De-scattering method based on single polarization image

Abstract

The invention discloses a scattering removal method based on a single polarization image, which is applied to enhancement of a target image in a turbid water body, and comprises the following steps: rotating a polaroid in a time-sharing manner, and obtaining a maximum polarization intensity image through a polarization camera; encoding the image with the maximum polarization intensity according to the Poincare ball trajectory, decoding based on the generalized Malus law, adjusting the polarization state and the input polarization state of a decoder, and obtaining an image with the minimum polarization intensity during orthogonality; carrying out frequency domain filtering processing by adopting a frequency domain low-pass filter to obtain maximum and minimum backscattered light intensities, and calculating the polarization degree of the backscattered light; and substituting into a single-polarization de-scattering model, and calculating a clear target object image. According to the method, the limitation of traditional multi-image processing is broken through, scattered light suppression can be achieved only through a single polarization image, the complexity of an imaging system is remarkably simplified, and the underwater image definition is effectively improved.

Description

Technical Field

[0001] The invention relates to the field of target image enhancement in a scattering medium environment, and in particular to a descattering method based on a single polarization image. Background Art

[0002] Underwater optical imaging has long faced the problem of image quality degradation due to the scattering effect of water bodies. The absorption and scattering of light by suspended particles causes underwater images to generally have defects such as low contrast, color distortion, and blurred details, which seriously restricts its application in fields such as marine exploration and ecological monitoring.

[0003] Traditional descattering methods based on physical models have achieved certain results, but there are still the following bottlenecks in actual scenarios: 1) Multi-image dependence and hardware limitations: Methods based on polarization characteristics usually require the acquisition of multiple sets of orthogonal polarization images or Stokes vector images, which require extremely high synchronization and stability of imaging equipment. 2) Scattered light calculation error: The intensity and polarization degree of backscattered light often have non-uniform distributions. Traditional polarization descattering technology assumes relevant information as a constant, which can easily lead to loss of target details in complex underwater scenes. 3) Usage scenario limitations: Traditional polarization descattering technology relies on the background area to calculate the polarization information of backscattered light, which is difficult to apply to target-dominated scenes.

[0004] In recent years, methods based on polarization physical models have shown significant advantages by separating the polarization characteristics of target light and scattered light. However, existing technologies are still limited by their reliance on multi-angle polarization data and the insufficient accuracy of estimating the low-frequency components of backscattered light. Summary of the invention

[0005] The problem to be solved by the present invention is to provide a descattering method based on a single polarization image, directly generate orthogonal polarization state images through the generalized Malus law, and realize frequency domain separation in combination with the low-frequency characteristics of backscattered light. While reducing the hardware complexity, it breaks through the limitations of traditional methods such as limited adaptability to scenes and inaccurate estimation of scattered light, and improves the imaging quality of the target.

[0006] The present invention adopts the following technical solution: a descattering method based on a single polarization image, using the generalized Malus law to generate an orthogonal polarization image pair, combined with a frequency domain analysis model to suppress backscattering interference, the specific steps are as follows: Step 1: Build a polarization descattering imaging system, rotate the linear polarizer in time, and obtain the brightest polarization image through the polarization camera as the maximum polarization intensity image ; Step 2: Input the maximum polarization intensity image , encoded according to the Poincare sphere trajectory and decoded based on the generalized Malus law, the decoding result is , according to the coding principle, adjust the polarization state of the decoder and the input polarization state, and obtain the minimum polarization intensity image when they are orthogonal ; Step 3: Based on the low-frequency nature of the backscattered light, a frequency domain low-pass filter is used to filter the maximum and minimum polarization intensity images. and Perform frequency domain filtering to obtain the maximum and minimum backscattered light intensities and , and calculate the polarization degree of the backscattered light ; Step 4: Polarization of backscattered light Bring in the single polarization descattering model to calculate a clear image of the target object .

[0007] Preferably, the polarization descattering imaging system in step 1 comprises: a light source, a linear polarizer, and a polarization camera; The light source emits active illumination light which passes through a linear polarizer to form linear polarized light, and irradiates the target object; the target object is submerged in turbid water containing a scattering medium, and a polarized image of the target object under the turbid water is obtained by the polarization camera; The linear polarizer is rotated in time division, and when the brightness of the polarization image obtained by the polarization camera is maximized, a maximum polarization intensity image is obtained. .

[0008] Preferably, in step 2, the polarization image is input , encoded according to the Poincare sphere trajectory, the light intensity distribution satisfies the generalized Malus law, and the decoding result is , which is expressed as follows: ; in, represents the total amplitude of the incident light, represents the ellipticity of the incident light, represents the azimuth of the incident light, , Represents a global parameter.

[0009] Preferably, in step 2, the total amplitude of the incident light The ellipticity of the incident light corresponds to the irradiance received by the polarization camera. Corresponding to the polarization state of the light wave received by the polarization camera, the azimuth angle of the incident light Corresponding to the polarization azimuth of the light wave received by the polarization camera.

[0010] Preferably, in step 2, the polarization state of the decoder and the input polarization state are adjusted to change for When the minimum polarization intensity image is obtained , which is expressed as follows: .

[0011] Preferably, in step 3, the maximum and minimum polarization intensity images are respectively subjected to frequency domain filtering processing by using Fourier transform and inverse Fourier transform and to obtain the maximum and minimum backscattered light intensities and , which is expressed as follows: ; ; wherein, , are respectively Fourier transform and inverse Fourier transform, is the transfer function of the frequency domain low-pass filter.

[0012] Preferably, in step 3, the degree of polarization of the backscattered light is calculated, and the formula is as follows: .

[0013] Preferably, in step 4, a clear target object image is obtained, which is expressed as: .

[0014] The technical solution of the present invention also provides: an electronic device, including: one or more processors; a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned de-scattering method based on a single polarization image.

[0015] The technical solution of the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the above-mentioned de-scattering method based on a single polarization image are implemented.

[0016] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects: 1. The method of the present invention innovatively combines the generalized Malus' law with frequency domain analysis. Only a single polarization image is required as input to achieve clear imaging in a scattering medium, without the need to collect complex input data such as multiple sets of orthogonal polarization images or Stokes vectors, greatly simplifying the data acquisition process and reducing the hardware complexity.

[0017] 2. The method of the present invention directly generates an orthogonally polarized state image through the generalized Malus' law, realizes frequency-domain separation by combining the low-frequency characteristics of backscattered light, and has an obvious scattering removal effect for the target-dominated scene.

[0018] 3. The method of the present invention is a fully automatic image processing method, which does not require background prior knowledge and human-computer interaction, can automatically obtain the scattering-removed imaging result, has strong robustness, and has a remarkable image information enhancement effect. It provides an innovative solution for underwater real-time high-definition imaging and can be applied to fields such as marine resource exploration, underwater ecological environment monitoring, and lost object recovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural diagram of the polarization scattering-removed imaging system in the embodiment of the present invention; Figure 2 It is a flowchart of the scattering-removed method based on a single polarization image of the present invention; Figure 3 It is a schematic diagram of the intensity imaging result in turbid underwater and the polarization scattering-removed imaging result based on the present invention in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the application will be further elaborated in detail below with reference to the accompanying drawings. The described embodiments are only a part of the embodiments involved in the present invention. All non-innovative embodiments made by other researchers in the field based on this embodiment belong to the protection scope of the present invention. At the same time, for the step numbers in the embodiments of the present invention, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0021] In an embodiment of the present invention, the polarization scattering-removed imaging system involved in the method of the present invention has a structure as Figure 1 shown, including: a polarization camera 1, an LED light source 2, a linear polarizer 3, a water tank 4, and a target object 5.

[0022] Specifically, the target object 5 is submerged in the water tank 4 filled with a scattering medium (whole milk). The light emitted by the LED light source 2 passes through the linear polarizer 3 to form linearly polarized light, irradiates on the target object 5 through the water tank 4, and a polarization camera 1 is used to obtain an image of the target object in turbid underwater.

[0023] The scattering-removed method of the present invention based on a single polarization image, as Figure 2 shown, the specific steps are as follows: 1. Fix the target object 5 at the bottom of the water tank 4, fix the LED light source 2 and the linear polarizer 3, rotate the direction of the linear polarizer 3 at different times, so that the brightness of the polarization image obtained by the polarization camera 1 is maximized and saved to obtain the maximum polarization intensity image. .

[0024] 2. Input the maximum polarization intensity image , encode it according to the Poincaré sphere trajectory, the polarization light intensity distribution of which satisfies the generalized Malus' law, decode the polarization image encoded with the known Poincaré sphere trajectory, and the decoding result is . When the polarization state of the decoder is adjusted to be orthogonal to the input polarization state, the image is the minimum polarization intensity image .

[0025] 3. Utilize the low-frequency property of the backscattered light and adopt the frequency-domain analysis method of the frequency-domain low-pass filter to estimate the maximum and minimum backscattered light intensities.

[0026] Specifically, use the Fourier transform and the inverse Fourier transform to perform frequency-domain filtering processing on the cross-polarization images and respectively. After that, obtain the backscattered light and . Furthermore, according to the definition of the degree of polarization, calculate the degree of polarization of the backscattered light .

[0027] 4. Substitute all the parameters into the single-polarization de-scattering model to calculate the clear target image and obtain the de-scattering result.

[0028] In this embodiment, the effectiveness of the method is verified by conducting experiments in scattering media with different concentrations. Taking the turbidity of 35 NTU as an example, as Figure 3 shown.

[0029] Specifically, the original intensity image of the turbid underwater obtained by the polarization camera 1 is as shown in (a) of Figure 3 . It can be seen that due to the interference of the backscattered light, the contrast of the target in the image is low, the details are blurred, and the recognition degree of the effective information is low. While the result obtained by the polarization de-scattering imaging method based on the present invention is as shown in (b) of Figure 3 . It can be seen that the method of the present invention effectively suppresses the degradation effect of the target information caused by the scattered light, realizes the enhancement effect of the target, and can easily identify the detailed information of the target.

[0030] Particularly, in the single-polarization de-scattering model of this embodiment, the underwater optical image is composed of the target light, the forward scattered light and the backscattered light. Ignoring the forward scattered part that causes image blurring in the total signal, the total intensity can be described as: ; wherein, represents the target light, represents the backscattered light.

[0031] The brightest maximum polarization intensity image and the darkest minimum polarization intensity image are respectively: ; ; wherein, , respectively represent the maximum and minimum target lights, , respectively represent the maximum and minimum backscattered lights.

[0032] The intensities of the target light and the backscattered light are: ; ;

[0033] According to the definition of the degree of polarization (DOP), the DOP of the target light and the DOP of the backscattered light are: ; ; wherein, , are respectively the DOP of the target light and the DOP of the backscattered light.

[0034] Since most of the target light is included in , the minimum target light intensity component is ignored: ; Therefore, the total intensity is: .

[0035] Thus, the clear target image is: ; Since in this embodiment, can be directly obtained. At the same time, according to the low-frequency property of the backscattered light, the intensity and the degree of polarization of the backscattered light can be estimated by using the frequency-domain analysis method based on the maximum and minimum polarization intensity images. Therefore, the key lies in obtaining the minimum polarization intensity image, that is, the cross-linear polarization image.

[0036] Furthermore, according to the generalized Malus' law, the decoding result of the polarization image encoding the known Poincaré sphere trajectory is: ; wherein, is the ellipticity of the incident light, is the azimuth angle of the incident light, which are respectively used to describe the elliptical shape and the spatial orientation of the polarization state; is the total amplitude of the incident light.

[0037] During the imaging process, corresponds to the irradiance received by the camera, corresponds to the polarization state of the light wave received by the camera, corresponds to the polarization azimuth angle of the light wave received by the camera.

[0038] and are global parameters, and their calculation formulas are respectively: ; ; wherein, and are the given coding trajectories: ; .

[0039] Therefore, by changing to , can be generated, which is expressed as: .

[0040] Furthermore, to enhance the target image and avoid human-computer interaction, it is necessary to estimate and remove the backscattered light. Since the backscattered light has low-frequency properties, in this embodiment, a frequency-domain low-pass filter is used to estimate the intensity of the backscattered light: ; ; wherein, and respectively represent the Fourier transform and the inverse Fourier transform, is the transfer function of the frequency-domain low-pass filter.

[0041] Preferably, is expressed as follows: ; wherein, represents the distance from the origin of frequency, is the standard deviation of the Gaussian distribution.

[0042] Furthermore, the DOP of the backscattered light is updated to: ; The clear target image is updated to: .

[0043] It can be seen that in view of the limitation that traditional polarization imaging requires the input of multiple polarization azimuth angle images, the present invention proposes a de-scattering imaging method based on a single polarization image, which combines the generalized Malus' law with frequency-domain analysis. Without the need for background and prior knowledge, clear imaging in a scattering medium can be achieved with only a single polarization input. By frequency-domain filtering to remove the backscattered light, in a scattering medium environment such as turbid underwater, the contrast and recognition of the target image are significantly improved, and finally the enhancement of the target image in the scattering medium is realized. The method of the present invention is simple and universal, has strong robustness, and has a remarkable image information restoration effect, and can be widely applied to underwater detection, ocean engineering monitoring, and underwater robot vision systems.

[0044] In an embodiment of the present invention, there is also provided an electronic device, including: one or more processors; a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors, the one or more processors implement the de-scattering imaging method based on a single polarization image described in the above embodiment.

[0045] In an embodiment of the present invention, there is also provided a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps in the de-scattering imaging method based on a single polarization image in the above embodiment are implemented.

[0046] The above is only a preferred embodiment of the present invention. It should be noted that: for those of ordinary skill in the art of the present technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A de-scattering method based on a single polarization image, characterized in that, Generate an orthogonal polarization image pair using the generalized Malus' law and suppress the backscattering interference by combining with the frequency-domain analysis model. The specific steps are as follows: Step 1: Construct a polarization de-scattering imaging system, rotate the linear polarizer in a time-sharing manner, and obtain the brightest polarization image through a polarization camera as the maximum polarization intensity image ; Step 2: Input the maximum polarization intensity image , encode according to the Poincaré sphere trajectory, decode based on the generalized Malus' law, and the decoding result is . Adjust the polarization state of the decoder and the input polarization state according to the encoding and decoding principle. When they are orthogonal, obtain the minimum polarization intensity image ; Step 3: Based on the low-frequency property of the backscattered light, use a frequency-domain low-pass filter to perform frequency-domain filtering on the maximum and minimum polarization intensity images and respectively, to obtain the maximum and minimum backscattered light intensities and , and calculate the degree of polarization of the backscattered light ; Step 4: Substitute the polarization degree of the backscattered light into the single-polarization depolarization model to calculate a clear image of the target object .

2. The de-scattering method based on a single polarization image according to claim 1, wherein The polarization de-scattering imaging system includes: a light source, a linear polarizer, and a polarization camera; The illumination light emitted by the light source forms linearly polarized light after passing through the linear polarizer and irradiates the target object; the target object is submerged in turbid water containing a scattering medium, and the polarization image of the target object under the turbid water is acquired by the polarization camera; Rotate the linear polarizer at intervals, obtain polarization images through a polarization camera, and select the brightest polarization image as the maximum polarization intensity image .

3. The de-scattering method based on a single polarization image according to claim 2, wherein, In step 2, the input polarized image , is encoded according to the Poincaré sphere trajectory, and the light intensity distribution satisfies the generalized Malus law. The decoding result is , which is expressed as follows: ； Among them, represents the total amplitude of the incident light, represents the ellipticity of the incident light, represents the azimuth angle of the incident light, 、 represent global parameters.

4. The de-scattering method based on a single polarization image according to claim 3, wherein In step 2, the total amplitude of the incident light corresponds to the irradiance received by the polarization camera, and the ellipticity of the incident light corresponds to the polarization state of the light wave received by the polarization camera, and the azimuth angle of the incident light corresponds to the polarization azimuth angle of the light wave received by the polarization camera.

5. The de-scattering method based on a single polarization image according to claim 3, characterized in that In step 2, adjust the polarization state of the decoder to the input polarization state and change to to obtain the minimum polarization intensity image , which is expressed as follows: 。 6. The de-scattering method based on a single polarization image according to claim 1, wherein In Step 3, the maximum and minimum polarization intensity images are respectively subjected to frequency-domain filtering processing using Fourier transform and inverse Fourier transform and to obtain the maximum and minimum backscattered light intensities and , which are expressed as follows: ； ； Among them, and are the Fourier transform and the inverse Fourier transform respectively, is the transfer function of the low-pass filter in the frequency domain.

7. The de-scattering method based on a single polarization image according to claim 6, characterized in that In step 3, calculate the degree of polarization of the backscattered light , and the formula is as follows: 。 8. The de - scattering method based on a single polarization image according to claim 7, wherein In step 4, a clear image of the target object is obtained based on the single-polarization depolarization model , which is expressed as: 。 9. An electronic device, characterized in that, including: One or more processors; A storage device storing one or more programs thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the de-scattering method based on a single polarization image as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and when the program is executed by a processor, the steps in the de-scattering method based on a single polarization image as described in any one of claims 1 to 8 are implemented.

Images