Image data processing method and device, equipment, storage medium and product

Abstract

The invention discloses an image data processing method and device, equipment, a storage medium and a product, and relates to the technical field of artificial intelligence, and the image data processing method comprises the steps: obtaining to-be-detected underwater image data; performing illumination compensation processing on the underwater image data through a nonlinear illumination compensation algorithm to obtain first image data; performing denoising processing on the first image data to obtain preprocessed data; and inputting the preprocessed data into a preset image detection model, performing prediction processing on the preprocessed data based on the preset image detection model, and outputting an image prediction result, the image prediction result being obtained by performing fusion processing based on image information of different modalities. According to the invention, the classification accuracy of the abnormal detection points of the underwater environment is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an image data processing method, device, equipment, storage medium and computer program product. Background Art

[0002] The cooling system of a nuclear power plant is a key component of its safe and stable operation. It is responsible for transferring the heat generated by the nuclear reactor to the external environment to maintain the normal operating temperature of the reactor. As the core part of the cooling system, the operating status of the cold source area is directly related to the safety and efficiency of the entire nuclear power plant. Therefore, real-time monitoring and abnormal detection of the cold source area is crucial.

[0003] Since the cold source area is in an underwater environment, the refraction and absorption of light in the water, coupled with the disturbance of water flow and the influence of suspended particles, together result in the acquired images often having poor quality, manifested as blur, shape distortion and uneven light distribution. These problems directly weaken the accuracy of subsequent target detection and recognition. At present, median filtering or Gaussian filtering is mainly used to filter out noise to improve image clarity, but the lighting conditions in the underwater environment are complex and changeable, which may cause local overexposure or underexposure of the image. Therefore, for some abnormal detection points, the current image processing methods cannot accurately identify and classify.

[0004] Therefore, how to improve the classification accuracy of abnormal detection points in underwater environments is a problem that needs to be solved urgently.

[0005] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The main purpose of the present application is to provide an image data processing method, device, equipment, storage medium and computer program product, aiming to solve the technical problem in the related technology that median filtering or Gaussian filtering is used to filter out noise to improve image clarity, but the lighting conditions in the underwater environment are complex and changeable, which may cause local overexposure or underexposure of the image. Therefore, for some abnormal detection points, the current image processing method cannot accurately identify and classify the technical problem.

[0007] To achieve the above objectives, the present application proposes an image data processing method, which includes: Acquire underwater image data to be detected; Performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; Performing denoising processing on the first image data to obtain preprocessed data; The preprocessed data is input into a preset image detection model, and based on the preset image detection model, the preprocessed data is predicted and processed, and an image prediction result is output, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0008] In one embodiment, the step of performing denoising on the first image data to obtain pre-processed data includes: Processing the first image data by using a preset optical flow estimation model to obtain offset correction data; Performing pixel denoising processing on the offset-corrected data to obtain second image data; Performing illumination adjustment processing on the second image data by using a light transmission model to obtain third image data; The third image data is segmented to obtain pre-processed data.

[0009] In one embodiment, the step of processing the first image data by using a preset optical flow estimation model to obtain offset correction data includes: Performing image frame analysis on the denoised image data by using a preset optical flow estimation model to obtain a water flow motion vector for each image frame; Correction processing is performed on the water flow motion vector to obtain offset correction data.

[0010] In one embodiment, the step of performing pixel denoising on the offset-corrected data to obtain the second image data includes: Determining a noise estimation result of the offset-corrected data based on an average value within a neighborhood of each pixel on the offset-corrected data; Based on the pixel value difference between each pixel neighborhood and the noise estimation result, the wave noise in the offset correction data is removed by a preset filtering algorithm to obtain the second image data.

[0011] In one embodiment, the step of performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain the first image data includes: Determining an overexposed area and an underexposed area of ​​the underwater image data; A nonlinear illumination compensation algorithm is used to perform illumination compensation operations on the overexposed area and the underexposed area to obtain first image data.

[0012] In one embodiment, the step of performing prediction processing on the preprocessed data based on the preset image detection model and outputting an image prediction result includes: Based on the preset image detection model, extracting color features, texture features and local brightness features of the preprocessed data; Performing splicing processing on the color feature, the texture feature and the local brightness feature to obtain fused feature data; After evaluating the predicted value of the fused feature data by a preset loss function, an image prediction result is output, wherein the preset loss function includes a loss term for image depth information.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes an image data processing device, the image data processing device comprising: An acquisition module, used for acquiring underwater image data to be detected; A compensation processing module, used for performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; A denoising processing module, used for performing denoising processing on the first image data to obtain pre-processed data; An input module is used to input the preprocessed data into a preset image detection model, perform prediction processing on the preprocessed data based on the preset image detection model, and output an image prediction result, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes an image data processing device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the image data processing method described above.

[0015] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the image data processing method described above are implemented.

[0016] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the image data processing method described above are implemented.

[0017] The present application proposes an image data processing method, device, equipment, storage medium and computer program product. The present application obtains underwater image data to be detected, uses a nonlinear illumination compensation algorithm to perform illumination compensation processing on the underwater image data, and obtains first image data. The nonlinear illumination compensation method is adapted to the dynamic illumination changes of the underwater environment, corrects the illumination unevenness of the image, improves the overall brightness and color consistency of the image, avoids the occurrence of uneven illumination distribution of the acquired underwater image data, performs denoising processing on the first image data, and then performs prediction processing on the obtained preprocessed data through a preset image detection model, combines image information of different modes, to improve the accuracy of target detection or image detection, thereby outputting accurate image prediction results, and further, those abnormal detection points underwater can be accurately identified and classified. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 A flowchart of the first embodiment of the image data processing method of the present application is provided; Figure 2 A flowchart of the second embodiment of the image data processing method of the present application is provided; Figure 3 This is a schematic diagram of the overall implementation process of the image data processing method of this application; Figure 4 This is a schematic diagram of the module structure of the image data processing device according to an embodiment of the present application; Figure 5 Schematic diagram of the device structure of the hardware operating environment involved in the image data processing method in the embodiment of the present application.

[0021] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0023] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The main solutions of the embodiments of this application are: Acquire underwater image data to be detected; Performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; Performing denoising processing on the first image data to obtain preprocessed data; The preprocessed data is input into a preset image detection model, and based on the preset image detection model, the preprocessed data is predicted and processed, and an image prediction result is output, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0025] In the prior art, median filtering or Gaussian filtering is mainly used to filter out noise and improve image clarity. However, the lighting conditions in the underwater environment are complex and changeable, which may cause local overexposure or underexposure of the image. Therefore, for some abnormal detection points, the current image processing method cannot accurately identify and classify them.

[0026] The present application proposes an image data processing method, device, equipment, storage medium and computer program product. The present application obtains underwater image data to be detected, uses a nonlinear illumination compensation algorithm to perform illumination compensation processing on the underwater image data, and obtains first image data. The nonlinear illumination compensation method is adapted to the dynamic illumination changes of the underwater environment, corrects the illumination unevenness of the image, improves the overall brightness and color consistency of the image, avoids the occurrence of uneven illumination distribution of the acquired underwater image data, performs denoising processing on the first image data, and then performs prediction processing on the obtained preprocessed data through a preset image detection model, combines image information of different modes, to improve the accuracy of target detection or image detection, thereby outputting accurate image prediction results, and further, those abnormal detection points underwater can be accurately identified and classified.

[0027] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, an image data processing device, etc. The following takes an image data processing device as an example to illustrate this embodiment and the following embodiments.

[0028] Based on this, the present application embodiment provides an image data processing method, referring to Figure 1 , Figure 1 This is a flowchart of the first embodiment of the image data processing method of the present application.

[0029] In this embodiment, the image data processing method includes steps S10 to S40: Step S10, acquiring underwater image data to be detected; It should be noted that this application is applicable to the recognition of abnormal status of the cold source area of ​​a nuclear power plant, and can also be used to process other complex underwater operation images, such as underwater building maintenance, underwater pipeline repair, and underwater detection, etc. The underwater image data can be underwater equipment data, pipeline data, and other data that needs to be detected, etc., captured by a camera, and the specific scenario is not limited.

[0030] Step S20, performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; It should be noted that the nonlinear compensation algorithm is a technology used to reduce or eliminate signal distortion caused by the nonlinear characteristics of the system during transmission. The underwater image data is affected by light refraction and scattering, which will cause the received image data to be unclear. Therefore, it is necessary to perform light compensation on the underwater image data.

[0031] It should be noted that the first image data is data after illumination compensation processing. It can be concluded that the illumination non-uniformity of the first image data is significantly reduced.

[0032] In a feasible implementation manner, the step S20 of performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain the first image data includes: Determining an overexposed area and an underexposed area of ​​the underwater image data; It should be noted that the overexposed area is an area with too much light, and the underexposed area is an area with lack of light. The overexposed area and the underexposed area can be determined by setting the lighting standard.

[0033] A nonlinear illumination compensation algorithm is used to perform illumination compensation operations on the overexposed area and the underexposed area to obtain first image data.

[0034] It should be noted that illumination compensation is performed on the overexposed and underexposed areas of the underwater image. In order to adapt to the nonlinear characteristics of uneven illumination distribution in actual situations, a nonlinear compensation algorithm is adopted. That is, the logarithm of the linear compensation algorithm is taken as the pixel value of each pixel after illumination compensation to obtain the first image data.

[0035] Step S30, performing denoising processing on the first image data to obtain pre-processed data; It should be noted that there are many factors that affect underwater image data, and it is also necessary to remove water flow effects, scattering effects and wave noise. The preprocessed data obtained is the data to be detected that is ultimately input into the neural network model. In the embodiment of the present application, the underwater image data is mainly denoised in a diverse manner to reduce the impact of the chaotic underwater environment on the collected image data, thereby improving the accuracy of the model prediction.

[0036] In a feasible implementation manner, the step S30 of performing denoising on the first image data to obtain pre-processed data includes: Processing the first image data by using a preset optical flow estimation model to obtain offset correction data; It should be noted that the preset optical flow estimation model can be a PWC-Net optical flow learning network model. PWC-Net adopts a pyramid feature extractor, which processes the input image through multiple layers of convolutional layers to generate feature representations of different scales. As the number of levels increases, the size of the feature map gradually decreases, while the number of channels increases, thereby capturing richer contextual information. The PWC-Net optical flow learning network model analyzes the relationship between consecutive frame images, accurately estimates the two-dimensional motion vector of the pixel points in each frame image in the time dimension, reveals the water flow speed and direction, and corrects the resulting impact on the image to obtain offset correction data.

[0037] In a feasible implementation manner, the step of processing the first image data by using a preset optical flow estimation model to obtain offset correction data includes: Performing image frame analysis on the denoised image data by using a preset optical flow estimation model to obtain a water flow motion vector for each image frame; It should be noted that the water flow motion vector may be a two-dimensional motion vector of a pixel point in each frame of the image in the time dimension, which is used to indicate the speed and direction of the water flow.

[0038] Correction processing is performed on the water flow motion vector to obtain offset correction data.

[0039] It should be noted that the preset optical flow estimation model can be a PWC-Net optical flow learning network model. PWC-Net adopts a pyramid feature extractor, which processes the input image through multiple layers of convolutional layers to generate feature representations of different scales. As the number of levels increases, the size of the feature map gradually decreases, while the number of channels increases, thereby capturing richer contextual information. The PWC-Net optical flow learning network model analyzes the relationship between consecutive frame images, accurately estimates the two-dimensional motion vector of the pixel points in each frame image in the time dimension, reveals the water flow speed and direction, and corrects the resulting impact on the image to obtain offset correction data.

[0040] Performing pixel denoising processing on the offset-corrected data to obtain second image data; It should be noted that the pixel denoising process is mainly to calculate the wave noise in the image. For the unique underwater wave noise, the local noise level is evaluated by calculating the average value in the neighborhood of each pixel to provide a basis for subsequent processing.

[0041] In a feasible implementation manner, the step of performing pixel denoising on the offset-corrected data to obtain the second image data includes: Determining a noise estimation result of the offset-corrected data based on an average value within a neighborhood of each pixel on the offset-corrected data; Based on the pixel value difference between each pixel neighborhood and the noise estimation result, the wave noise in the offset correction data is removed by a preset filtering algorithm to obtain the second image data.

[0042] It should be noted that the filter calculates the average value in the neighborhood of each pixel and evaluates the local noise level to provide a basis for subsequent processing; in the adaptive filter weight adjustment stage, the filter weight is dynamically adjusted according to the pixel value difference and noise estimation results to ensure effective noise suppression while protecting image details. Combined with the total variation regularization method based on sparse representation, it can accurately restore the image, minimize the sum of squared errors, and control the total variation of the image to achieve the effect of noise reduction and edge preservation. The red dark channel prior method targets the unique underwater wave noise. By analyzing the dark channel and transmittance, it effectively removes fog and stripes, restores image clarity, and presents natural colors.

[0043] It should be noted that the second image data may be image data from which wave noise has been removed.

[0044] Performing illumination adjustment processing on the second image data by using a light transmission model to obtain third image data; It should be noted that image correction is performed through physical models, and the light transmission model is used to compensate for the light attenuation of each pixel to achieve brightness balance and color restoration, thereby reducing the impact of the underwater environment on image clarity; adaptive filtering technology is used to eliminate image noise in a targeted manner. This method can dynamically adjust the filtering intensity according to the local characteristics of the image, effectively reducing the interference caused by water body unevenness and sensor noise, while maintaining the integrity of image details; taking into account the color deviation caused by the characteristics of underwater light sources and the optical properties of water bodies, white balance correction is implemented, and the color is corrected pixel by pixel by adjusting the gain of the red and blue channels relative to the green channel to ensure the natural and realistic color of the image.

[0045] It should be noted that the third image data may be image data after illumination adjustment processing is performed using a light transmission model.

[0046] The third image data is segmented to obtain pre-processed data.

[0047] It should be noted that the segmentation process may be performed by using a variational level set based method, which converts image segmentation into an image segmentation boundary represented by a zero level set by finding an evolution problem of a continuous function (level set function).

[0048] Specifically, the Chan-Vese model energy functional is used for optimization. The Chan-Vese model constructs an energy function that includes the difference between the average values ​​of the internal and external pixel values ​​of the two regions and a length term that depends on the boundary of the segmented region. The purpose of the model is to minimize this energy function to obtain the optimal segmentation result.

[0049] The functional contains the internal consistency term of the region, the discontinuity term between regions, and the regularization term of the level set function. The energy is minimized through iterative optimization, thereby gradually obtaining the segmentation result, that is, the precise boundary of the target object. In addition, the depth information is combined as an additional constraint to encourage the segmentation boundary to appear within the specified depth range, further improving the accuracy and adaptability of the segmentation. Finally, the obtained pre-processed data is input into the preset image detection model for detection processing.

[0050] It should be noted that in the process of denoising underwater image data, the Jaffe-McGlamery model is used to simulate the underwater image degradation process. The Jaffe-McGlamery model is a classic underwater imaging model that describes the physical formation process of underwater images and is widely used in underwater image enhancement and restoration tasks. Based on the principle of underwater optical imaging, the formation of underwater images is attributed to a linear combination of three main components, namely the direct component, the forward scattered component, and the backscattered component. This division reflects the complexity of light propagation in underwater environments, where each component has a different impact on the final image quality.

[0051] Step S40, inputting the preprocessed data into a preset image detection model, performing prediction processing on the preprocessed data based on the preset image detection model, and outputting an image prediction result, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0052] It should be noted that the preset image detection model can be a neural network model based on deep learning, which is used to perform predictive processing on the input preprocessed data and output classification results / image prediction results. When in use, the preset image detection model has been trained based on historical sample data.

[0053] It should be noted that the data of different modalities include the depth information, color features, texture features and local brightness features of underwater image data. By fusing the information of different modalities, the final prediction accuracy of the model can be improved. In the process of model prediction, by adding an optimization term based on local information entropy when constructing the segmentation energy function, the segmentation results are encouraged to produce more subdivisions in areas with higher information entropy, because these areas often contain rich details and edge information. By dynamically adjusting the weights of each feature, the algorithm can show stronger adaptability and robustness in different environments. The use of deep learning neural networks to fuse adaptive features is mainly reflected in network structure design and loss function optimization. A multi-path network structure is designed, and each path processes a feature, including color features, texture features and local brightness features, which are fused together through simple splicing operations, and the fused features are sent to the fully connected layer or other feature fusion layers for final classification or segmentation prediction.

[0054] The present application proposes an image data processing method, device, equipment, storage medium and computer program product. The present application obtains underwater image data to be detected, uses a nonlinear illumination compensation algorithm to perform illumination compensation processing on the underwater image data, and obtains first image data. The nonlinear illumination compensation method is adapted to the dynamic illumination changes of the underwater environment, corrects the illumination unevenness of the image, improves the overall brightness and color consistency of the image, avoids the occurrence of uneven illumination distribution of the acquired underwater image data, performs denoising processing on the first image data, and then performs prediction processing on the obtained preprocessed data through a preset image detection model, combines image information of different modes, to improve the accuracy of target detection or image detection, thereby outputting accurate image prediction results, and further, those abnormal detection points underwater can be accurately identified and classified.

[0055] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 The step S40 of performing prediction processing on the preprocessed data based on the preset image detection model and outputting the image prediction result comprises steps S41 to S43: Step S41, extracting color features, texture features, and local brightness features of the preprocessed data based on the preset image detection model; It should be noted that a multi-path deep learning network structure is added to the preset image detection model. The weight of each feature is dynamically adjusted through the learning mechanism to improve the adaptability and robustness of image segmentation. The network structure contains three branches: color features, texture features, and local brightness features. Each branch processes a feature. Different features in the image represent different attributes of the image data. After the three are integrated, various influencing factors of the image data can be fully considered.

[0056] Step S42, performing splicing processing on the color feature, the texture feature and the local brightness feature to obtain fused feature data; It should be noted that the color feature branch directly extracts RGB color features through the convolution layer; the texture feature branch uses a specially designed texture feature extraction layer to extract texture features; the local brightness feature branch uses the image gradient operator and Gaussian filter to highlight the local brightness contrast. After obtaining the color features, texture features and local brightness features, the three are spliced ​​together to obtain fused feature data.

[0057] Step S43: After evaluating the predicted value of the fused feature data by a preset loss function, an image prediction result is output, wherein the preset loss function includes a loss term of image depth information.

[0058] It should be noted that the fused feature data is sent to the fully connected layer or the feature fusion layer for final classification or segmentation prediction to obtain the predicted value. A regularization term based on depth information is added after the convolution layer, and a loss term is introduced to establish a relationship between certain feature responses predicted by the model and the given depth value. The depth information is used as an additional constraint in the energy function to encourage the segmentation boundary to appear only within the specified depth range, further improving the accuracy and adaptability of the segmentation.

[0059] It should be noted that after calculating the error between the predicted value and the true value through the loss function, the image prediction result is output.

[0060] In this embodiment, the three branches of color features, texture features and local brightness features are spliced ​​and fused together to obtain fused feature data, thereby improving the accuracy of image segmentation and the adaptability of the preset image detection model to complex scenes.

[0061] This application has been experimentally verified to prove its feasibility and effectiveness. The experimental design focuses on the implementation of the algorithm, performance evaluation and comparison with existing technologies to prove that the algorithm proposed in this paper has more advantages in performance than traditional algorithms.

[0062] The experiment is based on Python programming language, and uses Pytorch framework to build a deep learning model for underwater image processing. The final experimental results are given by Matplotlib image analysis tool, and Nvidia GeForce RTX3050 Laptop GPU server is used to accelerate the deep learning model training and reasoning process. The underwater image acquisition equipment includes a high-sensitivity underwater camera, a high-definition imaging system carried by an underwater robot, and necessary lighting equipment to adapt to the low-light environment of cold underwater.

[0063] The experimental data are all composed of RGB three-color images, which consist of two parts. One part is collected from the underwater operation site in the cold zone of the nuclear power plant, covering different time periods, lighting conditions and water quality conditions, and is real data; the other part comes from images collected non-underwater. Through the known underwater optical mechanism model, this part of the images is atomized and denoised to simulate image degradation under different depths, suspended particle concentrations, and light intensities. The combination of real data and simulated data enriches the diversity of training samples and improves the generalization ability of the deep learning model.

[0064] In terms of image dehazing performance, the peak signal-to-noise ratio (PSNR) is used to evaluate the image quality of DCP and physical model-based dehazing methods. The higher the signal-to-noise ratio, the smaller the proportion of noise signals in the image, and the better the image noise filtering effect. The results show that the dehazing effect of dark channel prior is better than that of the dehazing method based entirely on the physical model; the accuracy of target detection is used to determine the probability that the system correctly identifies the detected target. The higher the accuracy, the better the target detection performance. The experimental part compares the target detection accuracy of Faster R-CNN and YOLOv5 algorithms; the Dice coefficient is used to determine the pros and cons of image segmentation performance: the larger the Dice value, the more consistent the target detection segmentation result and the true segmentation result, and the better the segmentation effect.

[0065] The experimental results show that the system exhibits excellent stability and accuracy when processing underwater operation images in the cold zone of nuclear power plants. On the one hand, the image enhancement function significantly improves the visual clarity and detail presentation of underwater images, making the equipment status that was originally difficult to identify due to the interference of water media clear at a glance, greatly reducing the risk of misjudgment and improving operational safety.

[0066] For example, to help understand the implementation process of the image data processing method obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 3 , Figure 3 A brief flowchart of an image data processing method is provided, specifically: Taking the identification of the cold source area of ​​a nuclear power plant as an example, an underwater camera is used to collect underwater images of the cold source area of ​​the nuclear power plant. By preprocessing the underwater image data, preprocessed data is obtained. The deep learning model training only requires one training process. The model learns to identify the characteristics of abnormal conditions such as underwater foreign objects and pipeline cracks, and performs multiple iterative training. After the model training is completed, the real-time collected underwater images are input into the trained convolutional neural network model, so that the model analyzes the image data and identifies and classifies abnormal conditions in real time.

[0067] After obtaining the recognition results, information such as the abnormal location can be determined, and whether to trigger an alarm or guide maintenance work can be determined based on the recognition results.

[0068] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the image data processing method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0069] This application also provides an image data processing device, please refer to Figure 4 , the image data processing device comprises: An acquisition module 10 is used to acquire underwater image data to be detected; A compensation processing module 20, configured to perform illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; A denoising processing module 30, configured to perform denoising processing on the first image data to obtain pre-processed data; The input module 40 is used to input the preprocessed data into a preset image detection model, perform prediction processing on the preprocessed data based on the preset image detection model, and output an image prediction result, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0070] Optionally, the denoising processing module includes: A first processing unit, configured to process the first image data by using a preset optical flow estimation model to obtain offset correction data; A second processing unit, configured to perform pixel denoising on the offset-corrected data to obtain second image data; A third processing unit, configured to perform illumination adjustment processing on the second image data through a light transmission model to obtain third image data; The fourth processing unit is used to perform segmentation processing on the third image data to obtain pre-processed data.

[0071] Optionally, the first processing unit includes: An analysis subunit, configured to perform image frame analysis on the denoised image data by using a preset optical flow estimation model to obtain a water flow motion vector for each image frame; The correction processing subunit is used to perform correction processing on the water flow motion vector to obtain offset correction data.

[0072] Optionally, the second processing unit includes: a determination subunit, configured to determine a noise estimation result of the offset-corrected data based on an average value within a neighborhood of each pixel on the offset-corrected data; The removal processing subunit is used to remove the wave noise in the offset correction data through a preset filtering algorithm based on the pixel value difference between each pixel neighborhood and the noise estimation result to obtain the second image data.

[0073] Optionally, the compensation processing module includes: A determination unit, configured to determine an overexposed area and an underexposed area of ​​the underwater image data; An operating unit is used to perform an illumination compensation operation on the overexposed area and the underexposed area by using a nonlinear illumination compensation algorithm to obtain first image data.

[0074] Optionally, the input module includes: An extraction unit, configured to extract color features, texture features, and local brightness features of the preprocessed data based on the preset image detection model; A splicing processing unit, used for splicing the color feature, the texture feature and the local brightness feature to obtain fused feature data; An output unit is used to output an image prediction result after evaluating the predicted value of the fused feature data through a preset loss function, wherein the preset loss function includes a loss term of image depth information.

[0075] The image data processing device provided by the present application adopts the image data processing method in the above embodiment, which can solve the technical problem of image data processing. Compared with the prior art, the beneficial effects of the image data processing device provided by the present application are the same as the beneficial effects of the image data processing method provided by the above embodiment, and other technical features in the image data processing device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0076] The present application provides an image data processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the image data processing method in the above-mentioned embodiment one.

[0077] Reference below Figure 5 , which shows a schematic diagram of the structure of an image data processing device suitable for implementing an embodiment of the present application. The image data processing device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The image data processing device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0078] like Figure 5 As shown, the image data processing device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the image data processing device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the image data processing device to communicate with other devices wirelessly or by wire to exchange data. Although the image data processing device with various systems is shown in the figure, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0079] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0080] The image data processing device provided by the present application adopts the image data processing method in the above embodiment to solve the technical problem of image data processing. Compared with the prior art, the beneficial effects of the image data processing device provided by the present application are the same as the beneficial effects of the image data processing method provided by the above embodiment, and other technical features in the image data processing device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0081] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0082] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0083] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the image data processing method in the above-mentioned embodiment.

[0084] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0085] The computer-readable storage medium may be included in the image data processing device; or may exist independently without being installed in the image data processing device.

[0086] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the image data processing device, the image data processing device: Acquire underwater image data to be detected; Performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; Performing denoising processing on the first image data to obtain preprocessed data; The preprocessed data is input into a preset image detection model, and based on the preset image detection model, the preprocessed data is predicted and processed, and an image prediction result is output, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

[0087] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0088] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0089] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0090] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned image data processing method, and can solve the technical problems of image data processing. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the image data processing method provided in the above-mentioned embodiment, and will not be described in detail here.

[0091] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned image data processing method when executed by a processor.

[0092] The computer program product provided by the present application can solve the technical problem of image data processing. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the image data processing method provided by the above embodiment, which will not be repeated here.

[0093] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for processing image data, characterized in that: The method comprises: Acquire underwater image data to be detected; Performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; Performing denoising processing on the first image data to obtain preprocessed data; The preprocessed data is input into a preset image detection model, and based on the preset image detection model, the preprocessed data is predicted and processed, and an image prediction result is output, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

2. The method according to claim 1, characterized in that The step of performing denoising on the first image data to obtain pre-processed data comprises: Processing the first image data by using a preset optical flow estimation model to obtain offset correction data; Performing pixel denoising processing on the offset-corrected data to obtain second image data; Performing illumination adjustment processing on the second image data by using a light transmission model to obtain third image data; The third image data is segmented to obtain pre-processed data.

3. The method according to claim 2, characterized in that The step of processing the first image data by using a preset optical flow estimation model to obtain offset correction data includes: Performing image frame analysis on the denoised image data by using a preset optical flow estimation model to obtain a water flow motion vector for each image frame; Correction processing is performed on the water flow motion vector to obtain offset correction data.

4. The method according to claim 2, characterized in that The step of performing pixel denoising on the offset-corrected data to obtain second image data comprises: Determining a noise estimation result of the offset-corrected data based on an average value within a neighborhood of each pixel on the offset-corrected data; Based on the pixel value difference between each pixel neighborhood and the noise estimation result, the wave noise in the offset correction data is removed by a preset filtering algorithm to obtain the second image data.

5. The method according to claim 1, characterized in that The step of performing illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data comprises: Determining an overexposed area and an underexposed area of ​​the underwater image data; A nonlinear illumination compensation algorithm is used to perform illumination compensation operations on the overexposed area and the underexposed area to obtain first image data.

6. The method according to claim 1, characterized in that The step of performing prediction processing on the preprocessed data based on the preset image detection model and outputting an image prediction result comprises: Based on the preset image detection model, extracting color features, texture features and local brightness features of the preprocessed data; Performing splicing processing on the color feature, the texture feature and the local brightness feature to obtain fused feature data; After evaluating the predicted value of the fused feature data by a preset loss function, an image prediction result is output, wherein the preset loss function includes a loss term for image depth information.

7. An image data processing device, characterized in that: The device comprises: An acquisition module, used for acquiring underwater image data to be detected; A compensation processing module, used to perform illumination compensation processing on the underwater image data by using a nonlinear illumination compensation algorithm to obtain first image data; A denoising processing module, used for performing denoising processing on the first image data to obtain pre-processed data; An input module is used to input the preprocessed data into a preset image detection model, perform prediction processing on the preprocessed data based on the preset image detection model, and output an image prediction result, wherein the image prediction result is obtained after fusion processing based on image information of different modalities.

8. An image data processing device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the image data processing method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the image data processing method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the image data processing method according to any one of claims 1 to 6 are implemented.

Images