An image processing method and system for embankment inspection

Abstract

The present application relates to the field of dam inspection, especially to a kind of image processing method and system for dam inspection, the present application is by controlling unmanned aerial vehicle to fly over detection area according to predetermined trajectory, when unmanned aerial vehicle is pre-entered detection area, continuously obtain image information and carry out multidimensional analysis, based on the result of multidimensional analysis and mirror surface reflection index, calculate to obtain spectral characteristic interference representation value, set interference label for detection area, subsequent interference label based on the control of unmanned aerial vehicle in detection area adaptively obtains spectral information in monitoring area, subsequent generation perspective adjustment instruction, by sampling determination selection confidence sampling position obtains spectral information, the present application is by image information in advance sensing interference situation of spectral information extraction in detection area, adaptively select spectral information capture mode, under the premise of guaranteeing inspection efficiency, improve the quality and reliability of the obtained spectral information, provide effective data support for subsequent analysis dam potential anomaly.

Description

Technical Field

[0001] This invention relates to the field of dam inspection, and more particularly to an image processing method and system for dam inspection. Background Technology

[0002] Traditional dam inspections primarily rely on manual visual inspections, which are inefficient, have blind spots, and struggle to detect early-stage hazards such as internal seepage. While the development of remote sensing technology has improved the efficiency of large-scale inspections with visible light drones, their information dimensions are limited, lacking the ability to detect potential risks under vegetation cover or within the dam structure. Against this backdrop, the application of spectral information technology has become a key breakthrough in improving the accuracy of dam defect identification. For example, vegetation has high reflectivity in the near-infrared band. When seepage occurs in the dam, causing changes in soil moisture content, the physiological state of the vegetation changes accordingly, leading to characteristic changes in its spectral reflectance curve. Therefore, drones, as flexible low-altitude remote sensing platforms, can be equipped with multispectral or hyperspectral sensors to acquire high-resolution images of ground features in specific discrete or continuous narrow bands. By analyzing the correlation between specific spectral characteristics and the physical state of the dam, early identification of potential hazards can be achieved.

[0003] For example, Chinese Patent Publication No. CN119492355A discloses a method, equipment, and medium for detecting water conservancy engineering dams based on unmanned aerial vehicles (UAVs). This method belongs to the field of water conservancy engineering dam technology and addresses the technical problems of existing water conservancy engineering dams, which often rely on manual inspections, resulting in low detection efficiency, high labor costs, and a tendency to miss or misjudge inspections, hindering intelligent monitoring and early warning of dam risks. The method includes: acquiring real-time detection images of the dam based on a preset UAV flight plan; performing risk identification processing on the real-time detection images related to the main features of the dam to obtain dam main feature information; calculating and processing the real-time detection images related to the dam's shape features using preset dam lidar data to obtain dam shape feature information; and performing index judgment processing on the real-time detection images related to the dam's environmental features based on preset multispectral image reflectance data to obtain dam environmental feature information.

[0004] However, the following problems still exist in the existing technology:

[0005] 1. Existing technologies do not consider the impact of various interference factors in complex detection environments on the accuracy of data acquisition, resulting in poor reliability of detection results;

[0006] 2. Existing technologies fail to adaptively select the optimal detection point based on environmental characteristics, making it difficult to guarantee the quality of key data collection. Summary of the Invention

[0007] To address this issue, the present invention provides an image processing method and system for dam inspection, which overcomes the problem that the existing technology does not consider the impact of various interference factors in complex detection environments on the accuracy of data acquisition, resulting in poor reliability of detection results. It can adaptively select the optimal detection point based on environmental characteristics, but it is difficult to guarantee the quality of key data acquisition.

[0008] To achieve the above objectives, in one aspect, the present invention provides an image processing method for dam inspection, comprising:

[0009] The drone is controlled to carry a first acquisition unit for acquiring image information and a second acquisition unit for acquiring spectral information to pass through several detection areas along a predetermined trajectory.

[0010] In response to the UAV's pre-entry detection area, image information is continuously acquired for multi-dimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discreteness based on image information.

[0011] Extract image information to identify specular reflection pixels, and use the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index;

[0012] Based on the results of the multidimensional analysis and the specular reflection index, the spectral feature interference characterization value is calculated to set the interference label for the detection area;

[0013] Based on the interference tag, the drone is controlled to acquire spectral information within the detection area, including:

[0014] The second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0015] Alternatively, a viewing angle adjustment command can be generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit can be controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization values ​​of several moving segments and perform sampling judgment.

[0016] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position;

[0017] The spectral information is stored in the dam inspection database.

[0018] Furthermore, the process of continuously acquiring image information for multidimensional analysis includes,

[0019] Extract the chromaticity mean from each image information, construct a temporal mapping between the chromaticity mean and the image information acquisition time, and construct a temporal variation curve of the chromaticity mean;

[0020] The peak variance of several peaks is determined to obtain the time-domain fluctuation parameter of the color feature;

[0021] Determine the grayscale distribution histogram of the image information to determine the entropy value, and then determine the average entropy value corresponding to each image information as the texture feature discrete quantity.

[0022] Furthermore, the process of identifying specular reflection pixels includes,

[0023] Image information is extracted and converted to the HSV color space for clustering and segmentation to divide the image into several blocks that meet the consistency criteria in color and brightness features.

[0024] Calculate the average block brightness and average block saturation of all pixels within each block;

[0025] Blocks that simultaneously meet the conditions of having an average brightness value higher than a preset brightness threshold and an average saturation value lower than a preset saturation threshold are identified as specular reflection blocks.

[0026] All pixels belonging to the specular reflection block are marked as specular reflection pixels.

[0027] Furthermore, based on the results of the multidimensional analysis and the specular reflectance index, the process of calculating the spectral characteristic interferometry characterization value includes:

[0028] The ratio of the color feature temporal fluctuation parameter to the preset temporal fluctuation parameter threshold is calculated to obtain the temporal fluctuation interference factor;

[0029] The texture feature interference factor is obtained by calculating the ratio of the discrete amount of the texture feature to the preset discrete amount threshold.

[0030] The specular reflection index is calculated as a ratio to a preset reflection index threshold to obtain the specular reflection interference factor.

[0031] The spectral feature interference characterization value is obtained by weighted summation of the temporal fluctuation interference factor, texture feature interference factor, and specular reflection interference factor.

[0032] Furthermore, the process of setting interference labels on the detection area includes,

[0033] If the spectral feature interference characterization value of the detection area is greater than or equal to the preset spectral feature interference threshold, then a strong interference label is set for the detection area;

[0034] If the spectral feature interference characterization value of the detection area is less than the preset spectral feature interference threshold, then a weak interference label is set for the detection area.

[0035] Furthermore, based on the interference tag, the drone is controlled to acquire spectral information within the detection area, including:

[0036] If the detection area is a weak interference label, then the second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0037] If the detection area is a strong interference label, a viewing angle adjustment command is generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit is controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization value of several moving segments and perform sampling judgment.

[0038] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position.

[0039] Furthermore, the process of controlling the first acquisition unit to focus on the target visual point and acquire image information of different background areas corresponding to the target visual point during the movement of the drone includes,

[0040] Select the target visual point from the image information corresponding to the detection area, and determine the target visual point and the observation plane of the UAV;

[0041] Starting from the drone, select a direction of movement on the observation plane and move it while maintaining the first acquisition unit focused on the target visual point during the movement.

[0042] The observation plane is perpendicular to the plane of the virtual vector corresponding to the UAV and the target visual point, and the UAV is located on the observation plane.

[0043] Furthermore, the process of determining the changing trends of the spectral characteristic interferometric characterization values ​​of several moving segments and performing sampling judgment includes,

[0044] If the judgment criteria are met, the sampling judgment is deemed successful.

[0045] The determination condition is that the spectral feature interference characterization value is in a decreasing trend.

[0046] Furthermore, the process of determining the confidence sampling location includes,

[0047] Determine the shift segment corresponding to the minimum spectral characteristic interferometric characterization value;

[0048] Determine several positions of the UAV corresponding to the movement segment, and randomly select a confidence sampling position from each of the positions.

[0049] On the other hand, a system is provided for applying an image processing method for dam inspection, comprising:

[0050] The inspection control module is used to control the UAV to pass through several detection areas along a predetermined trajectory.

[0051] The multidimensional analysis module, in response to the UAV's pre-entry detection area, continuously acquires image information for multidimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discrete quantities based on image information.

[0052] The mirror analysis module is used to extract image information to identify specular reflection pixels and use the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index.

[0053] The label setting module is used to calculate the spectral feature interference characterization value based on the results of the multidimensional analysis and the specular reflection index, so as to set the interference label for the detection area;

[0054] A control module, used to control a drone to acquire spectral information within the detection area based on the interference label, including:

[0055] The second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0056] Alternatively, a viewing angle adjustment command can be generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit can be controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization values ​​of several moving segments and perform sampling judgment.

[0057] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position;

[0058] The dam inspection database is used to store the acquired spectral information.

[0059] Compared with existing technologies, this invention controls a drone to fly over a detection area along a predetermined trajectory. When the drone enters the detection area, it continuously acquires image information for multidimensional analysis. Based on the results of the multidimensional analysis and the specular reflection index, it calculates the spectral feature interference characterization value and sets an interference label for the detection area. Subsequently, based on the interference label of the detection area, the drone is controlled to adaptively acquire spectral information within the monitoring area. Then, it generates a viewing angle adjustment command and selects a confidence sampling position to acquire spectral information through sampling judgment. This invention can perceive the interference of the detection area on spectral information extraction in advance through image information and adaptively select the spectral information capture method. While ensuring inspection efficiency, it improves the quality and reliability of the acquired spectral information, providing effective data support for subsequent analysis of potential anomalies in the dam.

[0060] In particular, this invention continuously acquires image information for multidimensional analysis when the UAV pre-enters the detection area. Through multidimensional analysis, it determines the discreteness of texture features and the temporal fluctuation parameters of color features. In practice, the fundamental purpose of spectral analysis is to identify the composition of a substance by its absorption and reflection characteristics of light at specific wavelengths. However, in reality, different areas have different environments and lighting conditions, such as cloud cover and light reflection. Complex lighting can affect the accuracy of spectral information, leading to information distortion. Therefore, this invention performs multidimensional analysis along the UAV's predetermined trajectory to ensure efficiency, while simultaneously utilizing the continuous acquisition of image information during the UAV's movement. The changes in color features in the temporal dimension of image information reflect the intensity of light and shadow changes. At the same time, it captures the texture complexity in a single image, considering the formation of microscopic shadows and the mixing of multiple reflection angles. Furthermore, the extraction methods for color feature temporal fluctuation parameters and texture complexity are simple and can be analyzed quickly, making it suitable for continuous inspection processes by UAVs. This allows for efficient early detection of interference factors affecting spectral information acquisition in the next detection area along a predetermined trajectory, providing data support for subsequent calculation of spectral feature interference characterization values. In addition, interference labels can be set to adaptively acquire spectral information within the detection area, ensuring inspection efficiency and improving the quality and reliability of the acquired spectral information.

[0061] In particular, this invention obtains the specular reflection index. In reality, some water surfaces or smooth rocks may reflect light from other light sources, such as sunlight, obscuring the spectrum of the ground features themselves. Furthermore, when the texture features in the image information have high discreteness, the superposition effect is significant. Based on this, this invention considers that the combined superposition of temporal fluctuation interference factors, texture feature interference factors, and specular reflection interference factors can easily lead to spectral signal distortion. Therefore, it constructs a spectral feature interference characterization value to reflect the influence of the environmental conditions within the detection area on the spectral signal. Interference labels are set for the detection area to adaptively obtain spectral information within the detection area, ensuring inspection efficiency and improving the quality and reliability of the obtained spectral information.

[0062] In particular, this invention controls a drone to acquire spectral information within the detection area based on the interference tag. For weak interference tags, it directly acquires spectral information within the detection area. For strong interference tags, it generates a viewing angle adjustment command to acquire image information of different background areas corresponding to the target visual point. Then, using the spectral feature interference characterization value, it actively acquires the interference of spectral information at different positions. Combined with the sampling judgment mechanism, it can determine the confidence sampling position and find a sampling position that relatively reflects the true spectral information of the dam area. The dynamic adaptive process ensures the reliability and effectiveness of the spectral information while ensuring inspection efficiency, providing effective data support for subsequent analysis of potential dam anomalies. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the steps of an image processing method for dam inspection according to an embodiment of the invention.

[0064] Figure 2 This is a logic block diagram for determining the mirror reflection area in an embodiment of the invention;

[0065] Figure 3 A logic block diagram for setting interference labels in the detection area according to an embodiment of the invention;

[0066] Figure 4 This is a logic block diagram of an embodiment of the invention for controlling a drone to acquire spectral information within the detection area based on an interference tag. Detailed Implementation

[0067] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0068] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0069] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0070] Please see Figure 1 As shown, Figure 1 This is a schematic diagram illustrating the steps of an image processing method for dam inspection according to an embodiment of the present invention. The image processing method for dam inspection according to an embodiment of the present invention includes:

[0071] Step S1: Control the drone to carry a first acquisition unit for acquiring image information and a second acquisition unit for acquiring spectral information to pass through several detection areas along a predetermined trajectory;

[0072] Step S2, in response to the UAV's pre-entry detection area, continuously acquire image information for multi-dimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discreteness based on image information;

[0073] Step S3: Extract image information to identify specular reflection pixels, and use the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index.

[0074] Step S4: Based on the results of the multidimensional analysis and the specular reflection index, calculate the spectral feature interference characterization value to set an interference label for the detection area;

[0075] Step S5, based on the interference tag, control the UAV to acquire spectral information within the detection area, including,

[0076] The second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0077] Alternatively, a viewing angle adjustment command can be generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit can be controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization values ​​of several moving segments and perform sampling judgment.

[0078] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position;

[0079] Step S6: Store the spectral information in the dam inspection database.

[0080] Specifically, there is no limitation on the way the predetermined trajectory is set. It is determined based on the dam area to be inspected. Those skilled in the art can set the predetermined trajectory according to the actual situation so that the predetermined trajectory traverses each dam area.

[0081] Specifically, the first acquisition unit can be a high-definition visible light camera, whose lens should have autofocus and variable focal length functions. The first acquisition unit can be mounted on a gimbal to achieve continuous focusing on the target visual point.

[0082] Understandably, the effective pixel count of a high-definition visible light camera should not be too low to avoid affecting subsequent analysis due to insufficient image resolution. In practice, the effective pixel count of a high-definition visible light camera should not be lower than 12 million.

[0083] Specifically, the second acquisition unit can be a spectral imager, which can be mounted on a gimbal. The acquisition direction of the second acquisition unit can be synchronized with that of the first acquisition unit to ensure the correlation between image data and spectral information.

[0084] Specifically, the process of continuously acquiring image information for multidimensional analysis includes,

[0085] Extract the chromaticity mean from each image information, construct a temporal mapping between the chromaticity mean and the image information acquisition time, and construct a temporal variation curve of the chromaticity mean;

[0086] The peak variance of several peaks is determined to obtain the time-domain fluctuation parameter of the color feature;

[0087] Determine the grayscale distribution histogram of the image information to determine the entropy value, and then determine the average entropy value corresponding to each image information as the texture feature discrete quantity.

[0088] In practice, the horizontal axis of the chromaticity mean time-domain variation curve represents the time of image information acquisition, and the vertical axis represents the chromaticity mean.

[0089] It is understandable that a higher entropy value indicates higher uncertainty in image information, more complex and coarser texture, and a higher temporal fluctuation parameter in color features indicates poor overall color consistency in continuously acquired image information, with abrupt changes and fluctuations, reflecting flickering shadows or interlaced shadows caused by the interplay of light and shadow.

[0090] This invention continuously acquires image information and performs multidimensional analysis when the UAV pre-enters the detection area. Through multidimensional analysis, it determines the discreteness of texture features and the temporal fluctuation parameters of color features. In practice, the fundamental purpose of spectral analysis is to identify the composition of a substance by its absorption and reflection characteristics of light at specific wavelengths. However, in reality, different areas have different environments and lighting conditions, such as cloud cover and light reflection. Complex lighting can affect the accuracy of spectral information, leading to information distortion. Therefore, this invention performs multidimensional analysis along the UAV's predetermined trajectory to ensure efficiency, while simultaneously utilizing the continuous acquisition of image information during the UAV's movement. The method acquires and considers the changes in color features in the temporal dimension of image information, reflecting the intensity of light and shadow changes. It also captures the texture complexity in a single image, considering the formation of shadows at the microscale and the mixing of multiple reflection angles. Furthermore, the extraction methods for color feature temporal fluctuation parameters and texture complexity are simple and can be analyzed quickly. This allows for efficient early detection of interference factors affecting spectral information acquisition in the next detection area along a predetermined trajectory, providing data support for subsequent calculation of spectral feature interference characterization values. In addition, it allows for the adaptive acquisition of spectral information within the detection area by setting interference labels, ensuring inspection efficiency and improving the quality and reliability of the acquired spectral information.

[0091] Specifically, please refer to Figure 2 As shown, Figure 2 This is a logic block diagram for determining specular reflection blocks according to an embodiment of the invention. The process of identifying specular reflection pixels includes:

[0092] Image information is extracted and converted to the HSV color space for clustering and segmentation to divide the image into several blocks that meet the consistency criteria in color and brightness features.

[0093] Calculate the average block brightness and average block saturation of all pixels within each block;

[0094] Blocks that simultaneously meet the conditions of having an average brightness value higher than a preset brightness threshold and an average saturation value lower than a preset saturation threshold are identified as specular reflection blocks.

[0095] All pixels belonging to the specular reflection block are marked as specular reflection pixels.

[0096] In implementation, the consistency condition is that the difference ratio of color features between pixels within a block is less than 0.2 and the difference ratio of brightness features is less than 0.2, thereby clustering and segmenting blocks with similar colors and brightness.

[0097] Specifically, the preset brightness threshold is set to 180, and the preset saturation threshold is set to 0.2. These thresholds are based on the characteristics of the HSV color space: the brightness component ranges from [0, 255], and the saturation component ranges from [0, 255]. When the average brightness of a block is higher than 180, it indicates that the area has high reflection intensity and exhibits specular reflection characteristics; simultaneously, when the average saturation of a block is lower than 50, it indicates that the area lacks color information, consistent with the characteristic of color distortion caused by specular reflection. In practical applications, those skilled in the art can adjust the thresholds appropriately according to specific ambient lighting conditions, which will not be elaborated further here.

[0098] Specifically, the process of calculating the spectral characteristic interferometry characterization value based on the results of the multidimensional analysis and the specular reflectance index includes the following steps:

[0099] The ratio of the color feature temporal fluctuation parameter to the preset temporal fluctuation parameter threshold is calculated to obtain the temporal fluctuation interference factor;

[0100] The texture feature interference factor is obtained by calculating the ratio of the discrete amount of the texture feature to the preset discrete amount threshold.

[0101] The specular reflection index is calculated as a ratio to a preset reflection index threshold to obtain the specular reflection interference factor.

[0102] The spectral feature interference characterization value is obtained by weighted summation of the temporal fluctuation interference factor, texture feature interference factor, and specular reflection interference factor.

[0103] Specifically, the temporal fluctuation parameter threshold and the discrete quantity threshold are predetermined. A large number of image information without shadows in typical detection areas are obtained in advance as samples. Several color feature temporal fluctuation parameters and texture feature discrete quantities are determined. The mean value of the color feature temporal fluctuation parameter and the mean value of the texture feature discrete quantity are solved to determine the color feature temporal fluctuation parameter and texture feature discrete quantity corresponding to the image information in the detection area under normal conditions. In order to characterize the deviation under normal conditions, the temporal fluctuation parameter threshold is set as the product of the mean value of the color feature temporal fluctuation parameter and the first offset coefficient. The first offset coefficient is selected in the interval [1.5,2], preferably 1.6.

[0104] The discrete threshold is set as the product of the mean discrete value of the texture feature and the second offset coefficient, which is selected in the interval [1.3, 1.5], preferably 1.4.

[0105] The reflectance index threshold is preset. The specular reflectance index represents the proportion of specular reflective pixels. In practice, a certain number of specular reflective pixels can be allowed, but too many specular reflective pixels will affect the acquisition of spectral information. In implementation, the reflectance index threshold is selected in the range [0.1, 0.4], preferably 0.2.

[0106] To comprehensively consider the effects of temporal fluctuation interference, texture feature interference, and specular reflection interference, the weights for weighted summation are 0.3, 0.3, and 0.4, respectively.

[0107] Specifically, please refer to Figure 3 As shown, Figure 3 This is a logic block diagram illustrating the setting of interference labels on a detection area according to an embodiment of the invention. The process of setting interference labels on the detection area includes:

[0108] If the spectral feature interference characterization value of the detection area is greater than or equal to the preset spectral feature interference threshold, then a strong interference label is set for the detection area;

[0109] If the spectral feature interference characterization value of the detection area is less than the preset spectral feature interference threshold, then a weak interference label is set for the detection area.

[0110] In implementation, the spectral feature interference characterization value calculated under the conditions that the color feature temporal fluctuation parameter is equal to the preset temporal fluctuation parameter threshold, the texture feature discrete quantity is equal to the preset discrete quantity threshold, and the specular reflection index is equal to the preset reflection index threshold is set as the spectral feature interference threshold.

[0111] This invention obtains the specular reflection index. In reality, some water surfaces or smooth rocks may reflect light from other light sources, such as sunlight, obscuring the spectrum of the ground features themselves. Furthermore, when the texture features in the image information have high discreteness, the superposition effect is significant. Based on this, this invention considers that the combined superposition of temporal fluctuation interference factors, texture feature interference factors, and specular reflection interference factors can easily lead to spectral signal distortion. Therefore, it constructs a spectral feature interference characterization value to reflect the influence of environmental conditions within the detection area on the spectral signal. Interference labels are set for the detection area to adaptively acquire spectral information within the detection area, ensuring inspection efficiency and improving the quality and reliability of the acquired spectral information.

[0112] Specifically, please refer to Figure 4 The diagram shown is a logic block diagram of an embodiment of the invention, illustrating the acquisition of spectral information within a detection area by controlling a drone based on an interference tag. The acquisition of spectral information within the detection area by controlling the drone based on the interference tag includes...

[0113] If the detection area is a weak interference label, then the second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0114] If the detection area is a strong interference label, a viewing angle adjustment command is generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit is controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization value of several moving segments and perform sampling judgment.

[0115] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position.

[0116] Specifically, the process of controlling the first acquisition unit to focus on the target visual point and acquire image information of different background areas corresponding to the target visual point during the movement of the drone includes:

[0117] Select the target visual point from the image information corresponding to the detection area, and determine the target visual point and the observation plane of the UAV;

[0118] Starting from the drone, select a direction of movement on the observation plane and move it while maintaining the first acquisition unit focused on the target visual point during the movement.

[0119] The observation plane is perpendicular to the plane of the virtual vector corresponding to the UAV and the target visual point, and the UAV is located on the observation plane.

[0120] In practice, the acquisition direction of the first acquisition unit can be changed so that the acquisition direction corresponds to the target visual point, which is a point in the image information, preferably the image center.

[0121] The direction of movement can be any direction on the observation plane, and the distance of a single movement is between 5m and 10m. By moving on the observation plane, the background at the visual point of the observed target can be changed, thereby changing the interference of the environment on the acquisition of spectral information to a certain extent.

[0122] Specifically, the process of determining the changing trends of the spectral characteristic interferometric characterization values ​​of several moving segments and performing sampling and judgment includes,

[0123] If the judgment criteria are met, the sampling judgment is deemed successful.

[0124] The determination condition is that the spectral feature interference characterization value is in a decreasing trend.

[0125] In practice, the spectral characteristic interference characterization value corresponding to the initial moving segment and the spectral characteristic interference characterization value corresponding to the final moving segment can be determined. If the spectral characteristic interference characterization value corresponding to the final moving segment is less than the spectral characteristic interference characterization value corresponding to the initial moving segment, it is determined to be a downward trend.

[0126] The moving segments are divided based on the distance of a single movement, and a single moving segment is 0.1 times the distance of a single movement.

[0127] If the sampling determination fails, a new view adjustment command is generated, causing the drone to select a new direction of movement. After the movement is completed, the sampling determination is performed again, and this process is repeated until the sampling determination passes.

[0128] Specifically, the process of determining the confidence sampling location includes,

[0129] Determine the shift segment corresponding to the minimum spectral characteristic interferometric characterization value;

[0130] Determine several positions of the UAV corresponding to the movement segment, and randomly select a confidence sampling position from each of the positions.

[0131] In practice, when controlling the second acquisition unit to capture spectral information within the detection area at the corresponding confidence sampling position, the acquisition direction of the second acquisition unit can be the same as that of the first acquisition unit.

[0132] This invention controls a drone to acquire spectral information within the detection area based on the interference tag. For weak interference tags, it directly acquires spectral information within the detection area. For strong interference tags, it generates a viewing angle adjustment command to acquire image information of different background areas corresponding to the target visual point. Then, using the spectral feature interference characterization value, it actively acquires the interference of spectral information at different positions. Combined with a sampling judgment mechanism, it can determine the confidence sampling position and find a sampling position that relatively reflects the true spectral information of the dam area. The dynamic adaptive process ensures the reliability and effectiveness of the spectral information while ensuring inspection efficiency, providing effective data support for subsequent analysis of potential dam anomalies.

[0133] On the other hand, a system for image processing methods for dam inspection is also provided, including,

[0134] The inspection control module is used to control the UAV to pass through several detection areas along a predetermined trajectory.

[0135] The multidimensional analysis module, in response to the UAV's pre-entry detection area, continuously acquires image information for multidimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discrete quantities based on image information.

[0136] The mirror analysis module is used to extract image information to identify specular reflection pixels and use the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index.

[0137] The label setting module is used to calculate the spectral feature interference characterization value based on the results of the multidimensional analysis and the specular reflection index, so as to set the interference label for the detection area;

[0138] A control module, used to control a drone to acquire spectral information within the detection area based on the interference label, including:

[0139] The second acquisition unit is directly controlled to capture the spectral information of the detection area;

[0140] Alternatively, a viewing angle adjustment command can be generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit can be controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization values ​​of several moving segments and perform sampling judgment.

[0141] If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position;

[0142] The dam inspection database is used to store the acquired spectral information.

[0143] Specifically, there are no restrictions on the specific form of the inspection control module, multidimensional analysis module, mirror analysis module, tag setting module, and control module. They can be composed of logic components, including field-programmable processors, computers, or microprocessors in computers, which will not be elaborated further.

[0144] The dam inspection database can be a virtual or physical storage database, as long as it can store the acquired spectral information, which will not be elaborated further.

[0145] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An image processing method for dam inspection, characterized in that, include: The drone is equipped with a first acquisition unit for acquiring image information and a second acquisition unit for acquiring spectral information to pass through several detection areas along a predetermined trajectory. In response to the UAV's pre-entry detection area, image information is continuously acquired for multi-dimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discreteness based on image information. Extract image information to identify specular reflection pixels, and use the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index; Based on the results of the multidimensional analysis and the specular reflection index, the spectral feature interference characterization value is calculated to set the interference label for the detection area; The process of setting interference labels on the detection area includes, If the spectral feature interference characterization value of the detection area is greater than or equal to the preset spectral feature interference threshold, then a strong interference label is set for the detection area; If the spectral feature interference characterization value of the detection area is less than the preset spectral feature interference threshold, then a weak interference label is set for the detection area; Based on the interference tag, the drone is controlled to acquire spectral information within the detection area, including: If the detection area is a weak interference label, the second acquisition unit is directly controlled to capture the spectral information of the detection area; If the detection area is a strong interference label, a viewing angle adjustment command is generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit is controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization value of several moving segments and perform sampling judgment. If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position; The spectral information is stored in the dam inspection database.

2. The image processing method for dam inspection according to claim 1, characterized in that, The process of continuously acquiring image information for multidimensional analysis includes, Extract the chromaticity mean from each image information, construct a temporal mapping between the chromaticity mean and the image information acquisition time, and construct a temporal variation curve of the chromaticity mean; The peak variance of several peaks is determined to obtain the time-domain fluctuation parameter of the color feature; Determine the grayscale distribution histogram of the image information to determine the entropy value, and then determine the average entropy value corresponding to each image information as the texture feature discrete quantity.

3. The image processing method for dam inspection according to claim 2, characterized in that, The process of identifying specular reflection pixels includes, Image information is extracted and converted to the HSV color space for clustering and segmentation to divide the image into several blocks that meet the consistency criteria in color and brightness features. Calculate the average block brightness and average block saturation of all pixels within each block; Blocks that simultaneously meet the conditions of having an average brightness value higher than a preset brightness threshold and an average saturation value lower than a preset saturation threshold are identified as specular reflection blocks. All pixels belonging to the specular reflection block are marked as specular reflection pixels.

4. The image processing method for dam inspection according to claim 1, characterized in that, Based on the results of the multidimensional analysis and the specular reflectance index, the process of calculating the spectral characteristic interferometry characterization value includes: The ratio of the color feature temporal fluctuation parameter to the preset temporal fluctuation parameter threshold is calculated to obtain the temporal fluctuation interference factor; The texture feature interference factor is obtained by calculating the ratio of the discrete amount of the texture feature to the preset discrete amount threshold. The specular reflection index is calculated as a ratio to a preset reflection index threshold to obtain the specular reflection interference factor. The spectral feature interference characterization value is obtained by weighted summation of the temporal fluctuation interference factor, texture feature interference factor, and specular reflection interference factor.

5. The image processing method for dam inspection according to claim 1, characterized in that, The process of controlling the first acquisition unit to focus on the target visual point and acquire image information of different background areas corresponding to the target visual point during the movement of the drone includes, Select the target visual point from the image information corresponding to the detection area, and determine the target visual point and the observation plane of the UAV; Starting from the drone, select a direction of movement on the observation plane and move it while maintaining the first acquisition unit focused on the target visual point during the movement. The observation plane is perpendicular to the plane corresponding to the virtual vector of the UAV and the target visual point, and the UAV is located on the observation plane.

6. The image processing method for dam inspection according to claim 1, characterized in that, The process of determining the changing trends of the spectral characteristic interferometric characterization values ​​of several moving segments and performing sampling and judgment includes: If the judgment criteria are met, the sampling judgment is deemed successful. The determination condition is that the spectral feature interference characterization value is in a decreasing trend.

7. The image processing method for dam inspection according to claim 1, characterized in that, The process of determining the confidence sampling location includes, Determine the shift segment corresponding to the minimum spectral characteristic interferometric characterization value; Determine several positions of the UAV corresponding to the movement segment, and randomly select a confidence sampling position from each of the positions.

8. A system for applying the image processing method for dam inspection according to any one of claims 1-7, characterized in that, include: The inspection control module is used to control the UAV to pass through several detection areas along a predetermined trajectory, using a first acquisition unit for acquiring image information and a second acquisition unit for acquiring spectral information. The multidimensional analysis module, in response to the UAV's pre-entry detection area, continuously acquires image information for multidimensional analysis, including extracting color feature temporal fluctuation parameters based on continuous image information and determining texture feature discrete quantities based on image information. The mirror analysis module is used to extract image information to identify specular reflection pixels and uses the ratio of the number of specular reflection pixels to the total number of pixels in the image information as the specular reflection index. The label setting module is used to calculate the spectral feature interference characterization value based on the results of the multidimensional analysis and the specular reflection index, so as to set the interference label for the detection area; The control module, used to control the UAV to acquire spectral information within the detection area based on the interference label, includes: The second acquisition unit is directly controlled to capture the spectral information of the detection area; Alternatively, a viewing angle adjustment command can be generated to control the movement of the UAV. During the movement of the UAV, the first acquisition unit can be controlled to focus on the target visual point to acquire image information of different background areas corresponding to the target visual point, so as to determine the changing trend of the spectral feature interference characterization values ​​of several moving segments and perform sampling judgment. If the sampling is deemed successful, a confidence sampling position is determined, and the second acquisition unit is controlled to capture spectral information within the detection area at the corresponding confidence sampling position; The dam inspection database is used to store the acquired spectral information.

Images