A polarization dehazing method suitable for outdoor scenes with large depth of field

Abstract

This invention provides a polarization dehazing method suitable for outdoor scenes with large depth of field. The method includes: acquiring an original image to be dehazed; calculating the transmission coefficient of the original image based on the local atmospheric light intensity before correction; dividing the transmission coefficient according to the depth of the original image to obtain a transmission coefficient threshold; calculating a correction coefficient based on the transmission coefficient threshold; calculating the corrected local atmospheric light intensity based on the correction coefficient; and obtaining a dehazed restored image of the original image based on the corrected local atmospheric light intensity. This invention divides the transmission coefficient based on the depth of the original image to obtain a transmission coefficient threshold, and then calculates the correction coefficient and the corrected local atmospheric light intensity based on the transmission coefficient threshold. This achieves differentiated dehazing intensity for different depth regions of the same original image, enabling uniform and natural dehazing effects in outdoor scenes with large depth of field, achieving a natural transition in image restoration, and improving the quality of image restoration.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a polarization dehazing method suitable for outdoor scenes with large depth of field. Background Technology

[0002] Images acquired in adverse weather conditions are often affected by fog. Because light is interfered with by particles in the air, the quality of captured images is often significantly degraded, resulting in poor contrast, inaccurate color fidelity, and loss of scene detail. This degradation in image quality is a common obstacle for a wide range of computer vision tasks, such as urban traffic monitoring, outdoor video surveillance, and autonomous driving. Therefore, dehazing processing is necessary to improve the quality of images acquired under adverse weather conditions.

[0003] Polarization dehazing is a commonly used method. Based on an atmospheric scattering model, it attributes the degradation of hazy images to the attenuation of transmitted light from objects and interference caused by atmospheric scattered light participating in imaging. Since these two factors have different polarization characteristics, image restoration can be achieved through polarization analysis. This method requires no prior knowledge of images and scenes under different atmospheric conditions, has low algorithm complexity, produces good image restoration quality, and is easy to implement. However, due to the depth-of-field differences in fog distribution in outdoor scenes with large depth of field, existing polarization-based dehazing methods often result in distant scenes appearing washed out or foreground scenes appearing dark after dehazing, thus affecting the quality of image restoration. Figure 1a and Figure 1b The image shown is a schematic diagram of a dehazed image based on existing polarization methods. Figure 1a The image shown exhibits a phenomenon where the background appears washed out, such as... Figure 1b The image shown exhibits a darkening effect in the foreground. Summary of the Invention

[0004] The purpose of this invention is to provide a polarization dehazing method suitable for outdoor scenes with large depth of field, which achieves differentiated dehazing intensity for different depth regions of the same original image, thereby achieving a natural transition in image restoration and improving the quality of image restoration.

[0005] To address the aforementioned technical problems, this invention provides a polarization dehazing method suitable for outdoor scenes with large depth of field, the method comprising:

[0006] Obtain the original image to be dehazed;

[0007] The transmission coefficient of the original image is calculated based on the local atmospheric light intensity before correction;

[0008] The transmission coefficient threshold is obtained by dividing the transmission coefficients according to the depth of the original image;

[0009] Calculate the correction coefficient based on the transmission coefficient threshold;

[0010] Calculate the corrected local atmospheric light intensity based on the correction coefficient; and

[0011] The dehazing restored image of the original image is obtained based on the corrected local atmospheric light intensity.

[0012] Optionally, the formula for calculating the transmission coefficient is:

[0013]

[0014] Where A′ is the local atmospheric light intensity before correction, A ∞ It is the atmospheric light intensity at infinity.

[0015] Optionally, the method for dividing the transmission coefficients according to the depth of the original image to obtain the transmission coefficient threshold includes the K-means clustering algorithm.

[0016] Optionally, the transmission coefficients are divided into distant, mid-range, and near scenes based on the depth of the original image, and the transmission coefficient thresholds t1 for the distant and mid-range scenes and t2 for the mid-range and near scenes are obtained.

[0017] Optionally, the correction coefficient increases as the depth of the original image increases or the transmission coefficient decreases.

[0018] Optionally, the formula for calculating the correction coefficient is:

[0019]

[0020] Optionally, provided that the minimum value of the correction coefficient in the distant view is greater than the minimum value of the correction coefficient in the mid-view, the correction coefficient in the distant view decreases as the depth of the original image increases or the transmission coefficient decreases.

[0021] Optionally, the formula for calculating the correction coefficient is:

[0022]

[0023] Optionally, the formula for calculating the corrected local atmospheric light intensity is:

[0024]

[0025] Where, θ A It is the atmospheric light polarization angle, P A S is the atmospheric polarization degree, S0 is the Stokes parameter, and I is the original image.

[0026] Optionally, the calculation formula for the dehazed and restored image is:

[0027]

[0028] Where (x,y) are the pixel positions in the original image.

[0029] In summary, the polarization dehazing method for outdoor scenes with large depth of field provided by this invention first acquires the original image to be dehazed. Then, it calculates the transmission coefficient of the original image based on the local atmospheric light intensity before correction. Next, it divides the transmission coefficient according to the depth of the original image to obtain a transmission coefficient threshold. Then, it calculates the correction coefficient based on the transmission coefficient threshold, followed by calculating the corrected local atmospheric light intensity based on the correction coefficient. Finally, it obtains the dehazed restored image of the original image based on the corrected local atmospheric light intensity. This invention divides the transmission coefficient based on the depth of the original image to obtain a transmission coefficient threshold, and then calculates the correction coefficient and the corrected local atmospheric light intensity based on the transmission coefficient threshold. This achieves differentiated dehazing intensity for different depth regions of the same original image, enabling uniform and natural dehazing effects in outdoor scenes with large depth of field, achieving a natural transition in image restoration, and improving the quality of image restoration.

[0030] Furthermore, while ensuring that the minimum correction coefficient in the distant view is greater than the minimum correction coefficient in the mid-view, the correction coefficient in the distant view decreases as the depth of the original image increases or the transmission coefficient decreases, thereby allowing the distant sky to transition smoothly, while retaining a slight sense of fog in the distant view, making the restored image more in line with visual laws. Attached Figure Description

[0031] Those skilled in the art will understand that the accompanying drawings are provided to better understand the invention and do not constitute any limitation on the scope of the invention. Wherein:

[0032] Figure 1a and Figure 1b This is a schematic diagram of an image after dehazing using existing polarization methods.

[0033] Figure 2 This is a flowchart of a polarization defogging method for outdoor scenes with large depth of field provided by an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram illustrating the relationship between the correction coefficient and the transmission coefficient provided in one embodiment of the present invention.

[0035] Figure 4 This is a schematic diagram showing the relationship between the correction coefficient and the transmission coefficient provided in another embodiment of the present invention.

[0036] Figure 5aThis is a schematic diagram of the sky before smoothing, provided in another embodiment of the present invention.

[0037] Figure 5b This is a schematic diagram of a smoothed sky image provided in another embodiment of the present invention.

[0038] Figure 6a This is a schematic diagram of an image showing a white background after defogging using existing technology, provided by an embodiment of the present invention.

[0039] Figure 6b This is a schematic diagram of an image after defogging using the defogging method of this invention, provided in one embodiment of the present invention.

[0040] Figure 6c This is a schematic diagram of a close-up image that appears black after defogging using existing technology, provided by an embodiment of the present invention.

[0041] Figure 7a This is a schematic diagram of an image before defogging for a scene with a large brightness range at the same depth of field, provided by one embodiment of the present invention.

[0042] Figure 7b This is a schematic diagram of a scene with a large brightness range under the same depth of field after defogging, provided by an embodiment of the present invention. Detailed Implementation

[0043] The polarization dehazing algorithm is based on the atmospheric scattering model: I = D + A = Lt(z) + A ∞ [1-t(z)], where D=Lt(z) is the intensity of the direct light in the scene, and A=A ∞ [1-t(Z)] represents the intensity of atmospheric scattered light. The image dehazing model is:

[0044]

[0045] Where L is the dehazed image, i.e., the finally recovered haze-free image; I is the original image to be dehazed, i.e., the hazy image obtained by the detection system, specifically the polarized image obtained using a polarization camera; (x, y) are the pixel positions in the image; and A... ∞ t(z) is the atmospheric light intensity at infinity, and t(z) is the transmission coefficient, which is the proportion of light that reaches the detection system after particle attenuation. The transmission coefficient reflects the relative distance of objects in the image.

[0046] In the above dehazing model, I is a known quantity, and A... ∞ Since it is a global variable, the key to defogging lies in estimating the local atmospheric light intensity A(x,y).

[0047] Traditional dehazing algorithms, assuming the elimination of quantum noise, estimate atmospheric light intensity values ​​for different regions of the entire image to be very similar. However, in outdoor scenes with a large depth of field, fog in the distance is often thicker and denser, while fog in the foreground is relatively thinner and lighter, resulting in differences in fog distribution across image depth. If similar local atmospheric light intensity estimates are used to dehaze different depth regions of the same image, it can easily lead to insufficient dehazing intensity (whitening) in the distance or excessive dehazing intensity (blackening) in the foreground.

[0048] To address the aforementioned problems, the inventors conducted research and proposed a polarization dehazing method based on image depth. First, the transmission coefficients, which reflect the relative distances of objects in the image, are separated. Then, the polarization degree in the atmospheric light intensity calculation of each pixel is additionally corrected based on the distance.

[0049] Further research has led to the present invention providing a polarization dehazing method suitable for outdoor scenes with large depth of field. This method achieves differentiated dehazing intensity for different depth regions of the same original image, thereby enabling a natural transition in image restoration and improving the quality of image restoration.

[0050] This invention provides a polarization dehazing method suitable for outdoor scenes with large depth of field, comprising:

[0051] Obtain the original image to be dehazed;

[0052] The transmission coefficient of the original image is calculated based on the local atmospheric light intensity before correction;

[0053] The transmission coefficient threshold is obtained by dividing the transmission coefficients according to the depth of the original image;

[0054] Calculate the correction coefficient based on the transmission coefficient threshold;

[0055] Calculate the corrected local atmospheric light intensity based on the correction coefficient; and

[0056] The dehazing restored image of the original image is obtained based on the corrected local atmospheric light intensity.

[0057] The present invention provides a polarization dehazing method suitable for outdoor scenes with large depth of field. Based on the depth of the original image, the transmission coefficient is divided to obtain a transmission coefficient threshold. Then, the correction coefficient and the corrected local atmospheric light intensity are calculated based on the transmission coefficient threshold. This achieves differentiated dehazing intensity for different depth regions of the same original image, enabling uniform and natural dehazing effects in outdoor scenes with large depth of field, achieving a natural transition in image restoration, and improving the quality of image restoration.

[0058] To make the objectives, advantages, and features of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the drawings are all in a very simplified form and are not drawn to scale, and are only used to facilitate and clarify the explanation of the embodiments of this invention. Furthermore, the structures shown in the drawings are often part of the actual structures. In particular, different figures may emphasize different aspects and may sometimes use different scales.

[0059] As used in this invention, the singular forms “a,” “an,” and “the” include plural objects; the term “or” is generally used to mean “and / or”; the term “a number” is generally used to mean “at least one”; the term “at least two” is generally used to mean “two or more”; furthermore, the terms “first,” “second,” and “third” are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined as "first," "second," and "third" may explicitly or implicitly include one or at least two of those features. The term "proximal" usually refers to the end closer to the operator, and the term "distal" usually refers to the end closer to the patient. "One end" and "the other end," as well as "proximal" and "distal," usually refer to two corresponding parts, which include not only the endpoints. The terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can be fixed connections, detachable connections, or integral connections; they can be mechanical connections or electrical connections; they can be direct connections or indirect connections through an intermediate medium; they can be internal connections between two components or the interaction between two components.

[0060] Furthermore, as used in this invention, the phrase "one element is disposed on another element" generally only indicates that there is a connection, coupling, cooperation, or transmission relationship between the two elements, and the connection, coupling, cooperation, or transmission between the two elements can be direct or indirect through an intermediate element. It should not be construed as indicating or implying a spatial positional relationship between the two elements, i.e., one element can be located arbitrarily inside, outside, above, below, or to one side of the other element, unless otherwise explicitly stated. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0061] Figure 2 This is a flowchart of a polarization dehazing method for outdoor scenes with large depth of field, provided by an embodiment of the present invention. Please refer to it. Figure 2 As shown, this invention provides a polarization dehazing method suitable for outdoor scenes with large depth of field, comprising:

[0062] S1: Obtain the original image to be dehazed;

[0063] S2: Calculate the transmission coefficient of the original image based on the local atmospheric light intensity before correction;

[0064] S3: Divide the transmission coefficients according to the depth of the original image to obtain the transmission coefficient threshold;

[0065] S4: Calculate the correction coefficient based on the transmission coefficient threshold;

[0066] S5: Calculate the corrected local atmospheric light intensity based on the correction coefficient; and

[0067] S6: Obtain the dehazing restored image of the original image based on the corrected local atmospheric light intensity.

[0068] The following is a detailed description of a polarization defogging method for outdoor scenes with large depth of field provided by an embodiment of the present invention.

[0069] In step S1, the original image to be dehazed is acquired. This is achieved using a foggy image obtained by the detection system, which is then used as the original image to be dehazed. The original image can be represented by I. In this embodiment, the original image I is a polarized image acquired using a polarization camera.

[0070] In step S2, the transmission coefficient of the original image is calculated based on the local atmospheric light intensity before correction.

[0071] The formula for calculating the transmission coefficient is:

[0072]

[0073] Where A′ is the local atmospheric light intensity before correction, A ∞ It is the atmospheric light intensity at infinity.

[0074] In this embodiment, the local atmospheric light intensity A′ before correction and the atmospheric light intensity A at infinity are compared. ∞ The transmission coefficient t of the original image is calculated. The transmission coefficient t reflects the relative distances of objects in the original image; a decrease in the transmission coefficient indicates an increase in the depth of the original image.

[0075] In step S3, the transmission coefficients are divided according to the depth of the original image to obtain a transmission coefficient threshold.

[0076] For example, the K-means clustering algorithm is used to divide the transmission coefficients. The transmission coefficients reflecting the relative distances of objects in the original image are divided according to the depth of the original image, for example, into distant, mid-range and near scenes, thereby obtaining the transmission coefficient threshold t1 between the distant and mid-range scenes and the transmission coefficient threshold t2 between the mid-range and near scenes.

[0077] In step S4, a correction coefficient is calculated according to the transmission coefficient threshold.

[0078] Figure 3 is a schematic diagram showing the relationship between the correction coefficient and the transmission coefficient provided by an embodiment of the present invention. Figure 3 In the figure, the abscissa is the transmission coefficient t, and the ordinate is the correction coefficient w. re , the transmission coefficient is divided into a long shot, a medium shot, and a close shot from small to large, and the transmission coefficient thresholds for the long shot and the medium shot are t1, and the transmission coefficient thresholds for the medium shot and the close shot are t2. Please refer to Figure 3 As shown in an embodiment of the present invention, the correction coefficient w re will increase as the depth of the original image increases, that is, as the transmission coefficient t decreases, so that regions with different depths will present similar restoration effects after defogging.

[0079] The calculation formula for the correction coefficient is:

[0080]

[0081] The correction coefficient w can be calculated according to the transmission coefficient threshold t1 between the long shot and the medium shot, the transmission coefficient threshold t2 between the medium shot and the close shot, and the transmission coefficient t. re .

[0082] In addition, outdoor scenes with large depth of field often have another feature, including a sky area with relatively high brightness, which has the characteristics of both very large depth and very high brightness. According to the defogging model L = (I-A) / t, 0<t<1, when defogging in this area, if a larger local atmospheric light intensity is blindly used, it is easy to amplify noise and overexposure of brightness, resulting in a sudden and noisy overexposed boundary.

[0083] Therefore, in another embodiment of the present invention, on the premise of ensuring that the overall correction coefficient in the long shot is greater than that in the medium shot, the correction coefficient can be appropriately reduced as the depth of the original image increases. Figure 4 is a schematic diagram showing the relationship between the correction coefficient and the transmission coefficient provided by another embodiment of the present invention. Figure 4 In the figure, the abscissa is the transmission coefficient t, and the ordinate is the correction coefficient w. re , the transmission coefficient is divided into a long shot, a medium shot, and a close shot from small to large, and the transmission coefficient thresholds for the long shot and the medium shot are t1, and the transmission coefficient thresholds for the medium shot and the close shot are t2. Please refer to Figure 4 As shown in another embodiment of the present invention, on the premise of ensuring that the minimum value of the correction coefficient in the long shot is greater than the minimum value of the correction coefficient in the medium shot, the correction coefficient w in the long shot reIt decreases as the depth of the original image increases or the transmission coefficient decreases. In the middle view and the close view, the correction coefficient w re increases as the depth of the original image increases or the transmission coefficient decreases. In the far view, appropriately reducing the correction coefficient as the depth of the original image increases can make the transition of the far view sky smooth. At the same time, a slight fog effect in the relatively distant area is retained in the far view, making the restored image more in line with the visual law.

[0084] Figure 5a is a schematic diagram of the image before sky smoothing provided by another embodiment of the present invention, Figure 5b is a schematic diagram of the image after sky smoothing provided by another embodiment of the present invention. Please refer to Figure 5a and Figure 5b As shown, the image after sky smoothing is more natural and more in line with the visual law.

[0085] In this embodiment, the calculation formula of the correction coefficient is:

[0086]

[0087] According to the transmission coefficient threshold t1 between the far view and the middle view, the transmission coefficient threshold t2 between the middle view and the close view, and the transmission coefficient t, the correction coefficient w of the far view (t < t1) can be calculated re and the correction coefficient w of the middle view and the close view (t > t1) re .

[0088] In step S5, the corrected local atmospheric light intensity is calculated according to the correction coefficient.

[0089] In this embodiment, the calculation formula of the corrected local atmospheric light intensity is:

[0090] <r

[0091] where θ A is the atmospheric light polarization angle, P A is the atmospheric light polarization degree, S0 is the Stokes parameter, and I is the original image. According to the correction coefficient w calculated in step S4 re and the above formula, the corrected local atmospheric light intensity A can be calculated. Since the correction coefficient w calculated in step S4 re involves the regions of different depths of the original image (i.e., involves the far view, the middle view, and the close view), the corrected local atmospheric light intensity A calculated in this step also includes the regions of different depths of the original image, that is, the differential defogging intensity of different depth regions of the same original image is achieved.

[0092] In step S6, the dehazing restored image of the original image is obtained based on the corrected local atmospheric light intensity.

[0093] The formula for calculating the dehazed and restored image is:

[0094]

[0095] Where I is the original image to be dehazed, and (x, y) are the pixel positions in the original image. The dehazed image L can be calculated based on the corrected local atmospheric light intensity A obtained in step S5, thus obtaining the dehazed image.

[0096] Figure 6a This is a schematic diagram of an image showing a blurred background after defogging using existing technology, provided by an embodiment of the present invention. Figure 6b This is a schematic diagram of an image after defogging using the defogging method of this invention, provided in one embodiment of the invention. Figure 6c This is a schematic diagram of a close-up image that appears dark after dehazing using existing technology, provided by an embodiment of the present invention. Please compare. Figure 6a and Figure 6b and comparison Figure 6b and Figure 6c As can be seen (please refer mainly to the positions marked by circles in the figure), the polarization dehazing method described in this embodiment avoids the problems of distant scenes appearing white and close-up scenes appearing black. It can achieve a uniform and natural dehazing effect in outdoor scenes with large depth of field, realize a natural transition in image restoration, and improve the quality of image restoration.

[0097] Figure 7a This is a schematic diagram of an image before dehazing for a scene with a large brightness range at the same depth of field, provided by one embodiment of the present invention. Figure 7b This is a schematic diagram of a dehazed image of a scene with a large brightness range at the same depth of field, provided by an embodiment of the present invention. Please refer to it. Figure 7a and Figure 7b As shown, for certain scenes with a large brightness range at the same depth of field (such as a locally overexposed sky, marked by circles in the figure), the polarization dehazing method provided in this embodiment increases the brightness difference while dehazing, further improving the image restoration quality.

[0098] In summary, the polarization dehazing method for outdoor scenes with large depth of field provided by this invention first acquires the original image to be dehazed. Then, it calculates the transmission coefficient of the original image based on the local atmospheric light intensity before correction. Next, it divides the transmission coefficient according to the depth of the original image to obtain a transmission coefficient threshold. Then, it calculates the correction coefficient based on the transmission coefficient threshold, followed by calculating the corrected local atmospheric light intensity based on the correction coefficient. Finally, it obtains the dehazed restored image of the original image based on the corrected local atmospheric light intensity. This invention divides the transmission coefficient based on the depth of the original image to obtain a transmission coefficient threshold, and then calculates the correction coefficient and the corrected local atmospheric light intensity based on the transmission coefficient threshold. This achieves differentiated dehazing intensity for different depth regions of the same original image, enabling uniform and natural dehazing effects in outdoor scenes with large depth of field, achieving a natural transition in image restoration, and improving the quality of image restoration.

[0099] Furthermore, while ensuring that the minimum correction coefficient in the distant view is greater than the minimum correction coefficient in the mid-view, the correction coefficient in the distant view decreases as the depth of the original image increases or the transmission coefficient decreases, thereby allowing the distant sky to transition smoothly, while retaining a slight sense of fog in the distant view, making the restored image more in line with visual laws.

[0100] The above description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any person skilled in the art can make possible changes and modifications to the technical solutions of the present invention by utilizing the methods and techniques disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the protection scope of the technical solutions of the present invention.

Claims

1. A polarization dehazing method suitable for outdoor scenes with large depth of field, characterized in that, The method includes: Obtain the original image to be dehazed; The transmission coefficient of the original image is calculated based on the local atmospheric light intensity before correction and the atmospheric light intensity at infinity. The K-means clustering algorithm is used to divide the transmission coefficients according to the depth of the original image to obtain the transmission coefficient threshold; The correction coefficient is calculated based on the transmission coefficient threshold; wherein the transmission coefficient is divided into far-field, mid-field and near-field according to the depth of the original image, and the transmission coefficient threshold t1 between the far-field and the mid-field and the transmission coefficient threshold t2 between the mid-field and the near-field are obtained; the correction coefficient containing different depth regions of the original image is obtained based on the transmission coefficient threshold t1 between the far-field and the mid-field, the transmission coefficient threshold t2 between the mid-field and the near-field and the transmission coefficient. Calculate the corrected local atmospheric light intensity based on the correction coefficient; and The dehazing restored image of the original image is obtained based on the corrected local atmospheric light intensity.

2. The polarization dehazing method for outdoor scenes with large depth of field as described in claim 1, characterized in that, The formula for calculating the transmission coefficient is: ； in, This refers to the local atmospheric light intensity before correction. It is the atmospheric light intensity at infinity.

3. The polarization defogging method for outdoor scenes with large depth of field as described in claim 1, characterized in that, The correction coefficient increases as the depth of the original image increases or the transmission coefficient decreases.

4. The polarization defogging method for outdoor scenes with large depth of field as described in claim 3, characterized in that, The formula for calculating the correction coefficient is: 。 5. The polarization dehazing method for outdoor scenes with large depth of field as described in claim 1, characterized in that, Provided that the minimum correction coefficient in the distant view is greater than the minimum correction coefficient in the mid-view, the correction coefficient in the distant view decreases as the depth of the original image increases or the transmission coefficient decreases.

6. The polarization dehazing method for outdoor scenes with large depth of field as described in claim 5, characterized in that, The formula for calculating the correction coefficient is: 。 7. The polarization defogging method for outdoor scenes with large depth of field as described in claim 4 or 6, characterized in that, The formula for calculating the corrected local atmospheric light intensity is as follows: ； in, It is the atmospheric light polarization angle. It is the degree of atmospheric light polarization. It is a Stokes parameter. This is the original image.

8. The polarization dehazing method for outdoor scenes with large depth of field as described in claim 7, characterized in that, The formula for calculating the dehazed and restored image is: ； in, It refers to the pixel position in the original image.

Images