A method for obtaining background light of underwater visual image in artificial illumination environment
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING JIAOTONG UNIV
- Filing Date
- 2023-04-06
- Publication Date
- 2026-06-26
Smart Images
Figure CN116503722B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of underwater image processing technology, and in particular to a method for obtaining background light when an underwater robot in a deep-sea area relies solely on artificial light for target detection. Background Technology
[0002] Underwater robots require clear information about the underwater environment and targets when conducting underwater target detection and operations. Visual imaging is currently the primary means of acquiring information for close-range underwater operations. In the underwater environment, underwater visual images suffer severe degradation due to factors such as the scattering and refraction of light by the water medium and impurities, resulting in problems like strong background noise, blurry images, and color casts. Therefore, it is necessary to restore severely degraded underwater visual images to improve image clarity and lay the foundation for subsequent image processing such as target extraction. Background light directly affects the tone and detail of the restored image, making its determination crucial.
[0003] Currently, methods for solving background light parameters mainly include dark channel prior methods and deep learning-based methods. Dark channel prior methods are characterized by not relying on data outside the image and having excellent dehazing effects. However, the acquired background light is easily affected by self-luminous organisms or objects with extremely smooth surfaces, misidentifying foreground objects as the source of background light, resulting in severe color casts. Deep learning-based methods have powerful feature extraction capabilities and high accuracy, but their robustness is relatively low. Furthermore, current datasets of background light in underwater images are limited, making learning difficult. Additionally, there is limited research on acquiring background light in underwater images under artificial lighting conditions. Therefore, addressing these issues, researching the acquisition of underwater background light under artificial lighting conditions is a problem that needs to be solved by those skilled in the art for underwater image restoration. Summary of the Invention
[0004] To address the aforementioned shortcomings of existing technologies, the purpose of this invention is to provide a method for acquiring background light in underwater visual images under artificial lighting conditions. This method solves the problem that in existing technologies, underwater robots acquiring target images under artificial lighting conditions suffer from significant background light errors, which in turn affect the subsequent image restoration effect.
[0005] To achieve the above objectives, the present invention adopts the following technical solution:
[0006] A method for acquiring background light in underwater visual images under artificial lighting conditions includes the following steps:
[0007] (1) Use an underwater robot to acquire the original underwater image, and construct the first depth map through the bright channel map of the original image;
[0008] (2) Construct a second depth map using the red channel map of the original image;
[0009] (3) Based on the main color tone of the image, and combining the first depth map obtained in step (1) and the second depth map obtained in step (2), construct an initial depth map;
[0010] (4) Construct a third depth map using the blur map of the original image;
[0011] (5) Based on the initial depth map obtained in step (3), divide the depth map into two regions: foreground and background. Combined with the third depth map obtained in step (4), optimize the background region of the initial depth map to construct the final depth map.
[0012] (6) Determine whether there are other objects in the background area obtained through step (5) that interfere with the solution of the background light. If there are, use the Otsu algorithm to remove other objects in the background area; if not, proceed directly to step (7).
[0013] (7) Select the average value of the brightest 0.1% of the pixel values in the first 0.1% of the depth map after segmentation in step (6) as the background light.
[0014] Specifically, step (1) is as follows:
[0015] Bright channel is a priori method that indicates that in most local patches of a haze-free image, at least one color channel contains some pixels with very high intensity. Since the bright channel varies significantly between the foreground and background under artificial lighting, using bright channel can meet the requirements of subsequent foreground-background separation. Furthermore, a depth map based on bright channel is more consistent with the requirements of an underwater environment with sufficient artificial lighting. Therefore, the first depth map d is constructed using bright channel. L Right now:
[0016] d L (x)=1-F s (J Light (1)
[0017]
[0018] Among them, F s (X) represents the grayscale stretching function, J Light Indicates an open channel, that is:
[0019]
[0020] Among them, J c For the original underwater image, Ω(x) is typically set to a 7×7 pixel neighborhood around pixel x.
[0021] Step (2) specifically involves:
[0022] In underwater environments, red light is more sensitive to depth changes, thus accurately reflecting the depth of underwater images. However, the first depth map obtained in step (1) may misidentify blue areas in the background as foreground when the image has a blue color cast. Therefore, to compensate for this defect and prepare for the fusion of the two channels in the subsequent step (3), a second depth map d is constructed using the red channel map. R :
[0023]
[0024] Where R is the red channel, F s It is the grayscale stretching function, I R For the red channel of the image, it is typically set to a 7×7 pixel neighborhood around pixel x.
[0025] Step (3) specifically involves: combining the first depth map obtained in step (1) with the second depth map obtained in step (2) to construct an initial depth map d. f :
[0026] d f (x)=θd L (x)+(1-θ)d R (x) (5)
[0027] θ=S(abs(C(1)-C(0)),20) (6)
[0028] Where θ represents the color bias of the image, C is the main color tone of the image obtained by the median segmentation method (C(0), C(1) and C(2) represent the red, green and blue color channels of the main color tone respectively), S is the Sigmoid function, and abs is the absolute value.
[0029] When abs(C(1)-C(0))<<20 and θ=0, there is a blue color bias in the underwater environment. L As a depth indicator, it might mistakenly identify the blue background area as the foreground. Because the underwater red channel value is low, and at this point it's close to the green channel value, while differing significantly from the blue channel value, it can accurately distinguish the blue background. Therefore, d is used here. R As a depth map.
[0030] When abs(C(1)-C(0))>>20 and θ=1, the overall color tone of the underwater environment has no significant color bias or has a green color bias. At this time, d R This can lead to misidentifying blue foreground objects as background. Since the bright channel can still accurately distinguish foreground objects even with less blue light, d0 is used in this case. L(x) is used as a depth map.
[0031] Between these two extremes, the initial depth map is derived from a weighted combination of the two methods.
[0032] The main purpose of step (3) is to obtain a better initial depth map so that the foreground and background of the underwater image can be better distinguished by the depth map, in preparation for step (5).
[0033] Step (4) specifically refers to:
[0034] Because the artificial lighting intensity in the background area of underwater images is lower than that in the foreground, and considering the relatively flat and blurred characteristics of the background area, a third depth map is constructed using a blur map. First, the coarse blur map P is solved. r :
[0035]
[0036] Where G is the input image filtered by the Gaussian filter, r i =2 i n+1, where n is typically 4, I g For a grayscale image, Ω(x) is typically set to a 7×7 pixel neighborhood around pixel x. Then, P is filled using morphological reconstruction. r The holes in the P are used to refine the P. r And guided filtering is used for smoothing to generate an accurate blur map P. blr And construct a third depth map d B :
[0037] P blr (x)=F g {C r [P r (x)]} (8)
[0038] d B (x)=1-F s (P blr (x)) (9)
[0039] Among them, C r It is a void-filling morphological reconstruction operator, F g It is a guided filter function.
[0040] Step (5) specifically refers to:
[0041] The initial depth map obtained in step (3) is used as the depth map of the foreground region of the underwater image, and is merged with the third depth map obtained in step (4) as the depth map of the background region to construct the final depth map d. final :
[0042]
[0043] in, The threshold represents the initial depth map d. f The average of the top 90%.
[0044] Step (6) specifically involves:
[0045] Background light is represented by the brightest point of light in the water body (non-object) at infinity. If the background area contains overly bright objects or non-water parts, it will interfere with the determination of background light. Therefore, in order to eliminate the influence of other objects in the background area on the background light, firstly, the gradient map of the background area is used to determine whether there are other objects in the background area obtained through step (5). If there are, the Otsu algorithm is used to remove other objects in the background area. If there are no objects, no processing is done, and the process proceeds directly to step (7) to prevent the loss of background information in the underwater image.
[0046] The specific steps (7) are as follows:
[0047] To meet the requirements of both the furthest and brightest background light, the first 0.1% of the segmented final depth map is designated as the furthest region. Based on this, the average value of the brightest pixels within the first 0.1% of this region is selected as the final background light value B. ∞c :
[0048]
[0049] in, This represents the first 0.1% of pixel values in the final depth map. This represents the average value of the first 0.1% of pixels in the brightness channel of the original image. That is, (7) the average value of the brightest 0.1% of pixels in the first 0.1% of the depth map after segmentation in step (6) is selected as the background light.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] This invention addresses the unique working environment of underwater robots by constructing a depth map from raw underwater images acquired by a charge-coupled device (CCD) camera. Then, it uses the maximum inter-class variance (MOV) method to segment objects in the background region of the raw image that affect background light acquisition. Finally, it selects the top 0.1% of the segmented depth map as the farthest region, and based on this, the average value of the brightest pixels in the top 0.1% of this region is selected as the background light value. This invention enables underwater robots to quickly calculate their background light after acquiring images using a CCCD camera, and then send the background light data to the underwater robot's main controller, laying the foundation for subsequent underwater image processing.
[0052] This invention, starting from the actual working environment of underwater robots, focuses on the impact of artificial lighting on background light and designs a background light acquisition method suitable for artificial lighting environments. This provides more accurate background light values for subsequent image restoration, resulting in underwater images with natural colors and rich details.
[0053] Because underwater robots acquire deep-sea images under artificial lighting conditions, unlike natural light, the brightness of objects under artificial lighting is higher than that of the water at the same depth. Most existing background light acquisition methods rely on dark channel methods. However, the dark channel method can inflate background light values when processing bright objects in the image, making subsequent image restoration difficult. This invention addresses the problem of inaccurate background light calculation in underwater images under artificial lighting by providing a depth map fusion method for background light acquisition. The background light acquired using this method ensures that the restored underwater image achieves natural colors, high saturation, and rich detail. Attached Figure Description
[0054] Figure 1 This is a schematic diagram of the background light acquisition method when using artificial light for target detection;
[0055] Figure 2 It is an underwater optical imaging model;
[0056] Figure 3 This is a flowchart of the method for acquiring background light in underwater visual images under artificial lighting conditions according to the present invention;
[0057] Figure 4 These are the original underwater degradation images used in the experiments of this invention;
[0058] Figure 5 (a) corresponds to the use of the method of the present invention for Figure 4 The result of the processing;
[0059] Figure 5 (b) corresponds to the use of the UDCP algorithm proposed by Drews et al. Figure 4 The result of the processing;
[0060] Figure 5 (c) corresponds to the use of the red channel prior algorithm proposed by Galdran et al. Figure 4 The result of the processing;
[0061] Figure 5 (d) corresponds to the use of the MIP algorithm proposed by Bianco et al. Figure 4 The result of the processing;
[0062] Figure 5 (e) corresponds to the algorithm proposed by Peng et al. Figure 4 The result of the processing;
[0063] Figure 5 (f) corresponds to the use of the WCID algorithm proposed by Chiang et al. Figure 4 The result of the processing. Detailed Implementation
[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0065] It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the figures, or the orientation or positional relationship commonly used when the product is in use. They are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance. In addition, the terms "horizontal," "vertical," etc., do not indicate that the component is required to be absolutely horizontal or suspended, but can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted. In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0066] See Figure 1 and Figure 2 This paper discusses a method for obtaining background light when underwater robots in deep-sea areas rely solely on artificial light for target detection. When underwater robots acquire deep-sea images, they are under artificial lighting conditions, which differ from natural light conditions. At the same depth of field, the brightness of targets under artificial lighting is higher than that of the water. Most existing background light calculation methods rely on the dark channel method, but the dark channel method can cause the background light value to be too high when processing bright objects in the image, making subsequent image restoration difficult.
[0067] like Figure 3 As shown, the present invention provides a method for acquiring background light in underwater visual images under artificial lighting conditions, comprising the following steps:
[0068] Step (1): Construct the first depth map using the bright channel map of the original image.
[0069] Bright channel is a priori method that indicates that in most local patches of a hazy image, at least one color channel contains some pixels with very high intensity. Since the bright channel of the foreground and background of an image varies significantly under artificial lighting, using bright channel can meet the requirements of subsequent foreground-background separation. Furthermore, the depth map based on bright channel meets the specific requirements of well-lit artificial environments. Therefore, the first depth map d is constructed using bright channel. L Right now:
[0070] d L (x)=1-F s (J Light (1)
[0071]
[0072] Among them, F s J represents the grayscale stretching function. Light Indicates an open channel, that is:
[0073]
[0074] Among them, J c (y) is the original underwater image, and Ω(x) is generally set to a 7×7 pixel neighborhood around pixel x.
[0075] Step (2): Construct a second depth map using the red channel map of the original image.
[0076] In underwater environments, red light is more sensitive to depth changes, thus accurately reflecting the depth of underwater images. However, the first depth map obtained in step (1) may misidentify blue areas in the background as foreground when the image has a blue color cast. Therefore, to compensate for this defect and prepare for the fusion of the two channels in the subsequent step (3), a second depth map (Figure d) is constructed using the red channel map. R :
[0077] d R (x)=1-F s (R) (4)
[0078] Where R is the red channel, F s It is a grayscale stretching function.
[0079] Step (3): Based on the main color tone of the image, combine the first depth map obtained in step (1) and the second depth map obtained in step (2) to construct an initial depth map.
[0080] The main purpose of step (3) is to obtain a better initial depth map, so that the foreground and background of the underwater image can be better distinguished by the depth map, thus preparing for step (5). Combining the first depth map obtained in step (1) and the second depth map obtained in step (2), an initial depth map d is constructed. f :
[0081] d f (x)=θd L (x)+(1-θ)d R (x) (5)
[0082] θ=S(abs(C(1)-C(0)),20) (6)
[0083] Where θ represents the color bias of the image, C is the main color tone of the image obtained by the median segmentation method (C(0), C(1) and C(2) represent the red, green and blue color channels of the main color tone respectively), S is the Sigmoid function, and abs is the absolute value.
[0084] When abs(C(1)-C(0))<<20 and θ=0, there is a blue color bias in the underwater environment. L As a depth map, it might mistakenly identify the blue background area as the foreground. Because the underwater red channel value is low and close to the green channel value, while differing significantly from the blue channel value, it can accurately distinguish the blue background. Therefore, di is used here. R As a depth map.
[0085] When abs(C(1)-C(0))>>20 and θ=1, the overall color tone of the underwater environment has no significant color bias or has a green color bias. At this time, dR This can lead to misidentifying blue foreground objects as background. Since the bright channel can still accurately distinguish foreground objects even with less blue light, d0 is used in this case. L (x) is used as a depth map.
[0086] Between these two extremes, the initial depth map is derived from a weighted combination of the two methods.
[0087] Step (4): Construct a third depth map using the blur map of the original image.
[0088] Since the artificial lighting intensity in the background area of underwater images is lower than that in the foreground, and considering the relatively flat and blurry nature of the background area, a third depth map is constructed using a blur map.
[0089] First, solve the roughness ambiguity map P. r :
[0090]
[0091] Where G is the input image filtered by the Gaussian filter, r i =2 i n+1, where n is typically 4, I g For grayscale images, Ω(x) is typically set to a 7×7 pixel neighborhood around pixel x.
[0092] Then, P was filled using morphological reconstruction. r The holes in the P are used to refine the P. r And guided filtering is used for smoothing to generate an accurate blur map P. blr And construct a third depth map d B :
[0093] P blr (x)=F g {C r [P r (x)]} (8)
[0094] d B (x)=1-F s (P blr (x)) (9)
[0095] Among them, C r It is a void-filling morphological reconstruction operator, F g It is a guided filter function.
[0096] Step (5): Based on the initial depth map obtained in step (3), divide the depth map into two regions: foreground and background. Combined with the third depth map obtained in step (4), optimize the background region of the initial depth map to construct the final depth map.
[0097] The initial depth map obtained in step (3) is used as the depth map of the foreground region of the underwater image, and is merged with the third depth map obtained in step (4) as the depth map of the background region to construct the final depth map d. final :
[0098]
[0099] in, The threshold represents the initial depth map d. f The average of the top 90%.
[0100] Step (6): Determine whether there are other objects interfering with the background light in the background area obtained through step (5). If there are, use the Otsu algorithm to remove other objects in the background area; if not, proceed directly to step (7).
[0101] Background light is represented by the brightest point of light in the water body (non-object) at infinity. If the background area contains overly bright objects or non-water parts, it will interfere with the determination of background light. Therefore, in order to eliminate the influence of other objects in the background area on the background light, firstly, the gradient map of the background area is used to determine whether there are other objects in the background area obtained through step (5). If they exist, the Otsu algorithm is used to remove other objects in the background area; if they do not exist, no processing is done, and the process proceeds directly to step (7) to prevent the loss of background information in the underwater image.
[0102] Step (7): Select the average value of the brightest 0.1% of the pixel values in the first 0.1% of the depth map after segmentation in step (6) as the background light.
[0103] To meet the requirements of both the furthest and brightest background light, the first 0.1% of the segmented final depth map is designated as the furthest region. Based on this, the average value of the brightest pixels within the first 0.1% of this region is selected as the final background light value B. ∞c :
[0104]
[0105] in, This represents the first 0.1% of pixel values in the final depth map. This represents the average value of the first 0.1% of pixels in the luminance channel of the original image.
[0106] Application Cases
[0107] To verify the effectiveness of the underwater image restoration method based on multiple frames under artificial lighting described in this invention, the method of this invention was experimentally compared with the UDCP algorithm of Drews et al., the red channel prior algorithm of Galdran et al., the MIP algorithm of Bianco et al., the underwater image restoration algorithm based on image blur and light absorption of Peng et al., and the WCID algorithm of Chiang et al.
[0108] See Figure 4 The original underwater degradation images used in the experiments of this invention serve as a reference for comparison of the restoration effect.
[0109] See Figure 5 This is a comparison of the underwater image restoration results. Among them:
[0110] Figure 5 (a) Corresponding to the application of the method of the present invention to Figure 3 The result of the processing;
[0111] Figure 5 (b) Correspondingly, the UDCP algorithm proposed by Drews et al. is used for... Figure 3 The result of the processing;
[0112] Figure 5 (c) Correspondingly, the red channel prior algorithm proposed by Galdran et al. is used for... Figure 3 The result of the processing;
[0113] Figure 5 (d) Corresponding to the MIP algorithm proposed by Bianco et al. Figure 3 The result of the processing;
[0114] Figure 5 (e) Correspondingly, the algorithm proposed by Peng et al. is used to... Figure 3 The result of the processing;
[0115] Figure 5 (f) Corresponding to the WCID algorithm proposed by Chiang et al. Figure 3 The result of the processing.
[0116] Combined with appendix Figure 5 As can be seen from the comparison of experimental results in (a), the image processed by the method of the present invention has significantly improved clarity, the details of the target object are more obvious, and the color of the restored image is more natural compared with other methods.
[0117] To more objectively evaluate the image quality of the algorithm's experimental results, this invention selects two commonly used underwater image quality evaluation metrics, UCIQE (Underwater Color Image Quality Evaluation) and UIQM (Underwater Image Quality Measurement), as the quality evaluation metrics for underwater image restoration. UCIQE reflects the linear quantitative evaluation result of the relationship between color cast, blur, and contrast after underwater image restoration; a higher value indicates a better image processing effect. UIQM mainly uses a linear combination of the color measurement metric (UICM), sharpness measurement metric (UISM), and contrast measurement metric (UIConM) as the evaluation basis; a higher value indicates better color balance, sharpness, and contrast in the processed image.
[0118] Depend on Figure 5 The original color images (b) to (f) clearly show that the image processed by Drews et al.'s UDCP algorithm exhibits a noticeable halo on the target object, with excessive brightness enhancement, and the algorithm's error in red channel compensation causes the image to be reddish; the image processed by Galdran et al.'s red channel prior algorithm has an overall reddish background, and the blue sphere in the target object almost blends into the background after restoration; the image processed by Bianco et al.'s MIP algorithm has severe color distortion; Peng et al.'s underwater image restoration algorithm based on image blur and light absorption incorrectly enhances the dark background of the original underwater image to blue, and makes the overall color of the processed image darker; although the target object color is better in the image processed by Chiang et al.'s WCID algorithm, the background is grayish and the target object details are lost.
[0119] In summary, the quantitative analysis results of image restoration by different algorithms are shown in Table 1. Comparative analysis of the evaluation indicators of each method shows that the method of this invention is superior to other underwater image restoration algorithms in both the UCIQE and UIQM objective evaluation indicators. Therefore, for turbid water environments with artificial light sources, this invention can achieve underwater image restoration better than other underwater image restoration algorithms.
[0120] Table 1 Comparison of objective evaluation metrics among different algorithms
[0121]
[0122] In summary, this invention addresses the problem of inaccurate background light calculation when restoring underwater images under artificial lighting. It provides a method for obtaining background light through depth map fusion. The background light obtained by this method ensures that the restored underwater image achieves natural colors, high saturation, and rich details.
[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit the technical solutions. Those skilled in the art should understand that any modifications or equivalent substitutions to the technical solutions of the present invention without departing from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.
Claims
1. A method for acquiring background light in underwater visual images under artificial lighting conditions, characterized in that, Starting from the working environment of the underwater robot, a depth map is constructed using information from the original image obtained by the charge-coupled camera. Then, the Otsu's method is used to segment the objects in the background region of the original image that affect the acquisition of background light. Finally, the top 0.1% of the segmented depth map is selected as the farthest region, and the average value of the brightest pixels in the top 0.1% of this region is selected as the value of the background light. Specifically, it includes the following steps: Step 1: Use an underwater robot to acquire raw underwater images, and construct the first depth map using the bright channel map of the raw images. ; Step 2: Construct a second depth map using the red channel information from the original image. ; Step 3: Based on the intensity difference between the green and red channels of the image, and combining the first depth map obtained in Step 1 with the second depth map obtained in Step 2, construct an initial depth map. ; Step 4: Construct a third depth map using the blur information in the original image. ; Step 5: Combine the above and Construct the final depth map ; Step 6: Determine whether there are other objects in the background area obtained in Step 5 that interfere with the solution of the background light; if so, use the Otsu algorithm to remove other objects in the background area; if not, proceed directly to step (7). Step 7: Select the top 0.1% of the final depth map after segmentation in Step 6 as the farthest region, and based on this, select the average value of the brightest pixels in the top 0.1% of this region as the final background light value. ; The first depth map constructed in step 1 for: (1)； (2)； in, This represents the grayscale stretching function. Indicates an open channel, that is: (3)； in, For the original image, Set to a 7×7 pixel neighborhood around pixel x.
2. The method for acquiring background light in underwater visual images under artificial lighting conditions according to claim 1, characterized in that, in, In step 2, the red channel map is used to construct a second depth map. for: (4)； Where R is the red channel value, For the red channel of the image, Set to a 7×7 pixel neighborhood around pixel x.
3. The method for acquiring background light in underwater visual images under artificial lighting conditions according to claim 1, wherein step 3 is to construct an initial depth map by combining the first depth map from step 1 and the second depth map from step 2. : (5)； (6)； in, The color bias value represents the image, C is the main color tone of the image obtained by the median segmentation method, C(0), C(1) and C(2) represent the red, green and blue color channels of the main color tone respectively, S is the Sigmoid function, and abs is the absolute value; when , At that time, the underwater environment exhibited a blue color bias, which will As a depth-based detection method, it might incorrectly identify the blue background area as the foreground. In this case, it would use... As a depth map; when , At this time, the overall color tone of the underwater environment has no significant color deviation or has a greenish tint. It will misjudge blue foreground objects as background; in this case, use As a depth map; Between these two extremes, the initial depth map is derived from a weighted combination of the two methods.
4. The method for acquiring background light in underwater visual images under artificial lighting conditions according to claim 1, characterized in that, Step 4 uses a fuzzy map to construct a third depth map; First, solve the roughness fuzziness map. : (7)； in, The input image is filtered by a Gaussian filter. n takes the value 4. It is a grayscale image. Set to a 7×7 pixel neighborhood around pixel x; Then, filling was performed using morphological reconstruction. The holes in the middle, thus refining Guided filtering is then used for smoothing to generate accurate blur maps. And construct a third depth map : (8)； (9)； in, It is a void-filling morphological reconstruction operator. It is a guided filter function.
5. The method for acquiring background light in underwater visual images under artificial lighting conditions according to claim 1, characterized in that, Step 5 specifically involves: The initial depth map obtained in step 3 is used as the depth map of the foreground region of the underwater image, and it is merged with the third depth map obtained in step 4 as the depth map of the background region to construct the final depth map. : (10)； in, The threshold represents the initial depth map. The average of the top 90%.
6. The method for acquiring background light in underwater visual images under artificial lighting conditions according to claim 1, characterized in that, The final background light value for: (11)； in, This represents the first 0.1% of pixel values in the final depth map. This represents the average value of the first 0.1% of pixels in the brightness channel of the original image.