Sea surface anti-flare method based on rotation polarization of unmanned aerial vehicle

Abstract

The invention relates to a sea surface anti-flare method based on rotation polarization of an unmanned aerial vehicle, and the method comprises the steps: arranging a rotatable polarizing film in front of a lens of an image collection device, driving the polarizing film to rotate through a stepping motor, and enabling the image collection device to collect an image at a fixed frequency along with the rotation of the polarizing film; all the images collected by the polaroid in one rotation period are evaluated, the saturated pixel proportion, the contrast ratio and the information entropy of each image are calculated, a comprehensive evaluation function is established, the comprehensive evaluation functions of all the images in the period are compared, and the image with the optimal comprehensive evaluation function is screened out to serve as the optimal anti-blazing image. According to the method, the motor drives the polaroid to continuously rotate back and forth at a high speed, it is ensured that the polarization angle range of the polaroid is traversed in an extremely short time, image frames with the vibration transmission direction of the polaroid perpendicular to the main polarization direction of flare light can be captured inevitably, artifacts or information loss occurring in a traditional digital image processing method is avoided, and the image quality is improved. And efficient and reliable flare suppression capability is realized.

Description

Technical Field

[0001] This invention belongs to the field of UAV optical technology, specifically relating to a method for preventing glare on the sea surface based on the rotational polarization of a UAV. Background Technology

[0002] Natural sunlight, also known as unpolarized light, exhibits electromagnetic wave transverse vibrations in a plane perpendicular to its propagation direction, encompassing all possible directions with statistically evenly distributed amplitudes. When the vibrations of light waves lose this symmetry, it is called polarized light. Based on the direction of electric field vibration, polarized light can be classified as linearly polarized, circularly polarized, and elliptically polarized. The electric and magnetic fields of electromagnetic waves are perpendicular to each other; typically, the polarization direction of the electric field is defined as the polarization direction of the electromagnetic wave. Glare is essentially reflected light with a specific polarization direction. When natural light travels from air to the interface of a different medium (such as water), the polarization state of the light changes. For non-metallic surfaces, such as water, reflected light is typically partially polarized. This means that the intensity of the vibrational component perpendicular to the incident plane (s-polarized light) in the reflected light is higher than that of the vibrational component parallel to the incident plane (p-polarized light). This polarization phenomenon caused by reflection is the direct cause of the polarization characteristics of glare on the sea surface. A key physical phenomenon is Brewster's angle. When natural light is incident on the interface of a medium at Brewster's angle, the reflected light becomes completely linearly polarized (s-polarized), and the reflected and refracted rays are perpendicular to each other. For water, the refractive index is approximately 1.33, corresponding to a Brewster's angle of approximately 53.1°. Since solar flares are formed by the reflection of sunlight from the sea surface, they exhibit this distinct polarization characteristic.

[0003] When using drones for sea surface target detection, the complexity of the environment poses a severe challenge to detection efficiency. Sea glare is one of the main interference factors; sea glare is a bright area formed by intense reflection of sunlight off the sea surface. Within the glare area, the solar radiation intensity is extremely high, often causing the detector to reach saturation. This results in the glare completely obscuring the target signal, severely limiting the subsequent detection and tracking capabilities.

[0004] When drones conduct maritime patrols, their onboard cameras are essentially optoelectronic devices, making them highly susceptible to interference from glare. Because drones fly along specific routes and at certain altitudes, their camera observation angles are relatively limited, making them more prone to prolonged exposure to glare. This platform characteristic makes drone cameras more susceptible to prolonged exposure to glare, leading to severe image saturation and significant loss of target information. This severely impacts the target identification performance of drones in maritime patrol, reconnaissance, and rescue missions. Existing methods for overcoming the interference of sea surface glare on optoelectronic detection suffer from the following problems: 1. Complex system structure and high cost: Some existing polarization glare suppression solutions, such as Chinese patent CN202210477421.8, adopt complex hardware integration and require independent polarization cameras and imaging cameras. This significantly increases the payload, size and manufacturing cost of the UAV platform. This complexity not only raises the threshold for research and development and production, but also limits its application in cost and weight-sensitive UAVs.

[0005] 2. Insufficient real-time performance and heavy computational burden: Other solutions, such as Chinese patent CN202410865196.7, mainly rely on image post-processing algorithms for glare suppression, which cannot meet the real-time detection requirements. The method of Zhang Congli et al. (2023), which is closest to this invention, is designed for UAV platforms, but its solution relies on complex physical model construction and real-time polarization information calculation, resulting in a large system computation and high hardware requirements. It is difficult to deploy efficiently on UAVs with strict limitations on power consumption and processing capabilities. This computationally intensive method will consume a lot of airborne resources and affect the UAV's endurance and multi-tasking capabilities.

[0006] 3. Dependence on ideal environmental assumptions: The method proposed by Zhang Congli et al. (2023) makes certain assumptions about the light source (such as unbiased direct sunlight) and the polarization orientation (consistency) of the target / glare in the model derivation. Such a model based on ideal conditions may lead to poor suppression effect or residual glare in real complex marine environments (such as cloudy weather and atmospheric scattering light), which reduces its robustness and universality in real and variable scenarios.

[0007] 4. Limitations of manual adjustment: Traditional methods of adjusting polarizers manually or at preset angles cannot cope with the real-time dynamic changes in the camera's observation angle and the angle of sunlight incidence during UAV flight. This means that in actual missions, the polarizer often cannot remain at the optimal suppression angle, thus affecting the continuity and effectiveness of glare suppression and reducing the system's practicality in complex dynamic environments.

[0008] Therefore, there is an urgent practical need to develop a method that can effectively suppress sea surface glare and improve the detection performance of UAVs. Summary of the Invention

[0009] This invention belongs to the field of UAV optical technology and aims to solve the technical problem that the target recognition efficiency of UAVs is seriously reduced due to solar glare interference from the sea surface during tasks such as sea surface inspection, reconnaissance, and rescue. The method of this invention is to install a polarizer driven by a stepper motor in front of the camera lens of the UAV, and to comprehensively select the best image as the image with the best glare suppression effect in the current scene by evaluating whether there are overexposed areas in the current image, maximizing the image saturation contrast value, and evaluating the amount of information and detail contained in the image, thereby achieving effective suppression of glare reflected from the sea surface.

[0010] To achieve the above objectives, the present invention provides the following technical solution: A method for preventing glare from images based on rotational polarization involves placing a rotatable polarizer in front of the lens of an image acquisition device carried by a UAV. The polarizer is driven to rotate by a stepper motor. As the polarizer rotates, the image acquisition device acquires images at a fixed frequency. All images acquired by the polarizer in one rotation cycle are evaluated, and the saturation pixel ratio, contrast, and information entropy of each image are calculated. A comprehensive evaluation function is established, and the comprehensive evaluation functions of all images within the cycle are compared. The image with the optimal comprehensive evaluation function is selected as the image with the best glare prevention.

[0011] The method for calculating the saturated pixel ratio is to count the percentage of pixels with gray values ​​greater than the saturation threshold out of the total number of pixels in the image.

[0012] The method for calculating the contrast ratio is to select target pixels with grayscale values ​​higher than the contrast ratio threshold, then find the region with the largest connected area of ​​the target pixels as the glare region, and calculate the contrast ratio between the glare region and the set target region.

[0013] The method for calculating the information entropy is to first generate a grayscale histogram of the image, count the number of pixels at each grayscale level from 0 to 255 to obtain the probability distribution of the grayscale levels, and then substitute the probability distribution into the information entropy formula to obtain the information entropy of the image.

[0014] The formula for the comprehensive evaluation function is as follows:

[0015] In the formula, It is the normalized value of contrast in the current image period; It is the normalized value of information entropy in the current image period; It is a weighting coefficient for the proportion of saturated pixels. It is the weighting coefficient of information entropy, and .

[0016] Furthermore, the polarization angle range of the polarizer is determined to be... ~ Then the real-time polarization angle Number of image frames acquired with single-pass rotation N The relationship is:

[0017] In the formula, This is the minimum polarization angle of the polarizer; This represents the maximum polarization angle of the polarizer; This refers to the number of image frames acquired during one unidirectional rotation, specifically during one clockwise (or counterclockwise) rotation of the polarizer. Images with different polarization angles; i’ The frame number within the current period, i.e. ,in i It is the cumulative frame sequence number since system startup, that is, counting from 0. This indicates dividing the cumulative frame sequence number by The remainder obtained; Current time (seconds); Based on the polarization angle range and the periodic image acquisition frequency Define the step angle of a stepper motor for:

[0018] Based on the step angle of the stepper motor Define the angular velocity of the motor as:

[0019] In the formula, This refers to the camera frame rate (frames per second).

[0020] Furthermore, the saturated pixel ratio The calculation method is as follows:

[0021] In the formula ， The coordinates in the image are ( The grayscale value of the pixel; This is the saturation threshold, a standard used to define overexposure; The total number of pixels in the image; The function iterates through all pixels in an image and counts those that satisfy the condition " The number of pixels.

[0022] Furthermore, the contrast The calculation method is as follows:

[0023] This represents the average grayscale value of all pixels within the illuminated area; This represents the average grayscale value of all pixels within the target area.

[0024] Furthermore, the information entropy The calculation method is as follows:

[0025] In the formula, The gray level of the image; The pixel with a grayscale value of 0 The probability of it appearing in the entire image.

[0026] The invention has the following advantages: This invention discloses a method for preventing glare on the sea surface based on rotational polarization using a drone. By driving a polarizer to rotate continuously at high speed using a motor, the polarization angle range of the polarizer is traversed in a very short time. This ensures that image frames whose transmission direction of the polarizer is perpendicular to the main polarization direction of the glare are captured, achieving the best glare suppression effect. This direct physical filtering mechanism avoids artifacts or information loss that may occur in traditional digital image processing methods, providing efficient and reliable glare suppression capabilities.

[0027] This invention discloses a method for preventing glare on the sea surface based on the rotation polarization of a drone. A stepper motor is connected to the outer frame of the polarizer via precision gears, driving it to rotate at high speed and smoothly. The entire motor-polarizer assembly is designed to be compact and lightweight to meet the payload requirements of the drone. This installation method is simple, universal and easy to operate, without the need for major modifications to the drone or camera, greatly reducing the difficulty and cost of deployment.

[0028] This invention discloses a method for preventing glare on the sea surface based on the rotation polarization of unmanned aerial vehicles (UAVs). It comprehensively selects the saturated pixel ratio, contrast, and information entropy as evaluation indicators. For all images obtained in a unit rotation cycle, it comprehensively selects the best image as the image with the best glare suppression effect in the current scene by evaluating whether there are overexposed areas in the current image, maximizing the image saturated contrast value, and evaluating the amount of information and detail contained in the image. Attached Figure Description

[0029] To more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0030] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size should still fall within the scope of the technical content disclosed in the invention, provided that they do not affect the effectiveness and purpose that the invention can achieve.

[0031] Figure 1 An exploded view of an image anti-glare device based on rotational polarization according to an embodiment of the invention; Figure 2 This is a flowchart illustrating a method for preventing glare on the sea surface based on the rotational polarization of a drone, as an embodiment of the invention.

[0032] In the picture: 1. Unmanned aerial vehicle (UAV); 2. Gimbal; 3. Image acquisition device; 4. Stepper motor; 5. Polarizer; 6. Drive gear; 7. Driven ring gear. Detailed Implementation

[0033] The following specific embodiments illustrate the implementation of the invention. Those skilled in the art can easily understand other advantages and effects of the invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the invention. Based on the embodiments of the invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the invention.

[0034] like Figure 1 As shown, an image anti-glare method based on rotational polarization includes a drone 1, a gimbal 2 mounted on the drone 1, an image acquisition device installed on the gimbal 2, a stepper motor 4 on the gimbal 2, and a rotatable polarizer 5 placed in front of the lens of the image acquisition device carried by the drone 1. The polarizer 5 is driven to rotate by the stepper motor 4. As the polarizer 5 rotates, the image acquisition device acquires images at a fixed frequency.

[0035] The polarizer 5 is a linear polarizer, a special type of optical filter whose core physical characteristic lies in its internal structure. This polarizer 5 is typically made of polyvinyl alcohol plastic film, and during manufacturing, the long-chain molecules within the film are aligned parallel to each other in the same direction. This unique molecular arrangement forms a "grid" or "transmission axis." When light passes through the polarizer 5, only light vibrations whose vibration direction is parallel to the transmission axis of the polarizer 5 can pass through smoothly, while vibrations perpendicular to this direction are absorbed or blocked. This principle allows the polarizer 5 to selectively filter light with specific polarization directions. When glare reflected from the sea surface passes through the linear polarizer 5, if the transmission direction of the polarizer 5 is perpendicular to the main polarization direction of the glare, the glare component will be blocked or attenuated to the maximum extent, thereby eliminating or reducing the glare.

[0036] Due to the presence of sea breezes, the sea surface is not an ideal plane, but can be considered as an aggregate of many tiny specular elements. Its overall reflection characteristics lie between ideal specular reflection and diffuse reflection, and it also exhibits directionality. This technique references the advanced GGX model (also known as the Trowbridge-Reitz model) to describe sea surface reflection. The GGX model considers the undulating attitude of the sea surface micro-elements, and its core is the use of a two-way reflectance distribution function (BRDF) to connect the diffuse reflection term. and specular reflection term This allows for a quantitative description of the sea surface's reflectivity. (1) In the formula, It is a two-way reflection distribution function; For diffuse reflection; For specular reflection; Let the incident light vector be denoted as . To detect the light vector; The diffuse reflection term mainly reflects the debiasing effect, and its expression is: (2) In the formula, Fresnel reflectivity; It is a constant (approximately equal to) (or inherent parameters of the material). Functions related to materials and wavebands; To detect the zenith angle.

[0037] The specular reflection term is related to the pose of the micro-surface element, and its expression is: (3) In the formula, Let be the probability density function of the normal direction of the micro-surface element; The occlusion factor; The angle of incidence at the zenith; The zenith angle is the normal to the micro-element.

[0038] The attitude of the micro-element is determined by the slope standard deviation, which is related to wind speed, as follows: (4) In the formula, 、 They are micro-facets 、 Standard deviation of directional slope; Wind speed ( ).

[0039] Probability density function of normal direction of micro-element Based on the definition of slope standard deviation: (5) In the formula, The azimuth angle of the normal to the micro-surface element.

[0040] To quantify the polarization characteristics of the reflected light, the BRDF function is passed through... Muller Matrix vectorization yields the polarization BRDF function: (6) In the formula, For diffuse reflection polarization term ( (Since it is an identity matrix, the debiasing effect can be ignored). This refers to the specular reflection polarization term; Specular reflection Muller matrix; The rotation matrix between the sea level and the micro-element is given by the rotation angle determined by the relationship between the micro-element normal and the incident / detection direction through a spherical triangle.

[0041] Specular reflection Muller matrix The parameter is related to Fresnel reflectivity: (7) In the formula, 、 They are respectively p Light, s Light reflectance; 、 They are respectively p Light, s Light reflectance coefficient.

[0042] Assuming the incident light is natural light, its Stokes The vector is: (8) Natural light is reflected by the sea surface. Stokes The vector is composed of the polarization BRDF function and the incident vector. Stokes The vector action yields: (9) In the formula, , Light intensity; 、 、 This represents the polarization component.

[0043] Based on reflection Stokes Vectors can be used to solve for the polarization angle of glare. AOP The expression is: (10) Based on the above formula, the Monte Carlo method is used for simulation analysis: the standard deviation of the micro-surface slope under different wind speeds is generated by equation (4), and the probability distribution of the micro-surface normal is obtained by substituting it into equation (5); combined with the incident light and probe light vectors, the polarization BRDF is calculated by equation (6), and the reflection is obtained by equation (9). Stokes Vector; finally, the glare polarization angle is solved by equation (10). AOP .

[0044] Simulation results show that under different wind speeds, the effect of wind speed changes on... Q, U The effects of the components can cancel each other out, and the polarization angle fluctuates little (without significant change); when the detection azimuth difference is 180° (within the main reflecting surface), the glare phenomenon is obvious, and the glare polarization angle calculated by equation (10) is at this time. AOP The polarizer 5 should be rotated within a certain angle range to ensure that the image that most effectively blocks and eliminates glare is captured. This is the calculated glare polarization angle. AOP The angle range is set to the polarization angle range of polarizer 5.

[0045] This invention achieves anti-glare on the sea surface by adding a high-speed rotating linear polarizer 5 driven by a stepper motor 4 in front of the camera lens of a drone 1. The output section of the stepper motor 4 is equipped with a drive gear 6. The polarizer 5 is placed vertically in front of the camera lens. The outer edge of the polarizer 5 is equipped with a driven ring tooth 7. The drive gear 6 meshes with the driven ring tooth 7, so that the stepper motor 4 drives the polarizer 5 to rotate. The gimbal 2 or the camera is equipped with an arc-shaped groove to accommodate the polarizer 5. The notch of the arc-shaped groove exposes the meshing point of the drive gear 6 and the driven ring tooth 7. The arc-shaped groove is used to fix the polarizer 5 so that the polarizer 5 can only rotate in front of the camera lens. Alternatively, to accommodate lenses of different diameters, a size-matched adapter ring can be used for connection.

[0046] To ensure the polarization angle of polarizer 5 With camera frame rate (frames per second) and the number of frames per round trip. (Frames per second) Related, this technology incorporates the current acquisition frame number. and current time (seconds), determine the polarization angle range of polarizer 5 as follows: ~ Then the real-time polarization angle Number of image frames acquired with single-pass rotation N The relationship is:

[0047] In the formula, This is the minimum polarization angle of the polarizer; This represents the maximum polarization angle of the polarizer; This refers to the single-pass periodic image acquisition frequency, which is the frequency during one clockwise (or counterclockwise) rotation of the polarizer. Images with different polarization angles, a complete round-trip cycle contains 2 N Frame image; i’ The frame number within the current period, i.e. ,in Indicates the cumulative frame sequence number remove The remainder obtained by this operation ensures The value range is always from 0 to 2. N -1, for example when The modulo operation yields .

[0048] i This is the cumulative frame sequence number since system startup, starting from 0, and its value is determined by... Confirmed, among which This indicates that the product of the camera frame rate and the current time is rounded down to a multiple of 1.

[0049] The current time (in seconds) since the system started; This function ensures that the polarization angle changes from [value] over a complete 2N frame period. linearly increase to (First N frames), then from Linear reduction of return (The following N frames).

[0050] Based on the polarization angle range and the periodic image acquisition frequency Define the step angle of stepper motor 4 for:

[0051] Based on the step angle of stepper motor 4 Define the angular velocity of the motor as:

[0052] In the formula, The camera frame rate (frames per second) is given when the angular velocity of stepper motor 4 is less than the frame number. At this time, the polarizer 5 is rotating in the positive direction, and the angular velocity of the motor is... / second; when the angular velocity of stepper motor 4 is greater than or equal to the frame number At this time, polarizer 5 rotates in the opposite direction, and the angular velocity of the motor is... / second. Considering the performance differences between different motors and camera devices in practical applications, and the varying requirements for image frame rate and acquisition time period depending on the task requirements, here... and It can be dynamically adjusted according to the actual equipment performance and needs.

[0053] When the system is in operation, UAV 1 performs inspection missions at sea. The image acquisition device acquires images of the sea surface in real time. Light incident on the lens passes through a polarizer 5 driven by a stepper motor 4 before entering the sensor of the image acquisition device. Assuming that the stepper motor 4 continuously drives the polarizer 5 to rotate at a preset speed of 30° / second, with a step size of 1, the image acquisition device continuously acquires images at a high frame rate of 30 frames / second. The frame rate of the image acquisition device corresponds to the rotation of the stepper motor 4. The stepper motor 4 performs 30 rotations per second, and takes a picture with each rotation, thus obtaining a series of images taken at different polarization angles in each round-trip cycle.

[0054] Since ideal glare-free reference images are unavailable in real-world dynamic environments, this technique employs No-Reference Image Quality Assessment (NR-IQA) metrics to evaluate the quality of each polarization angle image in the image sequence. This selects images with clear maritime target outlines and optimal glare suppression. These metrics objectively measure the visual quality of the images, such as contrast, sharpness, saturation, and the presence of overexposure or underexposure, without requiring comparison with the original distortion-free image. Images are continuously captured within one cycle of polarizer 5's movement. This image data is then transmitted to an airborne or ground-based image processing module. The image processing module uses the NR-IQA metrics to evaluate the image sequence in real time, selecting the best image with the highest contrast and lowest saturation for subsequent target recognition and analysis tasks.

[0055] All images acquired by polarizer 5 in one rotation cycle are evaluated. The saturation pixel ratio, contrast and information entropy of each image are calculated, and a comprehensive evaluation function is established. The comprehensive evaluation functions of all images in the cycle are compared, and the image with the best comprehensive evaluation function is selected as the image with the best anti-glare.

[0056] The saturated pixel ratio is calculated by statistically analyzing the proportion of pixels with grayscale values ​​greater than the saturation threshold out of the total number of pixels in the image. This proportion is used to evaluate the degree of overexposure in the image; a lower proportion indicates fewer saturated pixels due to flare, resulting in higher image quality. The saturated pixel ratio value... The calculation method is as follows:

[0057] In the formula, The coordinates in the image are ( i, j The grayscale value of a pixel is 0 (pure black) to 255 (pure white) for a standard 8-bit grayscale image.

[0058] This is the saturation threshold, a standard used to define overexposure. In this technique, it can be... The threshold is set to 250 and can be adjusted as needed. This means that any pixel with a grayscale value greater than 250 is considered a saturated pixel due to glare.

[0059] This represents the total number of pixels in the image, i.e., the image resolution (e.g., 1920×1080).

[0060] The function iterates through all pixels in an image and counts those that satisfy the condition " The number of pixels.

[0061] This indicates the percentage of saturated pixels out of the total pixels. The lower the value, the better the glare suppression effect.

[0062] The contrast ratio is used to evaluate the distinguishability between the target and the background (flare area). The higher the contrast ratio, the clearer the target. The contrast ratio is calculated by selecting target pixels with grayscale values ​​higher than a contrast threshold, then finding the region with the largest connected area of ​​the target pixels as the flare area, and calculating the contrast ratio between the flare area and the defined target area. C The calculation method is as follows:

[0063] The grayscale value is the average grayscale value of all pixels within the glare area. All pixels in the image with a grayscale value higher than 240 are marked. This threshold can be adjusted as needed. The region with the largest connected area among these pixels is defined as the "glare area." If no pixels have a grayscale value higher than this threshold, the connected region with the highest grayscale value in the captured image is selected as the "glare area." If a target object is identified using a target recognition model such as YOLO and a bounding box is output, the region enclosed by this bounding box is defined as the "target region" of the image.

[0064] This is the average grayscale value of all pixels within the target area. If the object detection model does not identify any target, then the central 10% area of ​​the current image (the central 10% area of ​​the image is selected proportionally) is assumed to be the "target area".

[0065] Calculated and Substituting into the formula, you can obtain the contrast value. Contrast value The higher the value, the better the separation between the target and the glare background, and the higher the image quality.

[0066] The information entropy is used to evaluate the richness of detail in an image. A higher information entropy indicates that the image contains more detailed information and has clearer textures. The information entropy is calculated by first generating a grayscale histogram of the image, then counting the number of pixels at each grayscale level (0-255) to obtain the probability distribution of each grayscale level. This probability distribution is then substituted into the information entropy formula to obtain the image's information entropy. H The calculation method is as follows:

[0067] In the formula, The gray level of the image; Is the grayscale value The probability of a pixel appearing in the entire image. The calculation method is as follows: First, generate the grayscale histogram of the image, count the number of pixels at each grayscale level (0 to 255), and then divide the number of pixels at each grayscale level by the total number of pixels in the image. You can get .

[0068] Taking an 8-bit grayscale image as an example, first convert the color image / sampled image to an 8-bit grayscale image and calculate the grayscale histogram to obtain the grayscale levels. k (0) k L -1) number of pixels Let the total number of pixels in the image be N Then gray level k The probability of its occurrence is The information entropy of an image is defined as H Information entropy H The larger the value, the more uniform the grayscale distribution, the richer the image details, and the higher the image quality.

[0069] A comprehensive evaluation function is established. To unify the dimensions and comprehensively consider the above indicators, a comprehensive quality score is calculated for each image in the image sequence within the acquisition period. The formula for the comprehensive evaluation function is as follows:

[0070] in, It is the normalized value of contrast in the current image period, for example, processed to the [0, 1] interval by max-min normalization.

[0071] It is the normalized value of information entropy in the current image period, for example, processed to the [0, 1] interval by max-min normalization.

[0072] The maximum-min normalization can be performed according to the following formula:

[0073] in, x For values ​​that need to be normalized; ɛ To avoid dividing by zero for a very small positive number.

[0074] It is a weighting coefficient for the proportion of saturated pixels. It is the weighting coefficient of information entropy, and These weights can be preset according to task requirements; for example, in a strong glare environment, they can be appropriately increased. The weighting; in tasks requiring precise target identification, can be increased. and The weights; under normal circumstances, can be set. , , .

[0075] Finally, the best image is determined by the image processing module, which iterates through and calculates the overall quality score of all images acquired within the acquisition period. ,Will The image with the highest value is denoted as Preliminary selection The image is the best candidate image for the current period, and a minimum quality threshold is introduced. Under normal circumstances, it can be set to 0.5. This value can be set according to actual needs. If the best candidate image is selected, it is determined to be the final valid image and used for subsequent target recognition, tracking, or situational awareness tasks; if If the images captured in this round do not meet the requirements, the images captured in this round are discarded and no images are output to the subsequent modules. The system re-detects the polarization angle region of the glare according to the aforementioned formulas (1)-(10), updates the polarization angle region of the polarizer 5, calculates the angle rotation speed of the stepper motor 4, and then enters the acquisition cycle. The image processing module then processes all image sequences acquired in the new cycle.

[0076] The entire image evaluation and screening module process is as follows: read the image sequence of one acquisition cycle from local storage → convert each image into a grayscale image → calculate the saturation pixel ratio, image contrast, and information entropy formulas for each image → calculate the comprehensive quality score for each image. →Comparison and → Output / Do not output the final best image.

[0077] Although the invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, such modifications or improvements made without departing from the spirit of the invention are all within the scope of the claims.

Claims

1. A method for preventing image glare based on rotational polarization, characterized in that: A rotatable polarizer is placed in front of the lens of the image acquisition device carried by the drone. The polarizer is driven to rotate by a stepper motor. As the polarizer rotates, the image acquisition device acquires images at a fixed frequency. All images acquired by the polarizer in one rotation cycle are evaluated, and the saturation pixel ratio, contrast and information entropy of each image are calculated. A comprehensive evaluation function is established, and the comprehensive evaluation functions of all images in the cycle are compared. The image with the best comprehensive evaluation function is selected as the best image for glare protection. The method for calculating the saturated pixel ratio is to count the percentage of pixels with gray values ​​greater than the saturation threshold out of the total number of pixels in the image. The method for calculating the contrast is to select target pixels with gray values ​​higher than the contrast threshold, then find the region with the largest connected area of ​​the target pixels as the glare region, and calculate the contrast between the glare region and the set target region. The method for calculating the information entropy is to first generate a grayscale histogram of the image, count the number of pixels at each grayscale level from 0 to 255, obtain the probability distribution of the grayscale levels, and then substitute the probability distribution into the information entropy formula to obtain the information entropy of the image. The formula for the comprehensive evaluation function is as follows: ; In the formula, It is the normalized value of contrast in the current image period; It is the normalized value of information entropy in the current image period; It is the weighting factor for contrast. It is a weighting coefficient for the proportion of saturated pixels. It is the weighting coefficient of information entropy, and .

2. The image anti-glare method based on rotational polarization according to claim 1, characterized in that: The polarization angle range of the polarizer is determined as follows: Then the real-time polarization angle Number of image frames acquired with single-pass rotation N The relationship is: ； In the formula, This is the minimum polarization angle of the polarizer; This represents the maximum polarization angle of the polarizer; This refers to the number of image frames acquired during one unidirectional rotation, specifically during one clockwise (or counterclockwise) rotation of the polarizer. Images with different polarization angles; The frame number within the current period, i.e. ,in i It is the cumulative frame sequence number since system startup, i.e. i Counting starts from 0. This indicates dividing the cumulative frame sequence number by The remainder obtained; Current time (seconds); Based on the polarization angle range and the periodic image acquisition frequency Define the step angle of a stepper motor for: ； Based on the step angle of the stepper motor Define the angular velocity of the motor as: ； In the formula, This refers to the camera frame rate (frames per second).

3. The image anti-glare method based on rotational polarization according to claim 1, characterized in that, The saturation pixel ratio The calculation method is as follows: ； In the formula ， The coordinates in the image are The grayscale value of the pixel; This is the saturation threshold, a standard used to define overexposure; The total number of pixels in the image; The function iterates through all pixels in an image and counts those that satisfy the condition. The number of pixels.

4. The image anti-glare method based on rotational polarization according to claim 1, characterized in that, The contrast The calculation method is as follows: ； This represents the average grayscale value of all pixels within the illuminated area; This represents the average grayscale value of all pixels within the target area.

5. The image anti-glare method based on rotational polarization according to claim 1, characterized in that, The information entropy The calculation method is as follows: ； In the formula, The gray level of the image; The pixel with a grayscale value of 0 The probability of it appearing in the entire image.

Images