A method, storage medium, and device for adaptive underwater lighting in ROVs

Abstract

This invention relates to an underwater adaptive lighting method, storage medium, and device for ROVs. This method combines convolutional neural networks and generative adversarial networks to improve image quality and intelligently adjusts the current of the light source based on real-time acquired lighting data, thereby adjusting the light intensity to ensure optimal illumination under different water depths and water quality conditions. By intelligently adjusting the light source's operating mode, unnecessary energy consumption is reduced, thus improving the ROV's endurance during long-duration underwater missions. The underwater adaptive lighting method provided by this invention can significantly improve image clarity and visual effects under low-light conditions, and it possesses real-time response capabilities, ensuring rapid adjustment of lighting conditions in complex and variable underwater environments.

Description

Technical Field

[0001] This invention relates to underwater imaging technology, specifically to an ROV underwater adaptive lighting method, storage medium, and device. Background Technology

[0002] A remotely operated vehicle (ROV) is a device that can move underwater, has vision and perception systems, and uses robotic arms or other tools remotely to replace or assist humans in performing certain underwater tasks.

[0003] Light propagation in underwater environments has unique characteristics, primarily including light attenuation, scattering, and absorption. As light travels through water, its intensity rapidly decreases, especially since different wavelengths attenuate at different rates. Blue and green light have relatively strong penetrating power, while red light attenuates the fastest. Furthermore, suspended particles in the water cause light scattering, creating diffused light and reducing image contrast and clarity. Water also has a strong ability to absorb light, particularly long wavelengths such as red and ultraviolet light, which are absorbed by water molecules and suspended matter. This not only reduces the intensity of visible light but also affects the illumination effect of ROVs, making underwater photography and exploration more difficult.

[0004] Current underwater imaging technologies face challenges in dealing with complex lighting environments. Image quality is often affected by the attenuation, scattering, and absorption characteristics of underwater light. Common methods include physical model-based imaging techniques and data-driven imaging techniques. Physical model techniques, such as polarization imaging, structured light scanning, and light field imaging, can acquire high-dimensional information, but they face challenges in handling dynamic changes and complex nonlinear factors. Most existing underwater lighting systems use fixed light sources and cannot be dynamically adjusted according to environmental changes. Summary of the Invention

[0005] In view of this, the present invention provides an underwater adaptive lighting method, storage medium, and device for ROVs, thereby solving or at least mitigating one or more of the above-mentioned problems and other problems existing in the prior art.

[0006] To achieve the aforementioned objective, a first aspect of the present invention provides an underwater adaptive lighting method for ROVs, wherein the method includes the following steps:

[0007] Step S1: Input the initial underwater image;

[0008] Step S2: Use histogram analysis technology to detect the image quality. If the image quality detection result is unqualified, proceed to step S3. If the image quality detection result is qualified, proceed to step S4.

[0009] Step S3: Determine whether the light source is turned on. If the light source is not turned on, turn on the light source and proceed to step S1. If the light source is already turned on, proceed to step S4.

[0010] Step S4: Extract multiple features of the image based on a convolutional neural network;

[0011] Step S5: Based on the multiple features extracted from the image, perform 3D reconstruction, use a generative adversarial network to process the extracted multiple features, generate a 3D image, and the generative adversarial network improves the quality of the 3D image;

[0012] Step S6: Determine the quality of the three-dimensional image. If the quality of the three-dimensional image is unqualified, adjust the light intensity of the light source and proceed to step S1. If the quality of the three-dimensional image is qualified, complete the lighting adjustment and image processing for the current cycle and prepare to enter the next cycle.

[0013] In the method described above, step S1 may optionally include: acquiring an initial underwater image using the ROV's sensors and / or camera, wherein the light source is an LED light.

[0014] In the method described above, optionally, the step of detecting the image quality using histogram analysis in step S2 includes:

[0015] The input image is preprocessed, including denoising and normalization.

[0016] Calculate the grayscale histogram of the image and analyze the brightness distribution of the image;

[0017] The quality of the image is evaluated based on the uniformity and contrast of the grayscale histogram, and a target image quality score is set. Furthermore, a predetermined range is set, which defines the target image quality score. An acceptable fluctuation range is defined, and a score within the predetermined range indicates acceptable image quality. An image quality score is then calculated. If the image quality score Within the predetermined range, it is indicated that the image quality is acceptable; if the image quality score is... If the image is not within the predetermined range, it indicates that the image quality is substandard.

[0018] In the method described above, optionally, the convolutional neural network includes:

[0019] Input layer, which receives the image;

[0020] Convolutional layers extract features of different scales and levels from the image through convolution operations, including edges, textures, and higher-order features;

[0021] Pooling layer, which reduces the dimension of feature map and retains main features through max pooling or average pooling operations. The main features include low-order features and high-order features. The low-order features include edges and textures, and the high-order features include the shape and pose of the object.

[0022] The activation function includes the Rectified Linear Unit (ReLU), which is used to introduce nonlinear characteristics and enhance the expressive power of the model.

[0023] Optionally, in the method described above, the generative adversarial network includes:

[0024] A generator that generates a 3D image based on the multiple features extracted by the convolutional neural network;

[0025] A discriminator that evaluates the quality of the 3D image generated by the generator.

[0026] In the method described above, optionally, the loss function of the convolutional neural network includes a perturbation-resistant loss term, which is used to train the robustness of the convolutional neural network under noisy conditions, and the loss function is:

[0027]

[0028] in, It is the mean squared error loss, used to measure the difference in image pixel values:

[0029] ;

[0030] in, It is a structural similarity loss, used to measure the structural similarity of images:

[0031] ;

[0032] in, It is an adversarial loss used to enhance image details and contrast, implemented through a generative adversarial network:

[0033] ;

[0034] in, It is a perturbation resistance loss used to enhance the robustness of the model under different noise conditions:

[0035]

[0036] These are noise parameters;

[0037] in, These are loss items for specific situations, including turbidity and color deviation:

[0038]

[0039] in, and It is the weighting coefficient of the loss term.

[0040] In the method described above, optionally, the quality of the three-dimensional image is determined using an image quality evaluation equation, which is:

[0041] in, It is an image quality score;

[0042] in, It is an evaluation index for image contrast;

[0043] in, It is an evaluation index for edge fidelity;

[0044] in, It is a brightness consistency evaluation index;

[0045] in, It is a variable related to water turbidity;

[0046] in, It is a variable representing the deviation of light color;

[0047] in, and These are weighting coefficients, which are adjusted according to the actual situation.

[0048] In the method described above, optionally, the step of determining the quality of the three-dimensional image in step S6 includes:

[0049] Calculate real-time image quality score ,

[0050] If the real-time image quality score Less than the target image quality score This indicates that the current is insufficient. Increase the current intensity of the light source to increase the light intensity of the light source.

[0051] If the real-time image quality score Greater than the target image quality score This indicates that the current light intensity is too strong. The light intensity of the light source should be reduced appropriately to reduce the light intensity of the light source.

[0052] To achieve the foregoing objective, a second aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions or a computer program that, when executed by a processor, implement the method as described in any of the first aspects above.

[0053] To achieve the aforementioned objectives, a third aspect of the present invention provides a data processing apparatus, wherein the data processing apparatus comprises:

[0054] The memory is used to store computer-executable instructions or computer programs;

[0055] A processor, when executing computer-executable instructions or computer programs stored in the memory, implements the method as described in any of the preceding first aspects.

[0056] The ROV underwater adaptive lighting method provided by this invention combines convolutional neural networks and generative adversarial networks to improve image quality. It can intelligently adjust the current of the light source based on real-time acquired lighting data, thereby adjusting the light intensity to ensure optimal illumination under different water depths and water quality conditions. By intelligently adjusting the light source's operating mode, unnecessary energy consumption is reduced, thus improving the ROV's endurance during long-duration underwater missions. The ROV underwater adaptive lighting method provided by this invention can significantly improve image clarity and visual effects under low-light conditions, and it possesses real-time response capabilities, ensuring rapid adjustment of lighting conditions in complex and variable underwater environments.

[0057] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0058] The disclosure of this invention will become more apparent from the accompanying drawings. It should be understood that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings:

[0059] Figure 1 This is a flowchart illustrating an embodiment of the ROV underwater adaptive lighting method of the present invention;

[0060] Figure 2 Images that have not been processed by the ROV underwater adaptive lighting method of this invention;

[0061] Figure 3 For processing by the ROV underwater adaptive lighting method of the present invention Figure 2 The resulting effect diagram. Detailed Implementation

[0062] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While one embodiment of the invention is shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0063] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment". Definitions of other terms will be given in the description below.

[0064] Figure 1 This is a flowchart illustrating an embodiment of the ROV underwater adaptive lighting method of the present invention.

[0065] like Figure 1 As shown, the ROV underwater adaptive lighting method may include the following steps S1 to S6.

[0066] In step S1, the initial underwater image is input.

[0067] In some embodiments, the initial underwater image is a raw image captured in the underwater environment by the ROV's sensors and / or cameras. It is understood that the unique characteristics of the underwater environment, such as water absorption and scattering, can lead to a decrease in image quality; therefore, the ROV needs to be equipped with lighting equipment to optimize imaging. In optional embodiments, the type of lighting equipment is not limited and can be an LED light source. LED lights can provide high-brightness illumination, and because LED lights do not have fragile filaments, they are more resistant to vibrations and shocks during underwater operations.

[0068] In step S2, histogram analysis is used to detect image quality. If the image quality detection result is unqualified, proceed to step S3; if the image quality detection result is qualified, proceed to step S4.

[0069] It is understood that histogram analysis is an image processing and analysis method that evaluates image quality by statistically analyzing the frequency of occurrence of each pixel value in an image. A pixel is the basic unit of an image, and each pixel has one or more pixel values ​​corresponding to it. For example, in a color image, a pixel can have three pixel values: RGB (red, green, and blue). In a grayscale image, a pixel can have one pixel value representing the brightness of the pixel.

[0070] In an optional embodiment, the step of detecting image quality using histogram analysis technology includes:

[0071] The input image is preprocessed, including denoising and normalization.

[0072] Calculate the grayscale histogram of the image and analyze the brightness distribution of the image;

[0073] The image quality is evaluated based on the uniformity and contrast of the grayscale histogram, and a target image quality score is set. Furthermore, a predetermined range is set, which limits the target image quality score. An acceptable fluctuation range is defined, and a score within the predetermined range indicates acceptable image quality. An image quality score is then calculated. If the image quality score If the image quality is within the predetermined range, it indicates that the image quality is acceptable. If the image is outside the predetermined range, it indicates that the image quality is substandard.

[0074] Denoising is the process of reducing or eliminating noise in an image. Noise in an image refers to random variations unrelated to the original image that occur during image acquisition, transmission, or processing. It can manifest as random spots, stripes, or other irregular patterns in the image. Denoising can improve the resolution and quality of an image. In optional embodiments, the denoising method is not limited; filter-based methods, model-based methods, and other denoising methods are all considered to fall within the scope of protection of this invention.

[0075] Specifically, normalization restricts image data to a specific range, making image processing methods more stable and efficient. In optional embodiments, the normalization method is not limited, and normalization methods such as min-max normalization and nonlinear normalization are all considered to fall within the protection scope of this invention.

[0076] Specifically, a grayscale histogram is a graph used in image processing to represent the number of pixels at each grayscale level in an image. A grayscale level refers to the number of different brightness levels that each pixel in an image can represent. Each grayscale level corresponds to a specific brightness value; the smaller the brightness value, the lower the brightness and the closer the color is to black; the larger the brightness value, the higher the brightness and the closer the color is to white. A grayscale histogram can intuitively show the proportion of each grayscale level in an image. By calculating the grayscale histogram of an image, we can gain a direct understanding of the image's contrast, brightness, and grayscale distribution.

[0077] In step S3, it is determined whether the light source is turned on. If the light source is not turned on, the light source is turned on and step S1 is performed. If the light source is already turned on, step S4 is performed.

[0078] In optional embodiments, a light source is used to compensate for insufficient natural light in the underwater environment, enabling the ROV to acquire clearer images. The type of light source is not limited in different embodiments. In this embodiment, an LED light source can be used, which has the advantages of being efficient, durable, and energy-efficient, and is adjustable, providing suitable illumination in different underwater environments.

[0079] In step S4, multiple features of the image are extracted based on a convolutional neural network;

[0080] In an optional embodiment, the convolutional neural network includes:

[0081] Input layer, the input layer receives the image;

[0082] Convolutional layers extract features at different scales and levels from images through convolution operations, including edges, textures, and higher-order features.

[0083] Pooling layers reduce the dimensionality of feature maps and retain key features through max pooling or average pooling operations. Key features include low-order features and high-order features. Low-order features include edges and textures, while high-order features include the shape and pose of objects.

[0084] Activation functions, including the Rectified Linear Unit (ReLU), are used to introduce nonlinear characteristics and enhance the expressive power of the model.

[0085] As we understand it, a convolutional neural network (CNN) is a deep learning model where convolutional layers extract image features through convolution operations. The convolution operation involves sliding a learnable convolutional kernel (or filter) across the image, calculating the dot product with the local image region at each location to obtain a new feature map. The convolutional kernel is a small matrix, and its size determines the size of the image region covered by each convolution operation.

[0086] Specifically, low-order features of an image refer to the original or basic features extracted directly from the image data, including texture and edges. Low-order features of an image reflect the basic attributes and local structure of the image. High-order features of an image are features that can represent more complex semantics and concepts in the image, obtained through calculation and analysis based on low-order features, including the shape and posture of objects.

[0087] In an optional embodiment, the pooling layer uses pooling operations to replace local regions of the input feature map of the convolutional layer. Pooling operations include max pooling and average pooling. Max pooling retains the maximum value within a region, capturing the most salient features within a local region of the feature map. Average pooling calculates the average of all values ​​within a local region, smoothing the feature map and capturing the average features within the local region. The presence of pooling layers in a convolutional neural network helps reduce computational complexity and prevent overfitting. It can be understood that in this embodiment, the convolutional neural network can extract features from low-quality underwater images and help identify and preserve important image details.

[0088] In different embodiments, the type and number of activation functions are not limited. Activation functions that can introduce nonlinear characteristics into convolutional neural networks, such as the Rectified Luminaire (ReLU), hyperbolic tangent activation function, or exponential linear unit activation function, all fall within the scope of this invention. In convolutional neural networks, activation functions enable the model to perform nonlinear transformations, improving the network's expressive power and its ability to model complex functions.

[0089] In step S5, based on the multiple features extracted from the image, three-dimensional reconstruction is performed. The extracted multiple features are processed using a generative adversarial network to generate a three-dimensional image, and the generative adversarial network calculates the quality of the three-dimensional image.

[0090] It's important to understand that a 3D image is an image that represents an object or scene in three-dimensional space. A 3D image can display the three-dimensional shape, size, and spatial location of an object. Generative adversarial networks (GANs) can create 3D images based on multi-dimensional features in 2D image data.

[0091] In an optional embodiment, the generative adversarial network includes:

[0092] The generator generates 3D images based on multiple features extracted by a convolutional neural network.

[0093] The discriminator evaluates the quality of the 3D image generated by the generator.

[0094] It is understood that in generative adversarial networks (GANs), the generator learns to generate high-quality images through continuous improvement, while the discriminator helps the generator improve its generation capabilities by distinguishing between real and generated images, and ensures that the generated images are visually realistic and of high quality by evaluating their quality. In this embodiment, the GAN can be used to denoise and enhance the contrast and brightness of images, thereby significantly improving image clarity and visual effects under low-light conditions.

[0095] In step S6, the quality of the three-dimensional image is judged. If the quality of the three-dimensional image is unqualified, the light intensity of the light source is adjusted, and step S1 is performed. If the quality of the three-dimensional image is qualified, the lighting adjustment and image processing of the current cycle are completed, and the process is ready to enter the next cycle.

[0096] In an optional embodiment, the loss function of the convolutional neural network includes a perturbation-resistant loss term, which is used to train the robustness of the convolutional neural network under noisy conditions. The loss function is:

[0097]

[0098] in, It is the mean squared error loss, used to measure the difference in image pixel values:

[0099] ;

[0100] in, It is a structural similarity loss, used to measure the structural similarity of images:

[0101] ;

[0102] in, It is an adversarial loss used to enhance image details and contrast, implemented through a generative adversarial network:

[0103] ;

[0104] in, It is a perturbation resistance loss used to enhance the robustness of the model under different noise conditions:

[0105]

[0106] These are noise parameters;

[0107] in, These are loss items for specific situations, including turbidity and color deviation:

[0108]

[0109] in, and It is the weighting coefficient of the loss term.

[0110] Specifically, the robustness term refers to techniques used to prevent model overfitting. Overfitting refers to a model that performs well on training data but has poor generalization ability to new, unseen data. The robustness term helps reduce the model's sensitivity to noise and outliers in the training data. By adding a robustness term to the loss function, it helps train the convolutional neural network to be robust under noisy conditions, ensuring that the model works stably in the face of various environmental changes, which is especially important in situations where underwater lighting and environmental conditions may change rapidly.

[0111] In an optional embodiment, the ROV underwater adaptive lighting method of the present invention uses an image quality evaluation equation to determine the quality of a three-dimensional image. The image quality evaluation equation is used to calculate the quality of the image, and the image quality evaluation equation is as follows:

[0112] in, It is an image quality score;

[0113] in, It is an evaluation index for image contrast;

[0114] in, It is an evaluation index for edge fidelity;

[0115] in, It is a brightness consistency evaluation index;

[0116] in, It is a variable related to water turbidity;

[0117] in, It is a variable representing the deviation of light color;

[0118] in, and These are weighting coefficients, which are adjusted according to the actual situation.

[0119] In an optional embodiment, It is the contrast weighting coefficient, which controls the proportion of contrast in the score. The recommended range is 0.5 ≤ ≤1.5; This is the edge fidelity weighting coefficient, which controls the proportion of edge fidelity in the score. The recommended range for β is 0.5 ≤ ≤1.5; This is the brightness consistency weighting coefficient, which controls the proportion of brightness consistency in the score. The recommended range is 0.5 ≤ ≤2.0; It is the weighting coefficient for the water turbidity variable. The recommended range is 0.5 ≤ ≤1.5; It is a variable related to the color deviation of light. The recommended range is 0.5 ≤ ≤1.5.

[0120] In some embodiments, features extracted using a convolutional neural network are combined with feature quality evaluation metrics to calculate the overall image quality score. These quality evaluation metrics include contrast, edge fidelity, and brightness consistency. Contrast refers to the degree of difference between bright and dark areas in an image, which is related to the image's visual distinguishability; edge fidelity refers to the clarity and accuracy of edge information in an image, which is related to the image's recognition rate and analysis accuracy; brightness consistency refers to the uniformity of brightness in different areas of an image, affecting the image's visual effect and readability.

[0121] By using the aforementioned loss function and image quality evaluation equation, the quality of underwater images can be effectively improved, ensuring the acquisition of high-quality images under different lighting and special conditions, thereby enhancing the working efficiency and data acquisition capabilities of ROVs.

[0122] In an optional embodiment, step S6 of the present invention, which involves determining the quality of the three-dimensional image, includes:

[0123] Calculate real-time image quality score ,

[0124] If real-time image quality score Smaller than the target image quality score This indicates that the current is insufficient. Increase the current intensity of the light source to increase the light intensity of the light source.

[0125] If real-time image quality score Greater than the target image quality score This indicates that the current light intensity is too strong. The light intensity of the light source should be reduced appropriately to reduce the light intensity of the light source.

[0126] The method provided by this invention can intelligently adjust the current of the LED light source based on real-time collected illumination data, thereby adjusting the intensity of the light source to ensure optimal lighting effects under different water depths and water quality conditions. Furthermore, this intelligent light source adjustment mode can reduce unnecessary energy consumption.

[0127] Figure 2 Images that have not been processed by the ROV underwater adaptive lighting method of this invention.

[0128] Figure 3 For processing by the ROV underwater adaptive method of the present invention Figure 2 The resulting effect diagram.

[0129] exist Figure 3In the ROV underwater adaptive method processing of the present invention, the image evaluation quality equation is as follows:

[0130]

[0131] in, It is an image quality score;

[0132] in, It is an evaluation index for image contrast;

[0133] in, It is an evaluation index for edge fidelity;

[0134] in, It is a brightness consistency evaluation index;

[0135] in, It is a variable related to water turbidity;

[0136] in, It is a variable representing the deviation of light color;

[0137] in, and It is the weighting coefficient.

[0138] exist Figure 3 In the illustrated embodiment, weighting coefficients are set for the image evaluation quality equation, wherein, =1.0, =1.0, =1.5, =0.5, =0.5. Example: Figure 3 Image contrast The edge fidelity of the image is 0.8. The brightness uniformity of the image is 0.7. The turbidity level is 0.9. and color deviation of light The default value is 0. Calculated according to the image quality evaluation equation. Figure 3 Quality rating:

[0139] Q=1.0×0.8+1.0×0.7+1.5×0.9+0.5×0+0.5×0;

[0140] Therefore, Q = 0.8 + 0.7 + 1.35 + 0 + 0 = 2.85.

[0141] By adjusting the weighting coefficients of the image quality assessment equation, the influence of each parameter on the image quality score Q can be controlled. This implementation helps maintain consistent image quality across different environments.

[0142] like Figure 2 and Figure 3 As shown, compared to the original image Figure 2 After processing by the method provided by this invention, the image is clearer and retains the details of the original image, thus maintaining a high image quality.

[0143] Furthermore, the present invention also provides a computer-readable storage medium storing computer-executable instructions or computer programs, which, when executed by a processor, implement the ROV underwater adaptive lighting method as described in any of the foregoing embodiments.

[0144] It should be noted that the computer-readable medium described above in this invention can be a computer-readable signal medium, a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0145] Computer program code for performing the operations of the present invention can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages.

[0146] Furthermore, the present invention also provides a data processing device, which may include a memory and a processor, wherein the memory may be used to store computer-executable instructions or computer programs, and the processor is used to execute the computer-executable instructions or computer programs stored in the memory to implement the ROV underwater adaptive lighting method as described in any of the foregoing embodiments.

[0147] The flowcharts in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of the systems, methods, and computer program products according to embodiments of the present invention. In this regard, each block in the flowchart may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowchart, and combinations of blocks in the flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0148] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0149] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the invention. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0150] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for adaptive underwater illumination for ROVs, characterized in that, The method includes: Step S1: Input the initial underwater image; Step S2: Use histogram analysis technology to detect the image quality. If the image quality detection result is unqualified, proceed to step S3. If the image quality detection result is qualified, proceed to step S4. Step S3: Determine whether the light source is turned on. If the light source is not turned on, turn on the light source and proceed to step S1. If the light source is already turned on, proceed to step S4. Step S4: Extract multiple features of the image based on a convolutional neural network; Step S5: Based on the multiple features extracted from the image, perform 3D reconstruction, use a generative adversarial network to process the extracted multiple features, generate a 3D image, and the generative adversarial network improves the quality of the 3D image; Step S6: Determine the quality of the three-dimensional image. If the quality of the three-dimensional image is unqualified, adjust the light intensity of the light source and proceed to step S1. If the quality of the three-dimensional image is qualified, complete the lighting adjustment and image processing for the current cycle and prepare to enter the next cycle. The quality of a 3D image is determined using an image quality evaluation equation, which is used to calculate an image quality score. The image quality evaluation equation is as follows: Q=α•C+β•E+γ•L+δ•T+κ•K Where Q is the image quality score; Where C is the image contrast evaluation index; Where E is the edge fidelity evaluation index; Where L is the brightness consistency evaluation index; Where T is the water turbidity variable; Where K is the light color deviation variable; Among them, α, β, γ, δ and κ are weighting coefficients, which are adjusted according to the actual situation; The step of determining the quality of the three-dimensional image in step S6 includes: Calculate real-time image quality score , If the real-time image quality score Smaller than the target image quality score This indicates that the current is insufficient. Increase the current intensity of the light source to increase the light intensity of the light source. If the real-time image quality score Greater than the target image quality score This indicates that the current light intensity is too strong. The light intensity of the light source should be reduced appropriately to reduce the light intensity of the light source.

2. The ROV underwater adaptive lighting method as described in claim 1, characterized in that, Step S1 includes: acquiring initial underwater images using the ROV's sensors and / or cameras, wherein the light source is an LED light.

3. The ROV underwater adaptive lighting method as described in claim 1, characterized in that, The step of detecting the image quality using histogram analysis technology in step S2 includes: The input image is preprocessed, including denoising and normalization. Calculate the grayscale histogram of the image and analyze the brightness distribution of the image; The quality of the image is evaluated based on the uniformity and contrast of the grayscale histogram, and a target image quality score is set. Furthermore, a predetermined range is set, which defines the target image quality score. An acceptable fluctuation range is defined, and a score within the predetermined range indicates acceptable image quality. An image quality score is then calculated. If the image quality score Within the predetermined range, it is indicated that the image quality is acceptable; if the image quality score is... If the image is not within the predetermined range, it indicates that the image quality is substandard.

4. The ROV underwater adaptive lighting method as described in claim 1, characterized in that, The convolutional neural network includes: Input layer, which receives the image; Convolutional layers extract features of different scales and levels from the image through convolution operations, including edges, textures, and higher-order features; Pooling layer, which reduces the dimension of feature map and retains main features through max pooling or average pooling operations. The main features include low-order features and high-order features. The low-order features include edges and textures, and the high-order features include the shape and pose of the object. The activation function includes the Rectified Linear Unit (ReLU), which is used to introduce nonlinear characteristics and enhance the expressive power of the model.

5. The ROV underwater adaptive lighting method as described in claim 1, characterized in that, The generative adversarial network includes: A generator that generates a 3D image based on the multiple features extracted by the convolutional neural network; A discriminator that evaluates the quality of the 3D image generated by the generator.

6. The ROV underwater adaptive lighting method as described in claim 4 or 5, characterized in that, The loss function of the convolutional neural network includes a perturbation-resistant loss term, which is used to train the robustness of the convolutional neural network under noisy conditions. The loss function is: in, It is the mean squared error loss, used to measure the difference in image pixel values: ； in, It is a structural similarity loss, used to measure the structural similarity of images: ； in, It is an adversarial loss used to enhance image details and contrast, implemented through a generative adversarial network: ； in, It is a perturbation resistance loss used to enhance the robustness of the model under different noise conditions: These are noise parameters; in, These are loss items for specific situations, including turbidity and color deviation: in, and It is the weighting coefficient of the loss term.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions or computer programs that, when executed by a processor, implement the method as described in any one of claims 1 to 6.

8. A data processing device, characterized in that, The data processing device includes: The memory is used to store computer-executable instructions or computer programs; A processor, when executing computer-executable instructions or computer programs stored in the memory, implements the method as described in any one of claims 1 to 6.

Images