registered polarizer image are extracted respectively. In this embodiment, the scale-invariant feature transformation method is used. The scale-invariant feature transformation method has been described in Distinctive image features from scale-invariant keypoints[J].International Journal of Computer Vision, 2004, 60(2):91-110. Disclosure. For any feature point in the reference polarized sub-image, select the two feature points with the smallest Euclidean distance and the second smallest Euclidean distance in the rough registration polarized sub-image, according to the ratio of the smallest Euclidean distance to the second smallest Euclidean distance is not greater than the set The threshold condition is 0.5 to 0.8, and the threshold in this embodiment is 0.6.

The selected feature point in the reference polarization sub-image and the feature point with the smallest Euclidean distance in the coarse registration polarization sub-image form a feature point pair, and the judgment method for other feature points to form a feature point pair is the same as above.

(4) Determine the number of points in the polarizer image

Determine the affine transformation matrix A between the roughly registered polarization sub-image and the reference polarization sub-image according to formula (6):

和

分别为参考偏振子图像和粗配准偏振子图像的第一个特征点对的坐标；

和

分别为参考偏振子图像和粗配准偏振子图像的第二个特征点对的坐标；

和

分别为参考偏振子图像和粗配准偏振子图像的第三个特征点对的坐标。in,

and

are the coordinates of the first feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the second feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the third feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image, respectively.

Determine the transformation coordinates (x ^cal , y ^cal ) of the remaining feature points of the roughly registered polarized sub-image according to formula (7):

Among them, (x ^rough , y ^rough ) are the coordinates of the remaining feature points in the roughly registered polarized sub-image.

The absolute value of the difference between the feature point transformation coordinates of the roughly registered polarized sub-image and the corresponding feature point coordinates of the reference polarized sub-image is less than 0.5, and the feature point is recorded as an interior point to determine the number of interior points in the polarized sub-image.

(5) Obtain the optimal registration coefficient of the polarized sub-image

Select any remaining three feature point pairs in the feature point pair, and repeat step (4) until all feature point pairs are traversed.

Determine the optimal registration coefficient A' according to formula (8):

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第一个特征点对的坐标；

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第二个特征点对的坐标；

和

分别为内点数最多时的参考偏振子图像和粗配准偏振子图像的第三个特征点对的坐标。in,

and

are the coordinates of the first feature point pair of the reference polarized sub-image and the roughly registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the second feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the third feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest.

(6) Complete polarization image registration

Determine the coordinates (x ^reg , y ^reg ) of the polarized sub-image after registration according to formula (9):

According to formula (10), the registered polarization image I _reg (x′, y′) is obtained:

I _reg (x′,y′)=I _rough (x ^reg ,y ^reg ) (10)

Wherein, I _rough (x ^reg , y ^reg ) is the intensity of the roughly registered polarized sub-image at coordinates (x ^reg , y ^reg ).

Complete the image registration method of the split aperture polarization camera. A 45° linearly polarized sub-image after registration is obtained (see FIG. 3 ). It can be seen from FIG. 2 and FIG. 3 that the present invention realizes the image registration of the sub-image of the polarization sub-image with different apertures.

Example 2

The image registration method of the sub-aperture polarization camera of the present embodiment consists of the following steps:

(1) Acquisition of polarized sub-images

This procedure is the same as in Example 1.

(2) Coarse registration of polarized sub-images

This procedure is the same as in Example 1.

(3) Extract the feature points of polarized sub-images

Extract the feature points of the reference polarization sub-image and the coarse registration polarization sub-image respectively. For any feature point in the reference polarization sub-image, select the two feature points with the smallest Euclidean distance and the second smallest in the coarse registration polarization sub-image. , according to the condition that the ratio of the minimum Euclidean distance to the second minimum Euclidean distance is not greater than the set threshold condition, the threshold value is 0.5-0.8, and the threshold value in this embodiment is 0.5.

The selected feature point in the reference polarization sub-image and the feature point with the smallest Euclidean distance in the coarse registration polarization sub-image form a feature point pair, and the judgment method for other feature points to form a feature point pair is the same as above.

Other steps are the same as in Example 1.

Complete the image registration method of the split aperture polarization camera.

Example 3

The image registration method of the sub-aperture polarization camera of the present embodiment consists of the following steps:

(1) Acquisition of polarized sub-images

This procedure is the same as in Example 1.

(2) Coarse registration of polarized sub-images

This procedure is the same as in Example 1.

(3) Extract the feature points of polarized sub-images

Extract the feature points of the reference polarization sub-image and the coarse registration polarization sub-image respectively. For any feature point in the reference polarization sub-image, select the two feature points with the smallest Euclidean distance and the second smallest in the coarse registration polarization sub-image. , according to the condition that the ratio of the minimum Euclidean distance to the second minimum Euclidean distance is not greater than the set threshold condition, the threshold value is 0.5-0.8, and the threshold value in this embodiment is 0.8.

The selected feature point in the reference polarization sub-image and the feature point with the smallest Euclidean distance in the coarse registration polarization sub-image form a feature point pair, and the judgment method for other feature points to form a feature point pair is the same as above.

Other steps are the same as in Example 1.

Complete the image registration method of the split aperture polarization camera.

Example 4

In the above embodiments 1 to 3, the image registration method of the split aperture polarization camera of this embodiment consists of the following steps:

(1) Acquisition of polarized sub-images

This procedure is the same as in Example 1.

(2) Coarse registration of polarized sub-images

This procedure is the same as in Example 1.

(3) Extract the feature points of polarized sub-images

The feature points of the reference polarizer image and the roughly registered polarizer image are extracted respectively. In this embodiment, the accelerated robust feature method is used. For any feature point in the reference polarizer image, the Euclidean distance from the roughly registered polarizer image is selected. The minimum and second minimum feature points, according to the ratio of the minimum Euclidean distance to the second minimum Euclidean distance is not greater than the set threshold condition, the threshold value is 0.5 to 0.8, the threshold value of this embodiment and the corresponding implementation Example is the same.

The selected feature point in the reference polarization sub-image and the feature point with the smallest Euclidean distance in the coarse registration polarization sub-image form a feature point pair, and the judgment method for other feature points to form a feature point pair is the same as above.

Other steps are the same as in Example 1.

Complete the image registration method of the split aperture polarization camera.

In order to verify the beneficial effects of the present invention, the inventor conducts experiments on the images of the sub-aperture polarization camera using the image registration method of the sub-aperture polarization camera according to Embodiment 1 of the present invention, and the experimental conditions are as follows.

1. Experimental conditions

The experimental test environment is a HP computer with Windows l0 (64-bit) operating system, and its configuration is Inter Corei3-10105F, 16GB memory, and the experimental operation is performed on the MATLAB R2019b platform.

2. Introduction of experimental data

The original frame images captured by the SAR camera were captured by the SAR camera at the Advanced Optical Imaging Laboratory of Shaanxi Normal University.

3. Evaluation indicators

Structural similarity index (SSIM) and normalized mutual information (NMI) were used as evaluation metrics. Determine the structural similarity index according to formula (11):

SSIM(R,T)=L(R,T)C(R,T)S(R,T) (11)

Among them, R and T represent two images with the same resolution, L(R, T) represents the luminance correlation function of the two images, C(R, T) represents the contrast correlation function of the two images, and S(R, T) represents the two images. Structural correlation function of the image, μ _R and μ _T represent the respective gray mean values of the two images, σ _R and σ _T represent the respective gray standard deviations of the two images, and σ _RT represents the difference between the image R and the image T. The grayscale covariances, C ₁ , C ₂ and C ₃ are small positive constants designed to avoid instability when the denominator is 0 or close to 0.

Determine the normalized mutual information according to formula (12):

Among them, M and N represent two images of the same resolution and the same gray level, H(M) represents the entropy of image M, H(N) represents the entropy of image N, and H(M, N) represents the image The relative mean information between M and image N, m and n represent the gray value of any pixel in image M and image N, respectively, P _M (m) represents the edge probability density function of image M, P _N (n) represents the edge probability density function of image N, and _PMN (m,n) represents the joint probability density function between image M and image N.

Quantitative evaluation of the similarity between the pre-registered and unregistered images of the aperture polarization camera.

The test was carried out according to the method of Example 1, the structural similarity index between the polarization images with different apertures is shown in Table 1, and the normalized mutual information between the polarization images with different apertures is shown in Table 2.

Table 1 Structural Similarity Index (SSIM) of every two polarized images of sub-aperture

Table 2 Normalized Mutual Information (NMI) of every two polarized images of sub-aperture

In Table 1, SSIM_0_45 represents the structural similarity index between the 0° linearly polarized sub-image and the 45° linearly polarized sub-image; SSIM_0_90 represents the structural similarity index between the 0° linearly polarized sub-image and the 90° linearly polarized sub-image; SSIM_0_C represents 0 ° Structural similarity index between linearly polarized sub-image and circularly polarized sub-image; SSIM_45_90 represents the structural similarity index between 45° linearly polarized sub-image and 90° linearly polarized sub-image; SSIM_45_C represents 45° linearly polarized sub-image and circularly polarized sub-image The structural similarity index of ; SSIM_90_C represents the structural similarity index of the 90° linearly polarized sub-image and the circularly polarized sub-image.

In Table 2, NMI_0_45 represents the normalized mutual information of 0° linearly polarized sub-image and 45° linearly polarized sub-image; NMI_0_90 represents the normalized mutual information of 0° linearly polarized sub-image and 90° linearly polarized sub-image; NMI_0_C Represents the normalized mutual information of the 0° linearly polarized sub-image and the circularly polarized sub-image; NMI_45_90 represents the normalized mutual information of the 45° linearly polarized sub-image and the 90° linearly polarized sub-image; NMI_45_C represents the 45° linearly polarized sub-image and the The normalized mutual information of the circularly polarized sub-image; NMI_90_C represents the normalized mutual information of the 90° linearly polarized sub-image and the circularly polarized sub-image.

It can be seen from Table 1 and Table 2 that, compared with the four unregistered polarized images, the average structural similarity index of the four polarized images after registration in the present invention is increased by 72.08%, and the normalized mutual information is increased by 18.5% on average, indicating that The invention can well realize the image registration of the split-aperture polarization camera.

Claims (4)

1. an image registration method of a sub-aperture polarization camera is characterized in that being made up of the following steps: (1) Acquisition of polarized sub-images Shoot the target with a split-aperture polarization camera, obtain the original frame image, cut the original frame image into four polarization sub-images, select any one of the four polarization sub-images as the reference polarization sub-image, and the remaining three polarization sub-images as the polarized sub-image to be registered; (2) Coarse registration of polarized sub-images Determine the Fourier transform F(ω _x , ω _y ) of the reference polarizer image according to formula (1): F(ω _x , ω _y )=F{f(x, y)} (1) Among them, f(x, y) is the reference polarized sub-image of the coordinate (x, y), and F{} represents the Fourier transform; Determine the Fourier transform G(ω _x , ω _y ) of the polarized sub-image to be registered according to equation (2): G(ω _x , ω _y )=F{g(x, y)} (2) Wherein, g(x, y) is the polarized sub-image to be registered with coordinates (x, y); Determine the impulse function δ(x-dx, y-dy) according to formula (3):

Among them, dx is the displacement of the polarized sub-image to be registered in the horizontal direction, dy is the displacement of the polarized sub-image to be registered in the vertical direction, F ^-1 {} is the inverse Fourier transform, F ^* (ω _x , ω _y ) is the complex conjugate of F(ω _x , ω _y ); Determine the coordinates (x', y') of the roughly registered polarized sub-images according to formula (4):

Wherein, (x ₀ , y ₀ ) is the coordinate of the polarized sub-image to be registered; Determine the rough registration polarized sub-image I _rough (x ₀ , y ₀ ) according to formula (5): I _rough (x ₀ , y ₀ )=I(x′, y′) (5) Among them, I(x', y') is the intensity of the polarized sub-image to be registered at the coordinates (x', y'); (3) Extract the feature points of polarized sub-images Extract the feature points of the reference polarization sub-image and the coarse registration polarization sub-image respectively. For any feature point in the reference polarization sub-image, select the two feature points with the smallest Euclidean distance and the second smallest in the coarse registration polarization sub-image. , according to the condition that the ratio of the minimum Euclidean distance to the second minimum Euclidean distance is not greater than the set threshold condition, the selected feature point in the reference polarized sub-image and the feature point with the smallest Euclidean distance in the coarsely registered polarized sub-image form a feature point pair, The judgment method of other feature points forming feature point pairs is the same as above; (4) Determine the number of points in the polarizer image Determine the affine transformation matrix A between the roughly registered polarization sub-image and the reference polarization sub-image according to formula (6):

in,

and

are the coordinates of the first feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the second feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively;

and

are the coordinates of the third feature point pair of the reference polarizer image and the roughly registered polarizer image, respectively; Determine the transformation coordinates (x ^cal , y ^cal ) of the remaining feature points of the roughly registered polarized sub-images according to formula (7):

Wherein, (x ^rough , y ^rough ) are the coordinates of the remaining feature points in the roughly registered polarized sub-image; The absolute value of the difference between the feature point transformation coordinates of the roughly registered polarized sub-image and the corresponding feature point coordinates of the reference polarized sub-image is less than 0.5, and the feature point is recorded as an interior point, and the number of interior points in the polarized sub-image is determined; (5) Obtain the optimal registration coefficient of the polarized sub-image Select any remaining three feature point pairs in the feature point pair, and repeat step (4) until all feature point pairs are traversed; Determine the optimal registration coefficient A' according to formula (8):

in,

and

are the coordinates of the first feature point pair of the reference polarized sub-image and the roughly registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the second feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest;

and

are the coordinates of the third feature point pair of the reference polarized sub-image and the coarsely registered polarized sub-image when the number of inner points is the largest; (6) Complete polarization image registration Determine the coordinates (x ^reg , y ^reg ) of the polarized sub-image after registration according to formula (9):

The registered polarization image I _reg (x', y') is obtained according to formula (10): I _reg (x', y')=I _rough (x ^reg , y ^reg ) (10) Wherein, I _rough (x ^reg , y ^reg ) is the intensity of the roughly registered polarized sub-image at coordinates (x ^reg , y ^reg ). 2. The image registration method of a split-aperture polarization camera according to claim 1, characterized in that: in the step of (1) acquiring the polarized sub-image, the acquired original frame image comprises a frame containing a 0° linearly polarized image , 45° linearly polarized image, 90° linearly polarized image, divided aperture polarized image of circularly polarized image; described cutting the original frame image into four polarized sub-images including 0 ° linearly polarized sub-image, 45 ° linear polarized sub-image with the same size Polarized sub-image, 90° linearly polarized sub-image, circularly polarized sub-image. 3. the image registration method of the aperture polarization camera according to claim 1, is characterized in that: in (3) in the step of extracting polarized sub-image feature points, the described extraction of reference polarized sub-image and coarse registration polarization respectively The method of the feature points of the sub-image is the scale-invariant feature transformation method or the accelerated robust feature method. 4. the image registration method of the split aperture polarization camera according to claim 1, is characterized in that: in (3) in the step of extracting polarized sub-image feature points, described according to the minimum Euclidean distance and the second minimum Euclidean distance The ratio is not greater than the set threshold condition, and the threshold value ranges from 0.5 to 0.8.

Images