Intelligent water gauge identification system based on big data technology

Abstract

The invention relates to the technical field of water gauge identification, in particular to an intelligent water gauge identification system based on a big data technology, which comprises a data acquisition and calculation module, a denoising processing module and a texture restoration module, and is characterized in that the data acquisition and calculation module calculates a low-frequency illumination component and a high-frequency detail component of each pixel in a water gauge image by using a Gaussian filter; the denoising processing module selects each pixel in the image as a central pixel to be denoised, calculates a similarity weight between the central pixel and a pixel in a neighborhood, calculates a denoised pixel value according to the weight and a low-frequency illumination component of the pixel in the neighborhood, and combines the denoised pixel value with a high-frequency detail component to generate a complete denoised pixel; and the texture restoration module further selects all the denoised pixels as central pixels to be textured, calculates comprehensive weights of the central pixels and pixels in a neighborhood, calculates texture restoration pixels based on the weights and low-frequency illumination components, and restores texture missing or distortion caused by noise, so that a final image is more natural and real.

Description

Technical Field

[0001] The present invention relates to the technical field of water gauge recognition, and in particular to an intelligent water gauge recognition system based on big data technology. Background Art

[0002] Traditional intelligent water gauge recognition systems have the following major technical problems when processing images.

[0003] First, traditional denoising methods usually rely on local filters, such as mean filtering or median filtering, which can easily lead to the loss of image details while removing noise. Especially when dealing with complex water level images, edge and texture information may be over-smoothed, thus affecting the accuracy of water level readings. Secondly, traditional denoising algorithms are often too sensitive to the type and intensity of noise. When there is high-density or complex-structured noise in the image, the denoising effect is not ideal and may even introduce new artifacts, further distorting the image.

[0004] In addition, when processing images with uneven illumination, traditional water gauge recognition systems often fail to effectively separate illumination changes and texture information, resulting in some noise remaining in the denoised image, affecting the accuracy of subsequent recognition. These problems become more prominent in the context of big data technology applications, because big data processing requires higher image quality and robustness, and traditional denoising methods are difficult to meet these requirements. Therefore, traditional intelligent water gauge recognition systems have obvious deficiencies in denoising and texture restoration, and more advanced technologies are needed to improve the effect of image processing and ensure the accuracy and reliability of water gauge recognition. Summary of the invention

[0005] The purpose of the present invention is to provide an intelligent water gauge identification system based on big data technology to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: an intelligent water gauge recognition system based on big data technology, comprising a data acquisition and calculation module, a denoising processing module and a texture restoration module, wherein: The data acquisition and calculation module acquires the water gauge image and uses a Gaussian filter to calculate the low-frequency illumination component and high-frequency detail component of each pixel in the water gauge image; The denoising processing module selects a pixel in the water gauge image as the central pixel to be denoised, and selects a pixel in the neighborhood of the central pixel to be denoised as the local pixel to be denoised; calculates the similarity weight between the central pixel to be denoised and the local pixel to be denoised, and calculates the denoised pixel of the central pixel to be denoised according to the size of the similarity weight and the low-frequency illumination component of the local pixel to be denoised; combines the denoised pixel of the central pixel to be denoised with the high-frequency detail component of the central pixel to be denoised, and calculates the complete denoised pixel of the central pixel to be denoised; repeatedly selects each pixel in the water gauge image, and calculates the corresponding complete denoised pixel, until all pixels are processed; The texture restoration module selects one pixel from all complete denoised pixels as the central pixel to be textured, and selects a pixel in the neighborhood of the central pixel to be textured as the local pixel to be textured, calculates the similarity weight and texture similarity between the central pixel to be textured and the local pixel to be textured; calculates the comprehensive weight of the central pixel to be textured and the local pixel to be textured according to the size of the similarity weight and the texture similarity; calculates the texture restoration pixel of the central pixel to be textured according to the comprehensive weight and the low-frequency illumination component of the local pixel to be textured; repeatedly selects each pixel from all complete denoised pixels, and calculates the corresponding texture restoration pixel, until all complete denoised pixels are processed.

[0007] As a further improvement of the technical solution, the data acquisition and calculation module includes a data acquisition unit, and the data acquisition unit uses an image acquisition device to acquire a water gauge image. The acquired water gauge image is represented as ,in are the pixel coordinates in the water gauge image.

[0008] As a further improvement of the technical solution, the data acquisition and calculation module includes a data calculation unit, which uses a Gaussian filter to convert the water gauge image Decomposition into low-frequency illumination components and high frequency detail components , where the Gaussian filter is a low-pass filter used to smooth the image and remove high-frequency noise. The decomposition process is as follows: Choose a Gaussian kernel of fixed size ; Gaussian kernel Image with water gauge Perform convolution operations on each pixel and its neighborhood in the image to calculate the low-frequency illumination component of each pixel. ; Image via Water Gauge Subtract low-frequency lighting components , calculate the high-frequency detail component of each pixel .

[0009] As a further improvement of the technical solution, the process of generating a complete denoised pixel by the denoising processing module is as follows: From the water gauge image Select a pixel As the central pixel to be denoised; At the center pixel to be denoised Neighborhood Select a pixel As the local pixel to be denoised, the neighborhood Defined as a A square area centered on Calculate the center pixel to be denoised and the local pixel to be denoised The similarity weight between , the calculation formula is as follows: ,in and is the center pixel to be denoised and the local pixels to be denoised The low-frequency illumination component of is the similarity threshold parameter; where the similarity weight The center pixel to be denoised is measured and the local pixels to be denoised The similarity between pixels is calculated and the influence of noise on the central pixel is reduced by weighted averaging of similar pixels in the neighborhood. Based on similarity weight and the local pixels to be denoised The low-frequency illumination component , calculate the center pixel to be denoised Denoised pixels , the calculation formula is as follows: ; The center pixel to be denoised Denoised pixels and the center pixel to be denoised The high-frequency detail components Combine and calculate the center pixel to be denoised Complete denoised pixels , the calculation formula is as follows: ; Repeat the selection of the water gauge image For each pixel in the image, the corresponding complete denoised pixel value is calculated until all pixels are processed.

[0010] As a further improvement of the technical solution, the process of the texture repair module generating texture repair pixels is as follows: Select one pixel among all the complete denoised pixels as the center pixel to be textured , and select the center pixel to be textured Neighborhood A pixel within is used as the local pixel to be textured , calculate the center pixel of the texture The local pixel to be textured The similarity weight between And texture similarity, where the calculation process of texture similarity is as follows: Extract the center pixel of the texture and the local pixel to be textured The texture similarity vector and ; Computing texture similarity , ,in is the texture similarity threshold parameter; Combined similarity weights and texture similarity , calculate the comprehensive weight , where the calculation formula is as follows: ; According to the comprehensive weight And the local pixels to be textured The low-frequency illumination component , calculate the center pixel of the texture Texture repair pixels , the calculation formula is as follows: ; Repeatedly select each pixel in all complete denoised pixels and calculate the corresponding texture restoration pixel until all complete denoised pixels are processed.

[0011] Compared with the prior art, the present invention has the following beneficial effects: The intelligent water gauge recognition system based on big data technology can effectively measure the similarity between the central pixel and the pixels in the neighborhood by calculating the similarity weight, thereby reducing the interference of noise in the denoising process; the larger the similarity weight, the more similar the pixels in the neighborhood are to the central pixel, and the greater the contribution of these similar pixels in the denoising process, and vice versa; by weighted averaging the low-frequency illumination components in the neighborhood, it is possible to effectively remove high-frequency noise while retaining the overall structural information of the image. Subsequently, the denoised low-frequency illumination components are combined with the high-frequency detail components to retain the edge and texture information of the image and avoid image distortion caused by excessive smoothing; therefore, this process can not only significantly reduce the noise in the image, but also maintain the detailed features of the image and improve the overall quality of the image.

[0012] In addition, by combining similarity weights and texture similarity, the relationship between pixels in the neighborhood and the central pixel can be better reflected; the similarity weight measures the similarity of pixel values, while texture similarity takes into account the similarity of local texture features, which enables more accurate use of similar texture information in the neighborhood during the restoration process; by calculating the comprehensive weight, the texture can be further restored on the basis of denoising, so that the restored image not only removes the noise, but also restores the texture features of the original image; finally, by combining the denoised low-frequency illumination component with the restored texture information, the problem of texture loss or distortion caused by noise in the image can be effectively solved, making the restored image more natural and realistic. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the overall module of the present invention; Figure 2 It is a schematic diagram of the data acquisition and calculation module unit of the present invention.

[0014] In the figure: 100, data acquisition and calculation module; 101, data acquisition unit; 102, data calculation unit; 200, denoising processing module; 300, texture restoration module. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0016] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0017] The following are some definitions of terms: Low-frequency illumination component: refers to the smoother part of the water gauge image that reflects illumination changes, usually containing the brightness and chromaticity information of the water gauge image, but excluding high-frequency details; the low-frequency illumination component can be separated from the water gauge image by a smoothing filter such as a Gaussian filter; High-frequency detail component: refers to the high-frequency part of the water gauge image that contains edges, details and noise; the high-frequency detail component reflects the texture and structural information of the water gauge image, and combined with the low-frequency illumination component, it can restore the details of the water gauge image; the high-frequency detail component is usually obtained by subtracting the low-frequency illumination component from the water gauge image; Local pixels to be denoised: During the denoising process, pixels adjacent to the central pixel to be denoised are used to calculate the denoising value of the central pixel; these pixels are usually selected within the neighborhood of the central pixel; Similarity weight: a value used to measure the similarity between the central pixel to be denoised and the local pixel to be denoised. The larger the similarity weight, the more similar the two pixels are, and the greater the contribution of the local pixel to be denoised in the denoising process. The similarity weight can be calculated by the difference in pixel values, local structure or texture features.

[0018] Complete denoised pixel: In the denoising processing module 200, the pixel value obtained by recombining the denoised low-frequency illumination component with the retained high-frequency detail component; the complete denoised pixel is intended to retain the detailed features of the image while reducing noise interference; Center pixel to be textured: in the texture restoration module 300, the pixel whose texture needs to be restored is selected; this pixel is the target pixel in the restoration process, and the pixels in its neighborhood will be used to calculate the restoration value; Local pixels to be textured: In the texture restoration module 300, pixels adjacent to the central pixel to be textured are used to calculate the texture restoration value of the central pixel. These pixels are usually selected in the neighborhood of the central pixel; Texture similarity: a value used to measure the similarity of texture features between the central pixel to be textured and the local pixel to be textured. The greater the texture similarity, the more similar the texture features of the two pixels are, and the greater the contribution of the local pixel to be textured in the texture restoration process. Texture similarity can be calculated by the matching degree of local texture features. Comprehensive weight: In the texture restoration module 300, the weight value obtained by combining the similarity weight and the texture similarity; the comprehensive weight is used to more accurately reflect the relationship between the central pixel to be textured and the local pixel to be textured, so as to better utilize similar texture information in the restoration process; Texture repair pixel: In the texture repair module 300, the repaired pixel value is calculated by comprehensive weights and low-frequency illumination components of the local pixel to be textured; the texture repair pixel is intended to restore the texture loss or distortion caused by noise in the image, making the final image more natural and realistic.

[0019] Next, see Figure 1-Figure 2 The present invention provides a technical solution: an intelligent water gauge recognition system based on big data technology, including a data acquisition and calculation module 100, a denoising processing module 200 and a texture restoration module 300.

[0020] The data acquisition unit 101 in the data acquisition and calculation module 100 uses a high-definition camera or other image acquisition equipment to acquire a water gauge image. The water gauge image may be affected by environmental factors such as uneven illumination and noise. The acquired water gauge image is represented as ,in are the pixel coordinates in the water gauge image.

[0021] The data calculation unit 102 in the data acquisition and calculation module 100 uses a Gaussian filter to transform the water gauge image Decomposition into low-frequency illumination components and high frequency detail components , where the Gaussian filter is a low-pass filter used to smooth the image and remove high-frequency noise. The decomposition process is as follows: Choose a Gaussian kernel of fixed size , for example, a 5×55×5 or 7×77×7 kernel; Gaussian kernel Image with water gauge Perform convolution operations on each pixel and its neighborhood in the image to calculate the low-frequency illumination component of each pixel. ; Image via Water Gauge Subtract low-frequency lighting components , calculate the high-frequency detail component of each pixel ; The high-frequency detail component retains the edge, texture and other detail information in the image, which helps to restore the detailed features of the image after denoising; and the high-frequency detail component usually contains high-frequency noise, which will be further processed in the subsequent denoising process.

[0022] The denoising processing module 200 is used to process the water gauge image. Select a pixel As the central pixel to be denoised; At the center pixel to be denoised Neighborhood Select a pixel As the local pixel to be denoised, the neighborhood Defined as a A square area with a center such as a 5×55×5 or 7×77×7 neighborhood; Calculate the center pixel to be denoised and the local pixel to be denoised The similarity weight between , the calculation formula is as follows: ,in and is the center pixel to be denoised and the local pixels to be denoised The low-frequency illumination component of is the similarity threshold parameter; where the similarity weight The center pixel to be denoised is measured and the local pixels to be denoised The similarity between pixels is calculated and the influence of noise on the central pixel is reduced by weighted averaging of similar pixels in the neighborhood. Based on similarity weight and the local pixels to be denoised The low-frequency illumination component , calculate the center pixel to be denoised Denoised pixels , the calculation formula is as follows: ; The center pixel to be denoised Denoised pixels and the center pixel to be denoised The high-frequency detail components Combine and calculate the center pixel to be denoised Complete denoised pixels , the calculation formula is as follows: ; By adding the denoised low-frequency illumination component to the high-frequency detail component, the detailed features of the image are restored, while reducing the influence of noise, making the denoised image clearer, with less noise and more detailed features; Repeat the selection of the water gauge image For each pixel in the image, the corresponding complete denoised pixel value is calculated until all pixels are processed.

[0023] The texture restoration module 300 selects one pixel from all the complete denoised pixels as the center pixel to be textured. , and select the center pixel to be textured Neighborhood A pixel within is used as the local pixel to be textured , calculate the center pixel of the texture The local pixel to be textured The similarity weight between And texture similarity, where the calculation process of texture similarity is as follows: Extract the center pixel of the texture and the local pixel to be textured The texture similarity vector and ; Computing texture similarity , ,in is the texture similarity threshold parameter; by introducing texture similarity, the global texture information can be used in the subsequent restoration process to effectively restore the distorted texture and make the image more complete and beautiful; that is, in the restoration process, the algorithm not only removes the noise, but also retains the detailed features of the image, avoiding image distortion caused by excessive restoration; Combined similarity weights and texture similarity , calculate the comprehensive weight , where the calculation formula is as follows: ; According to the comprehensive weight And the local pixels to be textured The low-frequency illumination component , calculate the center pixel of the texture Texture repair pixels , the calculation formula is as follows: ; Repeatedly select each pixel in all complete denoised pixels and calculate the corresponding texture restoration pixel until all complete denoised pixels are processed.

[0024] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The intelligent water gauge recognition system based on big data technology is characterized by: The method comprises a data acquisition and calculation module (100), a denoising processing module (200) and a texture restoration module (300), wherein: The data acquisition and calculation module (100) acquires a water gauge image and uses a Gaussian filter to calculate a low-frequency illumination component and a high-frequency detail component of each pixel in the water gauge image; The denoising processing module (200) selects a pixel in the water gauge image as the central pixel to be denoised, and selects a pixel in the neighborhood of the central pixel to be denoised as the local pixel to be denoised; calculates the similarity weight between the central pixel to be denoised and the local pixel to be denoised, and calculates the denoised pixel of the central pixel to be denoised according to the size of the similarity weight and the low-frequency illumination component of the local pixel to be denoised; combines the denoised pixel of the central pixel to be denoised with the high-frequency detail component of the central pixel to be denoised, and calculates the complete denoised pixel of the central pixel to be denoised; repeatedly selects each pixel in the water gauge image, and calculates the corresponding complete denoised pixel, until all pixels are processed; The texture restoration module (300) selects one pixel from all complete denoised pixels as a central pixel to be textured, selects a pixel in the neighborhood of the central pixel to be textured as a local pixel to be textured, calculates a similarity weight and texture similarity between the central pixel to be textured and the local pixel to be textured; calculates a comprehensive weight of the central pixel to be textured and the local pixel to be textured according to the size of the similarity weight and the texture similarity; calculates a texture restoration pixel of the central pixel to be textured according to the comprehensive weight and the low-frequency illumination component of the local pixel to be textured; and repeatedly selects each pixel from all complete denoised pixels and calculates the corresponding texture restoration pixel until all complete denoised pixels are processed.

2. The intelligent water gauge identification system based on big data technology according to claim 1 is characterized in that: The data acquisition and calculation module (100) comprises a data acquisition unit (101), wherein the data acquisition unit (101) acquires a water gauge image using an image acquisition device, and the acquired water gauge image is represented as ,in are the pixel coordinates in the water gauge image.

3. The intelligent water gauge identification system based on big data technology according to claim 2 is characterized in that: The data acquisition and calculation module (100) comprises a data calculation unit (102), wherein the data calculation unit (102) uses a Gaussian filter to convert the water gauge image Decomposition into low-frequency illumination components and high frequency detail components , where the Gaussian filter is a low-pass filter used to smooth the image and remove high-frequency noise. The decomposition process is as follows: Choose a Gaussian kernel of fixed size ; Gaussian kernel Image with water gauge Perform convolution operations on each pixel and its neighborhood in the image to calculate the low-frequency illumination component of each pixel. ; Image via Water Gauge Subtract low-frequency lighting components , calculate the high-frequency detail component of each pixel .

4. The intelligent water gauge identification system based on big data technology according to claim 3 is characterized in that: The process of generating a complete denoised pixel by the denoising processing module (200) is as follows: From the water gauge image Select a pixel As the central pixel to be denoised; At the center pixel to be denoised Neighborhood Select a pixel As the local pixel to be denoised, the neighborhood Defined as a A square area with a center; Calculate the center pixel to be denoised and the local pixel to be denoised The similarity weight between , the calculation formula is as follows: ,in and is the center pixel to be denoised and the local pixels to be denoised The low-frequency illumination component of is the similarity threshold parameter; where the similarity weight The center pixel to be denoised is measured and the local pixels to be denoised The similarity between pixels is calculated and the influence of noise on the central pixel is reduced by weighted averaging of similar pixels in the neighborhood. Based on similarity weight and the local pixels to be denoised The low-frequency illumination component , calculate the center pixel to be denoised Denoised pixels , the calculation formula is as follows: ； The center pixel to be denoised Denoised pixels and the center pixel to be denoised The high-frequency detail components Combine and calculate the center pixel to be denoised Complete denoised pixels , the calculation formula is as follows: ； Repeat the selection of the water gauge image For each pixel in the image, the corresponding complete denoised pixel value is calculated until all pixels are processed.

5. The intelligent water gauge identification system based on big data technology according to claim 4 is characterized in that: The process of the texture restoration module (300) generating texture restoration pixels is as follows: Select one pixel among all the complete denoised pixels as the center pixel to be textured , and select the center pixel to be textured Neighborhood A pixel within is used as the local pixel to be textured , calculate the center pixel of the texture The local pixel to be textured The similarity weight between And texture similarity, where the calculation process of texture similarity is as follows: Extract the center pixel of the texture and the local pixel to be textured The texture similarity vector and ; Computing texture similarity , ,in is the texture similarity threshold parameter; Combined similarity weights and texture similarity , calculate the comprehensive weight , where the calculation formula is as follows: ； According to the comprehensive weight And the local pixels to be textured The low-frequency illumination component , calculate the center pixel of the texture Texture repair pixels , the calculation formula is as follows: ； Repeatedly select each pixel in all complete denoised pixels and calculate the corresponding texture restoration pixel until all complete denoised pixels are processed.

Images