River channel equidistant line pixel resolution evaluation method based on image recognition
Through the river channel isometric cell resolution evaluation method based on image recognition, the problem of difficult data quality and difficult to determine the river channel isometric cell length is solved, and high-accurate cell resolution evaluation is achieved, and the monitoring results are improved. reliability.
Patent Information
- Application Number
- CN202411225340.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-12-27
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is difficult to ensure data quality in image data processing, resulting in a decrease in the accuracy of cell resolution evaluation, especially when the true length of the equidistant line of the river channel is difficult to measure, which affects the application effect of image recognition.
Through the river isometric line cell resolution evaluation method based on image recognition, including image quality evaluation, river isometric line generation and cell resolution evaluation, the grayscale value, signal-to-noise ratio and water level information of remote sensing image data are generated. The river channel isometric and evaluates cell resolution to ensure data quality and accuracy.
It improves the accuracy and accuracy of the equidistant line monitoring data of the river channel, reduces errors caused by image quality problems, and can reflect the changes in the cell resolution of the river channel under different water levels in real time, enhancing the reliability of the monitoring results.
Smart Images
Figure CN119205647A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image data processing, and in particular to a method for evaluating the pixel resolution of river channel equidistant lines based on image recognition. Background Art
[0002] Currently, in the field of pixel resolution processing technology, image recognition technology is usually used to analyze the pixel resolution of river equidistant lines. However, this technology is difficult to guarantee the data quality in the process of acquiring image data. The acquisition process is easily interfered by various factors, resulting in a decrease in image quality, which in turn affects the accurate assessment of pixel resolution. At the same time, traditional pixel resolution assessment methods require specific information about river equidistant lines, but it is often difficult to measure their true length in the center of the river.
[0003] For example, the invention patent with the announcement number CN111127319B provides a method for calculating the ground pixel resolution of push-broom imaging in motion. In view of the influence of complex postures of push-broom imaging in motion, a ground pixel resolution calculation model suitable for push-broom imaging in motion is established, and a ground pixel resolution calculation method for calculating the ground pixel edge vector from the center of the earth to the ground pixel in an orbital coordinate system is proposed; the ground pixel resolution calculation method of the above application method can calculate the ground pixel resolution of all pixels in the entire process of push-broom imaging in motion and in the field of view, can be used to analyze the resolution changes of a single pixel during the imaging process in motion and the ground pixel resolution changes of all pixels in the entire field of view under any complex posture, and serve as a basis for evaluating the distortion of remote sensing images of push-broom imaging satellites in motion.
[0004] Combined with the above technical solutions, it is found that with the continuous development of image processing technology, it is necessary to organize more efficient, accurate and convenient pixel resolution analysis technical solutions. However, most pixel resolution analysis technologies are weak in judging the data quality during the evaluation and calculation of pixel resolution in image data. On the other hand, the methods used in pixel resolution analysis are relatively complex, which will lead to reduced accuracy and large errors in the process of pixel resolution evaluation and calculation, thus affecting the application of image recognition. Summary of the invention
[0005] In view of the deficiencies in the prior art, the present invention provides a method for evaluating pixel resolution of river equidistant lines based on image recognition, which can effectively solve the problems involved in the above-mentioned background technology.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for evaluating pixel resolution of river channel equidistant lines based on image recognition, comprising: S1. Image quality evaluation: performing image recognition on the river channel area, thereby obtaining remote sensing image data of the river channel area, evaluating the image quality index of the river channel area, and comparing it with a predefined image quality index threshold. If the image quality index of the river channel area is greater than or equal to the predefined image quality index threshold, then entering S2; S2. Generation of river channel equidistant lines: locating the pixel key points of the target river channel based on the remote sensing image data of the river channel area, connecting the pixel key points of the target river channel in a predetermined manner, and generating a target river channel. The equidistant lines of each channel are determined to determine the resolution evaluation data of each channel equidistant line of the target river; S3. Pixel resolution evaluation: According to the remote sensing image data of the river area, the resolution evaluation point of the target river is located, and the resolution evaluation data of the resolution evaluation point of the target river is obtained, and the resolution evaluation data are comprehensively processed with the resolution evaluation data of each channel equidistant line of the target river, so as to evaluate the pixel resolution corresponding to the resolution evaluation point of the target river; S4. Feedback on evaluation results: The pixel resolution corresponding to the resolution evaluation point of the target river is verified with the predefined measured pixel resolution, so as to provide feedback on the recognition process of the pixel resolution of the equidistant line of the target river.
[0007] As a further method, the remote sensing image data of the river area specifically includes the grayscale value of each pixel point of the remote sensing image of the river area, the spatial resolution mean of the remote sensing image of the river area, the useful signal intensity value of each pixel point of the remote sensing image of the river area, and the noise signal intensity value of each pixel point of the remote sensing image of the river area.
[0008] As a further method, the image quality index of the river area is evaluated, and the specific evaluation process is as follows: according to the grayscale value of each pixel of the remote sensing image of the river area, the grayscale value of each pixel of the remote sensing image of the river area in the first preset direction and the second preset direction is extracted, and the average gradient of the remote sensing image of the river area is obtained by comprehensive processing; the useful signal intensity value of each pixel of the remote sensing image of the river area and the noise signal intensity value of each pixel of the remote sensing image of the river area are comprehensively processed to obtain the signal-to-noise ratio of the remote sensing image of the river area; the average gradient of the remote sensing image of the river area, the signal-to-noise ratio of the remote sensing image of the river area, and the spatial resolution mean of the remote sensing image of the river area are comprehensively analyzed to obtain the image quality index of the river area, and the specific expression is:
[0009]
[0010] Where ZL is the image quality index of the river area, TD is the average gradient of the remote sensing image of the river area, XZ is the signal-to-noise ratio of the remote sensing image of the river area, FB is the mean spatial resolution of the remote sensing image of the river area, Y1 is the impact factor corresponding to the gradient unit value predefined in the river data cloud platform, Y2 is the impact factor corresponding to the signal-to-noise ratio unit value predefined in the river data cloud platform, Y3 is the impact factor corresponding to the resolution unit value predefined in the river data cloud platform, and e is a natural constant.
[0011] As a further method, the image quality index of the river channel area is compared with a predefined image quality index threshold. The specific comparison process is: compare the image quality index of the river channel area with the predefined image quality index threshold. If the image quality index of the river channel area is greater than or equal to the image quality index threshold, locate the key points of each pixel of the target river channel. If the image quality index of the river channel area is less than the image quality index threshold, re-recognize the image of the river channel area until the image quality index of the river channel area is greater than or equal to the image quality index threshold.
[0012] As a further method, the specific generation process of the channel equidistant lines of the target river is as follows: the pixel key points of the target river are numbered in a clockwise direction, and the required pixel key points are pixel key point P1, pixel key point P2, pixel key point P4, pixel key point P5, pixel key point P6, and pixel key point P8; after the pixel key point P1 is connected with the pixel key point P4 in a predetermined manner, the generated channel equidistant line is marked as the first channel equidistant straight line L1 of the target river; after the pixel key point P5 is connected with the pixel key point P8 in a predetermined manner, the generated channel equidistant line is marked as the first channel equidistant straight line L1 of the target river. line, marked as the second channel equidistant line L2 of the target channel; after the pixel key point P1 and the pixel key point P2 are connected in a predetermined manner, the channel equidistant line generated is marked as the third channel equidistant line L3 of the target channel; after the pixel key point P5 and the pixel key point P6 are connected in a predetermined manner, the channel equidistant line generated is marked as the fourth channel equidistant line L4 of the target channel; in this way, the channel equidistant lines of the target channel are generated, including the first channel equidistant line of the target river channel, the second channel equidistant line of the target river channel, the third channel equidistant line of the target river channel, and the fourth channel equidistant line of the target river channel.
[0013] As a further method, the resolution evaluation data of each channel equidistant line of the target river channel is determined, and the specific determination process is as follows: the resolution evaluation data of each channel equidistant line of the target river channel specifically includes the intercept, slope, and water level of each channel equidistant line of the target river channel; according to the first channel equidistant line L1 of the target river channel, the slope of the first channel equidistant line L1 of the target river channel is obtained by slope calculation, and is recorded as the slope K1 of the first channel equidistant line of the target river channel; according to the second channel equidistant line L2 of the target river channel, the slope of the second channel equidistant line L2 of the target river channel is obtained by slope calculation, and is recorded as the slope K2 of the second channel equidistant line of the target river channel; based on According to the first channel equidistant straight line L1 of the target river channel, the intercept of the first channel equidistant straight line L1 of the target river channel is obtained by intercept calculation, which is recorded as the intercept b1 of the first channel equidistant straight line of the target river channel; according to the second channel equidistant straight line L2 of the target river channel, the intercept of the second channel equidistant straight line L2 of the target river channel is obtained by intercept calculation, which is recorded as the intercept b2 of the second channel equidistant straight line of the target river channel; the height value between the third channel equidistant straight line L3 of the target river channel and the water level reference plane is obtained, which is recorded as the first water level sw1 of the target river channel, and the height value between the fourth channel equidistant straight line L4 of the target river channel and the water level reference plane is obtained, which is recorded as the second water level sw2 of the target river channel.
[0014] As a further method, the pixel resolution corresponding to the resolution point to be evaluated of the target river channel is evaluated in the following specific process: the slope corresponding to the resolution point to be evaluated of the target river channel, the intercept corresponding to the resolution point to be evaluated of the target river channel, the water level corresponding to the resolution point to be evaluated of the target river channel, and the intercept, slope and water level of each equidistant line of the target river channel are comprehensively processed to evaluate the pixel resolution corresponding to the resolution point to be evaluated of the target river channel.
[0015] As a further method, the identification process of the pixel resolution of the target river channel equidistant line is fed back. The specific feedback process is: the pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is verified with the predefined measured pixel resolution to generate a resolution evaluation label. If the pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is equal to the measured pixel resolution, the resolution evaluation label is defined as a resolution evaluation qualified label. If the pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is greater than or less than the measured pixel resolution, the resolution evaluation label is defined as a resolution evaluation abnormal label, thereby providing feedback on the identification process of the pixel resolution of the target river channel equidistant line.
[0016] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:
[0017] (1) The present invention provides a method for evaluating pixel resolution of river channel equidistant lines based on image recognition. First, the image quality index of the river channel area is evaluated. If it is greater than or equal to a predefined image quality index threshold, it means that the image quality has reached the pre-set image quality standard. Then, the pixel key points of the target river channel are located, and the channel equidistant lines of the target river channel are generated. The resolution evaluation data of the channel equidistant lines of the target river channel are determined, which is used to quantify and evaluate the fineness and accuracy of the river channel image data. According to the resolution evaluation point of the target river channel, the pixel resolution corresponding to the resolution evaluation point of the target river channel is evaluated, and the recognition process of the pixel resolution of the target river channel equidistant lines is fed back. By identifying the pixel resolution, it can be ensured that the monitoring data of the channel equidistant lines is more accurate, thereby improving the accuracy of the monitoring results.
[0018] (2) The present invention performs image recognition on the river area, thereby obtaining remote sensing image data of the river area and evaluating the image quality index of the river area, thereby ensuring that the quality of the acquired remote sensing image data meets the analysis requirements. High-quality image data can more accurately reflect the actual situation of the river area, reduce errors caused by image quality problems, and provide support for subsequent analysis of pixel resolution of the river area.
[0019] (3) The present invention solves the difficulty of measuring the true length of the river in the middle of the river in the traditional method by using the water level and the river equidistant lines to obtain pixel resolution. The water level data in the river area is usually obtained in real time. Combined with the river equidistant lines, the pixel resolution changes of the river at different water levels can be reflected in real time. By combining the water level and the river equidistant lines, the pixel resolution of any point in the river area can be calculated more accurately, which helps to reduce the deviation of monitoring results caused by resolution error. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.
[0021] Figure 1 It is a schematic diagram of the method flow of the present invention.
[0022] Figure 2 , Figure 3 All of them are plan schematic diagrams of the target river channel involved in the present invention.
[0023] Figure numerals: 1. target river channel; 2. pixel key point P1; 3. pixel key point P2; 4. pixel key point P4; 5. pixel key point P5; 6. pixel key point P6; 7. pixel key point P8; 8. resolution evaluation point of the target river channel; 9. first channel equidistant straight line L1 of the target river channel; 10. second channel equidistant straight line L2 of the target river channel; 11. third channel equidistant straight line L3 of the target river channel; 12. fourth channel equidistant straight line L4 of the target river channel; 13. first water level sw1 of the target river channel; 14. second water level sw2 of the target river channel. DETAILED DESCRIPTION
[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of 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.
[0025] Reference Figure 1 As shown, the present invention provides a method for evaluating pixel resolution of river equidistant lines based on image recognition, including: S1. Image quality evaluation: performing image recognition on the river area, thereby obtaining remote sensing image data of the river area, evaluating the image quality index of the river area, and comparing it with a predefined image quality index threshold. If the image quality index of the river area is greater than or equal to the predefined image quality index threshold, enter S2.
[0026] The above-mentioned equipment for image recognition of river areas is drone recognition equipment. Drones can quickly cover large areas of river areas and are more efficient than traditional manual inspections or ground vehicle inspections. This helps to complete the monitoring task of the entire river area in a shorter time; at the same time, the high-definition cameras and sensors carried by drones, combined with advanced image recognition algorithms, can accurately identify the overall status of the river area, improving the accuracy and reliability of recognition.
[0027] Specifically, the remote sensing image data of the river area includes the grayscale value of each pixel point of the remote sensing image of the river area, the mean spatial resolution of the remote sensing image of the river area, the useful signal intensity value of each pixel point of the remote sensing image of the river area, and the noise signal intensity value of each pixel point of the remote sensing image of the river area.
[0028] The above-mentioned pixel points are the smallest units for the drone sensor to scan and sample the ground scenes during the remote sensing data collection process, and are the basic units that constitute the remote sensing digital image.
[0029] The grayscale value of each pixel point mentioned above can be converted into a grayscale image by image processing software, so that the grayscale value of each pixel point can be viewed in the image processing software; the resolution, useful signal strength value and noise signal strength value can all be obtained by viewing the drone's identification report.
[0030] Furthermore, the image quality index of the river area is evaluated, and the specific evaluation process is as follows:
[0031] According to the grayscale value of each pixel of the remote sensing image of the river region, the grayscale value of each pixel of the remote sensing image of the river region in a first preset direction and a second preset direction is extracted, wherein the first preset direction is the east-west direction and the second preset direction is the north-south direction, and the grayscale value difference processing is performed on each pixel of the remote sensing image of the river region in the east-west direction and the north-south direction to obtain the grayscale difference of each pixel of the remote sensing image of the river region in the east-west direction and the grayscale difference in the north-south direction, and the average gradient of the remote sensing image of the river region is obtained through comprehensive processing.
[0032] In this embodiment, setting the grayscale value of the preset direction does not refer to a single grayscale value of a specific pixel, but is to indicate the changing trend or characteristics of the grayscale values in these directions. The grayscale values in the horizontal and vertical directions are still described based on the grayscale values of each pixel. In this embodiment, the focus is on how these pixels are related to and change in their directions (horizontal or vertical).
[0033] The specific processing process of the average gradient of the above river area remote sensing image is as follows:
[0034] The remote sensing image of the river channel area is marked as I, and its size is M×N, that is, there are M rows of pixels, representing the number of pixels in the north-south direction of the image, that is, from the top to the bottom of the image, there are M pixels arranged in rows; N columns, representing the number of pixels in the east-west direction of the image, that is, from the left edge to the right edge of the image, there are N pixels arranged in columns; take the intersection of M rows and N columns at the left edge as the starting point, for each pixel I(x,y) in the image, where 1≤x≤M, 1≤y≤N, the horizontal direction G of each pixel of the remote sensing image of the river channel area can be calculated. x and vertical direction G y Grayscale difference: G x (x,y)=I(x+1,y)-I(x,y),G y (x, y) = I (x, y + 1) - I (x, y). It should be explained that when x = M or y = N on the boundary, the grayscale difference cannot be calculated directly because it exceeds the image boundary, so the boundary grayscale difference is negligible.
[0035] According to the grayscale difference in the horizontal and vertical directions, the gradient amplitude G(x,y) of each pixel can be calculated and approximated by the Euclidean distance formula:
[0036] Finally, calculate the average value of the gradient amplitude of all pixels, that is, the average gradient:
[0037]
[0038] The useful signal intensity value of each pixel point of the river area remote sensing image and the noise signal intensity value of each pixel point of the river area remote sensing image are comprehensively processed to obtain the signal-to-noise ratio of the river area remote sensing image.
[0039] The signal-to-noise ratio of the above-mentioned river area remote sensing image is the average signal-to-noise ratio of the river area remote sensing image. The comprehensive processing process is: ratio processing is performed on the useful signal intensity value of each pixel point of the river area remote sensing image and the noise signal intensity value of each pixel point of the river area remote sensing image to obtain the signal-to-noise ratio of each pixel point of the river area remote sensing image, and the obtained value is marked as the signal-to-noise ratio of the river area remote sensing image through mean processing.
[0040] The signal-to-noise ratio of the above river area remote sensing image is shown in Table 1:
[0041] Table 1 Numerical change relationship table of signal-to-noise ratio
[0042] serial number 1 2 3 4 Useful signal strength value / dB 25 30 27 13 Noise signal strength value / dB 4 14 20 8 Signal-to-Noise Ratio 6.25 2.14 1.35 1.625
[0043] In this embodiment, according to the table of specific values of the signal-to-noise ratio of the river area remote sensing image mentioned above, the useful signal strength value of the river area remote sensing image and the noise signal strength value of the river area remote sensing image fluctuate greatly, resulting in a large difference in their signal-to-noise ratio. This is because the acquisition of signal strength values is easily affected by many factors, including atmospheric conditions, topography, water body characteristics, etc. The specific value of the signal-to-noise ratio of the river area remote sensing image in this embodiment is 2.84.
[0044] In this embodiment, it is obtained by comprehensive analysis based on the average gradient of the remote sensing image of the river area, the signal-to-noise ratio of the remote sensing image of the river area, and the mean spatial resolution of the remote sensing image of the river area, so as to determine the image quality index of the river area. In this embodiment, a more accurate calculation method is used to obtain it, and the specific expression is:
[0045]
[0046] Wherein ZL is the image quality index of the river area, and in this embodiment, it is used to evaluate the clarity, information integrity and other aspects of the remote sensing image of the river area.
[0047] TD is the average gradient of the remote sensing image of the river area, which is used to represent the rate of change of pixel values in the image. In grayscale images, it can be used to measure the edge strength or texture complexity of the image.
[0048] XZ is the signal-to-noise ratio of the remote sensing image of the river area, which refers to the proportional relationship between the useful signal of the image and the image noise signal, that is, the ratio of signal intensity to noise intensity. In the remote sensing image, the useful signal refers to the effective information of the river area that you want to obtain, such as river morphology, water quality, etc.; the noise signal refers to the interference information that you do not want to exist in the image, such as sensor noise, atmospheric interference, ground reflection, etc., which cause the image quality to deteriorate.
[0049] FB is the mean spatial resolution of remote sensing images of river areas, which refers to the size of the ground range represented by the pixels in the remote sensing image, that is, the instantaneous field of view of the scanner or the smallest unit that can be distinguished for ground objects. In remote sensing images of river areas, the level of spatial resolution directly affects the degree of river details that can be seen in the image.
[0050] Y1 is the impact factor corresponding to the gradient unit value predefined in the river data cloud platform.
[0051] In a specific embodiment, the influencing factor corresponding to the gradient unit value can be directly obtained from the river data cloud platform, and the corresponding relationship can be a pre-set mapping relationship. For example, the gradient and the influencing factor corresponding to the gradient unit value preset in the river data cloud platform form a mapping set, and the real-time gradient mapping set is used to obtain the influencing factor corresponding to the gradient unit value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, the value range is [0, 1].
[0052] Y2 is the impact factor corresponding to the signal-to-noise ratio unit value predefined in the river data cloud platform.
[0053] In a specific embodiment, the impact factor corresponding to the signal-to-noise ratio unit value can be directly obtained from the river data cloud platform, and the corresponding relationship can be a preset mapping relationship. For example, the signal-to-noise ratio and the impact factor corresponding to the signal-to-noise ratio unit value preset in the river data cloud platform form a mapping set, and the real-time signal-to-noise ratio mapping set is used to obtain the impact factor corresponding to the signal-to-noise ratio unit value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is [0, 1].
[0054] Y3 is the impact factor corresponding to the resolution unit value predefined in the river data cloud platform, and e is a natural constant.
[0055] In a specific embodiment, the influencing factor corresponding to the resolution unit value can be directly obtained from the river data cloud platform, and the corresponding relationship can be a pre-set mapping relationship. For example, the resolution and the influencing factor corresponding to the preset resolution unit value in the river data cloud platform form a mapping set, and the real-time resolution mapping set is used to obtain the influencing factor corresponding to the resolution unit value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is [0, 1].
[0056] In this embodiment, the average gradient is an important indicator for measuring the clarity of an image. The larger the average gradient, the more image layers, the richer the details, the clearer the image, and the better the image quality. An image with a high average gradient often has a lower noise level, that is, a higher signal-to-noise ratio, because noise interferes with grayscale changes in the image. By reducing the average gradient value, the less noise in the image, the more prominent the effective information, and the better the image quality. Remote sensing images with high spatial resolution often require a higher signal-to-noise ratio to ensure image quality, because as the spatial resolution increases, the detailed information in the image becomes richer, so effective measures are taken to improve the signal-to-noise ratio, thereby increasing the spatial resolution. In summary, images with high average gradient, high signal-to-noise ratio, and high spatial resolution can more clearly display the geographical features and dynamic changes of the river area, and provide more accurate and reliable information support for subsequent analysis of pixel resolution.
[0057] Specifically, the image quality index of the river area is compared with a predefined image quality index threshold, and the specific comparison process is as follows:
[0058] The image quality index of the river area is compared with the image quality index threshold predefined in the river data cloud platform. If the image quality index of the river area is greater than or equal to the image quality index threshold, it means that the image quality of the river area has reached the preset quality standard and can better reflect the actual situation of the river area. The key points of each pixel of the target river are located. If the image quality index of the river area is less than the image quality index threshold, it means that the image quality of the river area has not reached the preset standard and it is difficult to accurately reflect the actual situation of the river area, which increases the difficulty of subsequent image analysis and processing. The image recognition of the river area is then performed again until the image quality index of the river area is greater than or equal to the image quality index threshold.
[0059] The data stored in the above-mentioned river data cloud platform is obtained by averaging the historical image data of the river area based on statistics of multiple times. For example, the image quality index threshold is obtained by obtaining the historical image quality indicators of the river area for multiple times, and the surveying personnel determine that the index that meets the best image analysis quality under different quality indicators is averaged and used as the image quality index threshold. In this way, the value in the river data cloud platform is obtained, and the value may change due to different river environment, actual needs, and the status of the acquisition equipment. This embodiment does not make any special limitation on this.
[0060] It should be explained that if the image quality index of the river area is always less than the image quality index threshold, it means that the image quality taken in the river area continues to fail to meet the preset standard, and there are problems such as insufficient clarity, low contrast, poor resolution or excessive noise. You can re-select a drone camera or sensor with higher resolution and lower noise for shooting; choose a more appropriate shooting time and angle, and avoid shooting under adverse conditions such as haze, rain or strong direct sunlight.
[0061] S2. Generation of river channel equidistant lines: Based on the remote sensing image data of the river channel area, the pixel key points of the target river channel are located, and the pixel key points of the target river channel are connected in a predetermined manner to generate the channel equidistant lines of the target river channel, and the resolution evaluation data of the channel equidistant lines of the target river channel are determined.
[0062] The above-mentioned positioning method for locating the pixel key points of the target river channel is that the evaluators randomly arrange the pixel key points to ensure that the selection of pixel key points is not affected by the preset mode, thereby increasing the diversity and representativeness of the data. At the same time, by randomly selecting a part of the key points for processing, the processing efficiency can be improved while ensuring a certain accuracy.
[0063] Furthermore, the specific generation process of generating the equidistant lines of each channel of the target channel is as follows:
[0064] The pixel key points of the target river are numbered in a clockwise direction. The required pixel key points are pixel key points P1, such as Figure 2 As shown in 2, the pixel key point P2 is Figure 2 As shown in 3, the pixel key point P4 is Figure 2 As shown in 4, the pixel key point P5 is Figure 2 As shown in 5, the pixel key point P6 is Figure 2 As shown in 6, the pixel key point P8 is Figure 2 As shown in 7.
[0065] According to the key points of each pixel of the target river, a plane schematic diagram of the target river is constructed, such as Figure 2 , Figure 3As shown, according to the plan schematic diagram of the target river channel, the predetermined method of connecting the straight lines set in this embodiment is that the distance is equal and the straight lines are parallel. For example, the straight line formed by connecting the pixel key point P1 and the pixel key point P4 and the straight line formed by connecting the pixel key point P5 and the pixel key point P8 have the same distance and are parallel, and the requirements for the remaining point groups are the same; and in the embodiment of the present invention, only the pixel resolution on the left straight lines L1 and L2 is analyzed, and the analysis method of the pixel resolution on the right straight line is the same as the analysis method on the left, and no excessive elaboration is made in this embodiment.
[0066] It should be explained that the target river embankment is an inclined surface. Figure 2 , Figure 3 This is a plan view of the target river channel. The figure formed by the connected straight lines shows a trend of being larger when near and smaller when far.
[0067] After the pixel key point P1 and the pixel key point P4 are connected in a predetermined manner, the generated river channel equidistant line is marked as the first river channel equidistant line L1 of the target river channel, as shown in FIG. Figure 3 As shown in Figure 9.
[0068] After the pixel key point P5 and the pixel key point P8 are connected in a predetermined manner, the generated river channel equidistant line is marked as the second river channel equidistant line L2 of the target river channel, as shown in FIG. Figure 3 As shown in Figure 10.
[0069] After the pixel key point P1 and the pixel key point P2 are connected in a predetermined manner, the generated river channel equidistant line is marked as the third river channel equidistant line L3 of the target river channel, as shown in FIG. Figure 3 As shown in Figure 11.
[0070] After the pixel key point P5 and the pixel key point P6 are connected in a predetermined manner, the generated river channel equidistant line is marked as the fourth river channel equidistant line L4 of the target river channel, as shown in FIG. Figure 3 As shown in Figure 12.
[0071] In this way, the channel equidistant lines of the target channel are generated, including the first channel equidistant straight line of the target channel, the second channel equidistant straight line of the target channel, the third channel equidistant straight line of the target channel, and the fourth channel equidistant straight line of the target channel.
[0072] Specifically, the resolution evaluation data of each channel equidistant line of the target channel is determined by the following specific determination process:
[0073] The resolution assessment data of each channel equidistant line of the target river channel specifically includes the intercept, slope, and water level of each channel equidistant line of the target river channel.
[0074] It should be explained that the resolution assessment data of each channel equidistant line of the target river channel include but are not limited to the intercept, slope and water level of each channel equidistant line of the target river channel. In this embodiment, only the intercept, slope and water level are analyzed. In the analysis of channel equidistant lines, the intercept usually reflects the offset of the data under certain benchmark conditions, and the slope represents the rate at which these parameters change with space or time. In the analysis of channel equidistant lines, the change in slope may reflect the trend of water flow velocity, water level rise or fall, and changes in river morphology; changes in water level directly reflect the flow rate, water level and possible hydrological events of the river channel, and will also directly affect the intercept and slope of the channel equidistant line. For example, when the water level rises, the intercept of the equidistant line may increase, while the slope may change due to multiple factors such as water flow velocity and riverbed morphology. In summary, by comprehensively analyzing the intercept, slope and water level, more detailed and reliable information such as river morphology and water flow dynamics is provided for pixel resolution assessment.
[0075] According to the first channel equidistant straight line L1 of the target channel, the slope of the first channel equidistant straight line L1 of the target channel is obtained by slope calculation, which is recorded as the slope K1 of the first channel equidistant straight line of the target channel.
[0076] The specific process of the above slope calculation is: according to the remote sensing image data of the river area, the coordinates of the pixel key points P1 and P4 are obtained in the drone identification report, P1 (x1, y1), P4 (x2, y2), and the coordinates of the two points are substituted into the slope calculation formula: In this way, the slope K1 of the first channel equidistant straight line of the target channel is obtained.
[0077] It should be explained that the coordinates involved in the above-mentioned pixel key points are based on the vertical line where the pixel key point P5 is located as the Z axis, the intersection of the vertical line where the pixel key point P5 is located and the horizontal plane as the origin O, the east-west direction of the intersection as the X axis, and the north-south direction as the Y axis to establish a three-dimensional rectangular coordinate system. The Z axis is only used to obtain the specific values of the first water level sw1 of the target river channel and the second water level sw2 of the target river channel. The XOY plane is used to obtain the coordinates of the pixel key points involved in the embodiment of the present invention, and the specific values of the coordinates are the projection values of the first river channel equidistant straight line L1 and the second river channel equidistant straight line L2 on the XOY plane.
[0078] According to the second channel equidistant straight line L2 of the target channel, the slope of the second channel equidistant straight line L2 of the target channel is obtained by slope calculation, which is recorded as the slope K2 of the second channel equidistant straight line of the target channel.
[0079] The calculation process of the slope K2 is the same as that of the slope K1, and will not be described in detail in this embodiment.
[0080] According to the first channel equidistant straight line L1 of the target channel, the intercept of the first channel equidistant straight line L1 of the target channel is obtained by intercept calculation, which is recorded as the intercept b1 of the first channel equidistant straight line of the target channel.
[0081] The specific process of the above intercept calculation is: substitute the coordinates of the above pixel key point P1 and the slope K1 into the straight line equation y=k1*x+b to obtain the value of the intercept b1.
[0082] According to the second channel equidistant straight line L2 of the target river channel, the intercept of the second channel equidistant straight line L2 of the target river channel is obtained by intercept calculation, which is recorded as the intercept b2 of the second channel equidistant straight line of the target river channel.
[0083] The calculation process of the intercept b2 is the same as that of the intercept b1, and will not be described in detail in this embodiment.
[0084] The height value between the third channel equidistant line L3 of the target channel and the water level reference plane is obtained and recorded as the first water level sw1 of the target channel, such as Figure 3 As shown in FIG13, the height value between the fourth channel equidistant straight line L4 of the target channel and the water level reference plane is obtained, which is recorded as the second water level sw2 of the target channel. Figure 3 As shown in Figure 14.
[0085] The height value between the third channel equidistant straight line L3 of the above-mentioned target river channel and the water level reference plane, and the height value between the fourth channel equidistant straight line L4 of the target river channel and the water level reference plane can be obtained through remote sensing images, wherein the water level reference plane is the starting surface for calculating the water level, which is used to determine the reference height of the free water surface of the water body relative to each straight line.
[0086] S3. Pixel resolution assessment: Based on the remote sensing image data of the river area, locate the resolution points to be assessed in the target river, obtain the resolution assessment data of the resolution points to be assessed in the target river, and conduct a comprehensive processing with the resolution assessment data of each equidistant line of the target river to assess the pixel resolution corresponding to the resolution points to be assessed in the target river.
[0087] The above-mentioned positioning method of the resolution points to be evaluated for locating the target river is also randomly arranged by the evaluators, such as Figure 2 As shown in Figure 8.
[0088] Furthermore, the resolution assessment data of the target river channel resolution point to be assessed specifically includes the slope corresponding to the target river channel resolution point to be assessed, the intercept corresponding to the target river channel resolution point to be assessed, and the water level corresponding to the target river channel resolution point to be assessed.
[0089] The water level corresponding to the above-mentioned target river channel resolution assessment point is obtained by real-time measurement through a radar water level gauge.
[0090] The slope corresponding to the target channel resolution to be evaluated point and the intercept corresponding to the target channel resolution to be evaluated point are calculated by the water level corresponding to the target channel resolution to be evaluated point. The specific calculation process satisfies the expression:
[0091] Specifically, the resolution of the target river channel to be evaluated is the pixel resolution corresponding to the point to be evaluated. The specific evaluation process is as follows:
[0092] In this embodiment, the target river resolution is obtained by comprehensively analyzing the slope corresponding to the target river resolution to be evaluated point, the intercept corresponding to the target river resolution to be evaluated point, the water level corresponding to the target river resolution to be evaluated point, and the intercept, slope, and water level of each river equidistant line of the target river, so as to determine the pixel resolution corresponding to the target river resolution to be evaluated point. In this embodiment, a more accurate calculation method is used to obtain it, and the specific expression is:
[0093]
[0094] Where R is the resolution of the target river channel. The pixel resolution corresponding to the point to be evaluated refers to the actual ground area represented by each pixel in the image at the evaluation point. The pixel resolution determines the minimum size of the ground target that can be distinguished by the image.
[0095] k is the slope corresponding to the target river resolution point to be evaluated, which refers to the degree of inclination of the equidistant line of the river at that location as the spatial position changes when analyzing the point to be evaluated.
[0096] k1 is the slope of the first channel equidistant straight line of the target river channel, which refers to the degree of inclination of the first channel equidistant straight line of the target river channel as the spatial position changes.
[0097] k2 is the slope of the second channel equidistant straight line of the target river channel, which refers to the degree of inclination of the second channel equidistant straight line of the target river channel as the spatial position changes.
[0098] b is the intercept corresponding to the target river resolution point to be evaluated, which refers to the specific length of the river equidistant line at that location obtained when analyzing the point to be evaluated.
[0099] b1 is the intercept of the first channel equidistant straight line of the target river channel, which refers to the specific length of the first channel equidistant straight line of the target river channel.
[0100] b2 is the intercept of the second channel equidistant straight line of the target river channel, which refers to the specific length of the second channel equidistant straight line of the target river channel.
[0101] sw is the water level corresponding to the target river resolution point to be evaluated, which refers to the height value between the river equidistant line and the water level reference surface at the location obtained when analyzing the point to be evaluated.
[0102] sw1 is the first water level of the target river channel, which refers to the height value between the third channel equidistant straight line L3 of the target river channel and the water level reference plane.
[0103] sw2 is the second water level of the target river channel, which refers to the height value between the fourth channel equidistant straight line L4 of the target river channel and the water level reference plane.
[0104] It can be seen from the above expression that in this embodiment, the pixel resolution corresponding to the resolution point to be evaluated of the target river channel can be obtained through only one set of data, and multiple sets of data are used for calculation to ensure that each parameter can accurately correspond to the proportion in the actual parameter during data processing to avoid calculation deviation; for example, the slope corresponding to the target river channel resolution point to be evaluated and the water level corresponding to the target river channel resolution point to be evaluated are to ensure that the above expression calculates the water level and slope at the correct position. The complete calculation using all parameters can make the calculation result of the above expression based on an accurate and consistent relationship, and can also verify the correctness of the process of obtaining each parameter. In summary, the result calculated by one set of data cannot guarantee that its calculation result is accurate, but the relationship between multiple sets of data is utilized to ensure accuracy in the calculation through complete parameters.
[0105] S4. Feedback on evaluation results: The pixel resolution corresponding to the target river channel’s evaluation point is verified with the predefined measured pixel resolution, so as to provide feedback on the recognition process of the pixel resolution of the target river channel’s equidistant line.
[0106] Specifically, the recognition process of the pixel resolution of the target river channel equidistant line is fed back, and the specific feedback process is:
[0107] The pixel resolution corresponding to the resolution point to be evaluated of the target river channel is verified with the measured pixel resolution predefined in the river channel data cloud platform to generate a resolution evaluation label. If the pixel resolution corresponding to the resolution point to be evaluated of the target river channel is equal to the measured pixel resolution, the resolution evaluation label is defined as a qualified resolution evaluation label, which means that during the evaluation process, the degree of detail of the image data of the evaluation point observed matches the degree of detail specified in the standard, and the analysis of the pixel resolution has achieved the expected effect. The label may include the location information, evaluation time, and evaluation results of the point to be evaluated. If the pixel resolution corresponding to the resolution point to be evaluated of the target river channel is greater than or less than the measured pixel resolution, the resolution evaluation label is defined as a resolution evaluation abnormal label, so as to provide feedback on the recognition process of the pixel resolution of the target river channel equidistant line.
[0108] The above-mentioned measured pixel resolution is obtained by averaging the historical pixel resolutions of the target river equidistant lines under multiple optimal evaluation conditions, and the value thereof may vary due to the river environment, actual needs, and the status of the acquisition equipment. This embodiment does not impose any special limitation on this.
[0109] The pixel resolution corresponding to the evaluation point of the above-mentioned target river channel is greater than the measured pixel resolution, which means that at this evaluation point, the image's ability to capture ground details is lower than the actual required accuracy. In other words, each pixel in the image covers a larger ground area than expected, resulting in "blurring" or "averaging" of detailed information about the river channel and its surrounding environment, making it impossible to accurately identify subtle changes in the river channel, such as water flow velocity, water depth changes, riverbank erosion, etc., and affecting the precise measurement of river channel width, morphology and other characteristics.
[0110] The pixel resolution corresponding to the point to be evaluated of the target river channel is smaller than the measured pixel resolution, and the spatial resolution of the image at this point is higher than the actual measured resolution, which means that the redundancy of image data is increased in the process of pixel resolution evaluation. When processing a large amount of high-resolution data, more computing resources and time are required, which increases the complexity of the analysis.
[0111] The specific process of the above feedback is: if the above verification process generates a resolution assessment abnormality label, the evaluator will be reminded of the resolution assessment abnormality through SMS, phone calls, emails, etc., and an abnormality identification report will be generated so that the evaluator can take necessary measures to resolve the abnormality.
[0112] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.
Claims
1. A method for evaluating pixel resolution of river channel equidistant lines based on image recognition, characterized in that: include: S1. Image quality assessment: Perform image recognition on the river area, thereby obtaining remote sensing image data of the river area, assess the image quality index of the river area, and compare it with the predefined image quality index threshold. If the image quality index of the river area is greater than or equal to the predefined image quality index threshold, enter S2; S2. Generating river channel equidistant lines: locating each pixel key point of the target river channel according to the remote sensing image data of the river channel area, connecting each pixel key point of the target river channel in a predetermined manner, generating each river channel equidistant line of the target river channel, and determining the resolution evaluation data of each river channel equidistant line of the target river channel; S3. Pixel resolution assessment: Based on the remote sensing image data of the river area, locate the resolution point to be assessed of the target river, obtain the resolution assessment data of the resolution point to be assessed of the target river, and perform comprehensive processing with the resolution assessment data of each river equidistant line of the target river to assess the pixel resolution corresponding to the resolution point to be assessed of the target river; S4. Feedback on evaluation results: The pixel resolution corresponding to the target river channel’s evaluation point is verified with the predefined measured pixel resolution, so as to provide feedback on the recognition process of the pixel resolution of the target river channel’s equidistant line.
2. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 1, characterized in that: The remote sensing image data of the river area specifically includes the grayscale value of each pixel point of the remote sensing image of the river area, the spatial resolution mean of the remote sensing image of the river area, the useful signal intensity value of each pixel point of the remote sensing image of the river area, and the noise signal intensity value of each pixel point of the remote sensing image of the river area.
3. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 2, characterized in that: The specific evaluation process of evaluating the image quality index of the river area is as follows: According to the grayscale value of each pixel of the river area remote sensing image, the grayscale value of each pixel of the river area remote sensing image in the first preset direction and the second preset direction is extracted, and the average gradient of the river area remote sensing image is obtained through comprehensive processing; Comprehensively process the useful signal intensity value of each pixel point of the river channel area remote sensing image and the noise signal intensity value of each pixel point of the river channel area remote sensing image to obtain the signal-to-noise ratio of the river channel area remote sensing image; The average gradient of the river area remote sensing image, the signal-to-noise ratio of the river area remote sensing image, and the mean spatial resolution of the river area remote sensing image are comprehensively analyzed to obtain the image quality index of the river area. The specific expression is: Where ZL is the image quality index of the river area, TD is the average gradient of the remote sensing image of the river area, XZ is the signal-to-noise ratio of the remote sensing image of the river area, FB is the mean spatial resolution of the remote sensing image of the river area, Y1 is the impact factor corresponding to the gradient unit value predefined in the river data cloud platform, Y2 is the impact factor corresponding to the signal-to-noise ratio unit value predefined in the river data cloud platform, Y3 is the impact factor corresponding to the resolution unit value predefined in the river data cloud platform, and e is a natural constant.
4. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 1, characterized in that: The image quality index of the river area is compared with a predefined image quality index threshold, and the specific comparison process is as follows: The image quality index of the river area is compared with the predefined image quality index threshold. If the image quality index of the river area is greater than or equal to the image quality index threshold, the key points of each pixel of the target river are located. If the image quality index of the river area is less than the image quality index threshold, the image recognition of the river area is performed again until the image quality index of the river area is greater than or equal to the image quality index threshold.
5. The method for evaluating pixel resolution of river equidistant lines based on image recognition according to claim 1, characterized in that: The specific generation process of generating the equidistant lines of each channel of the target channel is as follows: Number the pixel key points of the target river channel in a clockwise direction, and the required pixel key points are pixel key point P1, pixel key point P2, pixel key point P4, pixel key point P5, pixel key point P6, and pixel key point P8 respectively; After the pixel key point P1 and the pixel key point P4 are connected in a predetermined manner, the generated river channel equidistant line is marked as the first river channel equidistant line L1 of the target river channel; After the pixel key point P5 and the pixel key point P8 are connected in a predetermined manner, the generated river channel equidistant line is marked as the second river channel equidistant line L2 of the target river channel; After the pixel key point P1 and the pixel key point P2 are connected in a predetermined manner, the generated river channel equidistant line is marked as the third river channel equidistant line L3 of the target river channel; After the pixel key point P5 and the pixel key point P6 are connected in a predetermined manner, the generated river channel equidistant line is marked as the fourth river channel equidistant line L4 of the target river channel; In this way, the channel equidistant lines of the target channel are generated, including the first channel equidistant straight line of the target channel, the second channel equidistant straight line of the target channel, the third channel equidistant straight line of the target channel, and the fourth channel equidistant straight line of the target channel.
6. The method for evaluating pixel resolution of river equidistant lines based on image recognition according to claim 5 is characterized by: The specific determination process of the resolution evaluation data of each channel equidistant line of the target channel is as follows: The resolution assessment data of each channel equidistant line of the target river channel specifically includes the intercept, slope and water level of each channel equidistant line of the target river channel; According to the first channel equidistant straight line L1 of the target channel, the slope of the first channel equidistant straight line L1 of the target channel is obtained by slope calculation, which is recorded as the slope K1 of the first channel equidistant straight line of the target channel; According to the second channel equidistant straight line L2 of the target channel, the slope of the second channel equidistant straight line L2 of the target channel is obtained by slope calculation, which is recorded as the slope K2 of the second channel equidistant straight line of the target channel; According to the first channel equidistant straight line L1 of the target channel, the intercept of the first channel equidistant straight line L1 of the target channel is obtained by intercept calculation, which is recorded as the intercept b1 of the first channel equidistant straight line of the target channel; According to the second channel equidistant straight line L2 of the target channel, the intercept of the second channel equidistant straight line L2 of the target channel is obtained by intercept calculation, which is recorded as the intercept b2 of the second channel equidistant straight line of the target channel; The height value between the third channel equidistant line L3 of the target channel and the water level reference plane is obtained, which is recorded as the first water level sw1 of the target channel, and the height value between the fourth channel equidistant line L4 of the target channel and the water level reference plane is obtained, which is recorded as the second water level sw2 of the target channel.
7. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 1, characterized in that: The resolution assessment data of the target river channel resolution point to be assessed specifically include the slope corresponding to the target river channel resolution point to be assessed, the intercept corresponding to the target river channel resolution point to be assessed, and the water level corresponding to the target river channel resolution point to be assessed.
8. The method for evaluating pixel resolution of river equidistant lines based on image recognition according to claim 1, characterized in that: The resolution of the target river channel to be evaluated is the pixel resolution corresponding to the point to be evaluated. The specific evaluation process is as follows: The slope corresponding to the target river channel resolution to be evaluated point, the intercept corresponding to the target river channel resolution to be evaluated point, the water level corresponding to the target river channel resolution to be evaluated point, and the intercept, slope, and water level of each channel equidistant line of the target river are comprehensively processed to evaluate the pixel resolution corresponding to the target river channel resolution to be evaluated point.
9. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 8, characterized in that: The resolution of the target river is the pixel resolution corresponding to the point to be evaluated. The specific expression is: Where R is the pixel resolution corresponding to the target river resolution point to be evaluated, k is the slope corresponding to the target river resolution point to be evaluated, k1 is the slope of the first channel equidistant straight line of the target river, k2 is the slope of the second channel equidistant straight line of the target river, b is the intercept corresponding to the target river resolution point to be evaluated, b1 is the intercept of the first channel equidistant straight line of the target river, b2 is the intercept of the second channel equidistant straight line of the target river, sw is the water level corresponding to the target river resolution point to be evaluated, sw1 is the first water level of the target river, and sw2 is the second water level of the target river.
10. The method for evaluating pixel resolution of river channel equidistant lines based on image recognition according to claim 1, characterized in that: The recognition process of the pixel resolution of the target river channel equidistant line is fed back, and the specific feedback process is: The pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is verified with the predefined measured pixel resolution to generate a resolution assessment label. If the pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is equal to the measured pixel resolution, the resolution assessment label is defined as a resolution assessment qualified label. If the pixel resolution corresponding to the resolution-to-be-assessed point of the target river channel is greater than or less than the measured pixel resolution, the resolution assessment label is defined as a resolution assessment abnormal label, thereby providing feedback on the recognition process of the pixel resolution of the target river channel equidistant line.