A method for image enhancement processing of UAV remote sensing mapping

Abstract

The invention discloses an image enhancement processing method for remote sensing and mapping of unmanned aerial vehicles, and relates to the technical field of image enhancement processing. During the operation of the system, the flight attitude and position information are recorded by GNSS and IMU equipment, the flight path of the unmanned aerial vehicle is controlled, and the collected original image is preliminarily processed to calculate and obtain: flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf, and the non-local mean NLM algorithm is used to remove random noise in the remote sensing image, optimize the edge area, and use the Laplace pyramid algorithm to improve the edge clarity and texture details of the terrain image, strengthen the mountain and hilly landform features, correct the color distortion caused by the complex lighting conditions in the mountains, make the image color truly restore the terrain features, avoid the loss of important landform information due to excessive brightness or darkness, optimize the terrain feature extraction, and generate the final index Z final ,Quality assessment of enhanced and rectified images.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement processing, and in particular to an image enhancement processing method for unmanned aerial vehicle remote sensing mapping. Background Art

[0002] UAV remote sensing mapping is a technical method that uses drones as a platform and carries a variety of sensor equipment (such as multispectral cameras, lidar, light sensors, etc.) to efficiently collect data and extract features of the surface environment. Compared with traditional ground mapping and satellite remote sensing, UAV remote sensing mapping has the characteristics of low cost, high flexibility, excellent spatial resolution and temporal resolution, and is widely used in terrain mapping, disaster assessment, environmental monitoring and other fields. However, for UAV remote sensing mapping of mountain and hill mapping, due to factors such as complex terrain environment, changing lighting conditions and unstable flight dynamics of drones, the collected images often have problems such as blurred edges, uneven light and shadow, and loss of details, which seriously affect the quality and reliability of mapping data. Therefore, enhancing the mapping images to improve their edge clarity, color authenticity and terrain feature retention is an important technical means to ensure the accuracy of mountain and hill mapping.

[0003] Existing UAV remote sensing mapping image processing methods are mostly based on single-dimensional optimization, such as suppressing image noise or enhancing details alone, lacking comprehensive consideration of multiple factors such as flight status, lighting conditions, and terrain complexity, resulting in obvious limitations in processing effects. In addition, traditional image processing methods are insufficient in representing terrain features in high-slope areas and complex texture areas, and are prone to edge blur and detail loss. Especially in mountain and hilly mapping scenes, due to complex lighting conditions, color distortion and unnatural brightness transitions often occur, affecting the authenticity and usability of the final mapping results. These deficiencies not only reduce the overall quality of image processing, but also limit the precise extraction and analysis of terrain features.

[0004] The shortcomings of existing UAV remote sensing image processing methods mainly come from the following aspects: First, the altitude fluctuation and attitude change of the UAV during flight (such as excessive inclination) will lead to poor spatial consistency of image acquisition, which will cause blur or distortion problems; second, the uneven illumination of complex terrain (such as the strong contrast between shadow areas and highlight areas) is difficult to effectively solve by simple brightness correction or color balancing methods; third, a single algorithm for noise suppression and detail enhancement cannot avoid introducing artifacts or over-smoothing while retaining terrain features. These defects may lead to the following abnormal effects: blurred image boundaries lead to unclear expression of terrain features (such as gullies and ridges), imbalance of light and shadow makes it difficult to restore the real landform in surveying and mapping data, and insufficient expression of terrain features leads to reduced accuracy of slope and texture information extraction, thus affecting subsequent terrain analysis and decision support. The above problems urgently need to be solved through multi-dimensional comprehensive optimization of image enhancement methods. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides an image enhancement processing method for unmanned aerial vehicle remote sensing mapping, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for enhancing image processing of UAV remote sensing mapping, comprising the following steps:

[0007] Step 1: Use GNSS and IMU devices to record flight attitude and position information, control the flight path of the drone, adjust the altitude and angle, and ensure coverage of the target area to collect high-precision flight status, image sampling, lighting conditions, and terrain feature data;

[0008] Step 2: Perform preliminary processing on the collected original images, remove interference information, preliminarily mark the peaks and gullies in the images, and calculate and obtain: flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf;

[0009] The flight stability coefficient Fs is calculated by the following formula:

[0010]

[0011] In the formula, H f Indicates the flight altitude, ΔH f Indicates the flight altitude fluctuation range, Havg indicates the average flight altitude, Ws indicates the wind speed, Wmax indicates the maximum allowable wind speed, Ap indicates the shooting inclination angle, k1 and k2 indicate the weight coefficients of each parameter, respectively, to adjust the effect of wind speed and inclination angle on flight stability;

[0012] The image acquisition quality coefficient Iq is calculated by the following formula:

[0013]

[0014] Where Pr represents the resolution accuracy, P ideal Indicates the ideal resolution accuracy, L u Indicates the uniformity of illumination, L ideal Ideal illumination uniformity, L i Indicates light intensity, L min represents the minimum received light intensity, k3 and k4 represent the weight coefficients of each parameter, which are used to adjust the contribution of resolution and light to image acquisition quality;

[0015] The terrain feature retention coefficient Tf is calculated by the following formula:

[0016]

[0017] In the formula, S g Indicates the slope change gradient, S max Indicates the maximum slope gradient, V r represents the vegetation coverage, V max represents the maximum vegetation coverage, T c Represents texture complexity, T max represents the maximum texture complexity, H d Indicates the elevation difference, H max represents the maximum elevation difference, k5, k6, k7 and k8 represent the weight coefficients of each parameter, which are used to adjust the contribution of each parameter to the retention of terrain features;

[0018] Step 3: Use the non-local means (NLM) algorithm to identify high-frequency noise points and remove random noise from remote sensing images, while retaining terrain information, optimizing edge areas, and avoiding boundary blur.

[0019] Step 4: Use the Laplace pyramid algorithm to improve the edge clarity and texture details of the terrain image and enhance the features of mountainous and hilly landforms;

[0020] Step 5: Correct the color distortion caused by the complex lighting conditions in the mountains. Adjust the color balance of the image based on the grayscale world hypothesis and tone mapping, and adjust the brightness of the shadow and highlight areas so that the image color can truly restore the terrain characteristics and avoid the loss of important landform information due to excessive brightness or darkness.

[0021] Step 6: Optimize terrain feature extraction and generate the final index Z final ,quality assessment of enhanced and rectified images;

[0022] The image quality is quantified by the clarity coefficient, color balance coefficient and terrain detail retention coefficient, and the comprehensive evaluation index Z is used. finalCalculated by the following formula:

[0023] Z final =α*Fs+β*Iq+γ*Tf;

[0024] Where, Fs represents the flight stability coefficient, Iq represents the image acquisition quality coefficient, Tf represents the terrain feature retention coefficient, α, β and γ represent the weight values ​​of the flight stability coefficient Fs, the image acquisition quality coefficient Iq and the terrain feature retention coefficient Tf respectively;

[0025] The enhanced and corrected terrain features, including slope distribution, vegetation coverage and texture feature information, are extracted and evaluated by the comprehensive evaluation index Z final Compare with the preset standard threshold Q to determine whether the image processing quality meets expectations, and formulate targeted improvement measures to improve the overall image quality and the accuracy of terrain mapping results, and obtain the comparison strategy:

[0026] When the comprehensive evaluation index Z final ≥Preset standard threshold Q, indicating that the image quality meets the standard, the processing effect is good, no further optimization is required, and the output results include image data and terrain feature extraction analysis report;

[0027] When the comprehensive evaluation index Z final <Preset standard threshold Q, image quality is below the standard, further analysis of the cause of the problem is needed, and improvement measures should be taken;

[0028] Further analysis questions include:

[0029] The image edges are blurred or details are missing. Reasons: The flight stability is not up to standard, resulting in blurred images, and the noise suppression or detail enhancement links are not fully optimized;

[0030] Improvement measures: Increase the stability of the flight path, reduce the flight altitude fluctuation ΔHf and the shooting inclination angle Ap, optimize the flight speed, and avoid image blur caused by flying too fast;

[0031] The image has obvious color deviation or uneven light and shadow. The reason is: the color correction process does not accurately adjust the light or color temperature deviation, and the light and shadow balance does not meet expectations;

[0032] Improvement measures: Enhance the accuracy of the color temperature sensor and record the actual color temperature value Tc for reference in the color correction process;

[0033] The slope, texture and vegetation in terrain features are not well represented. The reason is that the Laplacian pyramid enhancement fails to highlight the complex terrain features, and the parameter adjustment does not fully consider the complexity of the terrain.

[0034] Improvement measures: For high-slope terrain, adjust the Laplace pyramid enhancement coefficient k to highlight the edges of terrain features and texture details.

[0035] Preferably, the real-time position and attitude information of the UAV is collected through GNSS and IMU devices, and the flight path planning is generated in combination with the coordinate points of the terrain target area. The flight path planning includes horizontal position calibration, target area coverage analysis and path overlap rate adjustment to ensure that the UAV covers the target area with high precision. The control system dynamically adjusts the flight altitude and angle according to the real-time environmental parameters to avoid interference from obstacles.

[0036] The flight status acquisition is based on the data fusion of the barometer, laser rangefinder and environmental sensor. The flight stability status is obtained through multi-dimensional data analysis, and the key parameters of the flight status are recorded, including the flight altitude Hf, wind speed Ws, wind direction Wd and flight inclination angle Ap;

[0037] The multi-spectral camera, light sensor and lidar system are used to collect terrain feature data, lighting condition data and image data. The data collected includes the light intensity Li, light uniformity Lu, slope change gradient Sg and elevation difference Hd of the target area. After basic verification and formatting, the collected data is passed to the preprocessing module as the system initial data.

[0038] Preferably, the collected image frame data is analyzed by an algorithm to determine the quality of images caused by flight altitude fluctuations, wind speed influence or abnormal lighting. Image frames with quality below the standard threshold will be marked as abnormal data frames and removed from the data set. The image data retained after interference removal is used for subsequent terrain feature analysis and processing;

[0039] The edge detection algorithm is used to preliminarily mark the peaks and gullies of the significant terrain features in the image. The terrain feature marking is automatically completed by calculating the local brightness variance Vl and the image gradient Eg. The marking results are stored in vector form. The marked feature information is used to assist in terrain feature retention analysis and detail enhancement.

[0040] Based on the flight status and image data, the flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf are calculated.

[0041] Preferably, the non-local means NLM algorithm is used to detect high-frequency noise points in the image and identify random noise introduced by the sensor or the external environment. Noise detection is based on pixel similarity weights combined with the original image gradient E g and resolution accuracy P r , mark the noise area;

[0042] For the identified high-frequency noise area, a smoothing algorithm is used to remove noise while retaining the edge and texture details of the image. Noise removal is achieved through the following formula:

[0043]

[0044] The weight w(p, q) is used for dynamic adjustment to ensure the adaptability of noise removal;

[0045] The image after noise removal is edge optimized to prevent blurred boundaries and ensure the integrity of the terrain feature contours of peaks and gullies. Edge optimization is combined with local contrast adjustment.

[0046] Preferably, the edge portion of the image is enhanced at multiple scales using the Laplace pyramid algorithm, and the edge enhancement is achieved using the following formula:

[0047] I enhanced =I d +k*(I d -Gaussian(I d ));

[0048] I enhanced represents the enhanced image, I d Represents the denoised image, Gaussian(I d ) indicates that I d The image after Gaussian blur, I d -Gaussian(I d ) represents the high-frequency information of the image, k represents the enhancement coefficient, and Tf represents the terrain feature retention coefficient;

[0049] Among them, the enhancement coefficient k is determined by the terrain feature retention coefficient Tf. The Tf in the calculation formula integrates slope gradient, vegetation coverage, texture complexity and elevation difference.

[0050] Preferably, the grayscale world hypothesis is used to adjust the color deviation of the image, correct the color distortion caused by the complex lighting conditions in the mountains, optimize the brightness of the shadow and highlight areas, avoid information loss through dynamic adjustment, and adjust the transition effect of light and shadow in combination with the illumination uniformity Lu parameter to make the image more natural. The overall image hue and brightness are fine-tuned to conform to the visual perception habits of the human eye and enhance the image's ornamental value and practicality.

[0051] The present invention provides an image enhancement processing method for UAV remote sensing mapping, which has the following beneficial effects:

[0052] (1) When the system is running, the flight attitude and position information are recorded through GNSS and IMU devices to control the flight path of the UAV. The collected original images are preliminarily processed to calculate and obtain: flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf. The non-local mean NLM algorithm is used to remove random noise in remote sensing images, optimize the edge area, and use the Laplace pyramid algorithm to improve the edge clarity and texture details of the terrain image, strengthen the mountain and hilly landform features, correct the color distortion caused by the complex lighting conditions in the mountains, make the image color truly restore the terrain features, avoid the loss of important landform information due to excessive brightness or darkness, optimize the terrain feature extraction, and generate the final index Z final ,Quality assessment of enhanced and rectified images.

[0053] (2) Through the implementation of the above six steps, the UAV remote sensing image enhancement processing method effectively completed the whole process from data collection to image optimization. Steps 1 and 2 accurately collected flight status data, lighting conditions, and terrain feature information through the multi-sensor equipment and precise algorithms carried by the UAV, and removed interference and preliminarily labeled the collected original images, laying a high-quality data foundation for subsequent processing; Steps 3 and 4 effectively suppressed the random noise of the image, and at the same time enhanced the edge clarity and texture details of terrain features such as peaks and gullies, making the features of complex terrain more prominent; Steps 5 and 6 further optimized the color distortion and light and shadow balance of the image, and used the quality evaluation comprehensive index Z to evaluate the image quality. final The calculation and comparison of the final processing results ensured the reliability and applicability of the final processing results. In general, this method achieved the accurate extraction and true restoration of mountain and hill terrain features, providing high-precision data support for terrain mapping and analysis.

[0054] (3) Compared with the current single-dimensional optimization or traditional image enhancement technology, this method has significant multi-dimensional comprehensive optimization advantages. First, by introducing the flight stability coefficient Fs, the image acquisition quality coefficient Iq and the terrain feature retention coefficient Tf, this method quantitatively analyzes and adjusts the image quality from multiple angles such as flight status, lighting conditions, and terrain features, avoiding the shortcomings of the existing technology in limiting the processing of acquired data; secondly, the non-local mean NLM algorithm is used to accurately identify and remove high-frequency noise, while effectively retaining the terrain edge and texture details; finally, through the combined application of the Laplacian pyramid algorithm and the grayscale world hypothesis, not only the edge clarity and detail expression of complex terrain are enhanced, but also the color distortion problem caused by the complexity of lighting is solved. These improvements have enabled this method to achieve a comprehensive improvement in surveying and mapping accuracy, image authenticity, and detail expression.

[0055] (4) This method has achieved significant improvements in multiple dimensions. First, in the image acquisition stage, the consistency of image acquisition and the multi-dimensional integrity of data are guaranteed through dynamic adjustment of the flight path and fusion of multi-sensor data. Second, in the image processing stage, the joint optimization of noise suppression and detail enhancement makes the edges of the surveyed images sharper, the texture clearer, and the complex terrain features truly restored. Finally, in the overall evaluation stage, the comprehensive index Z is used to final The calculation and comparison ensures the high quality of the output results and avoids the mapping distortion problems caused by noise, blur or color deviation in traditional methods. Overall, this method not only improves the observation accuracy and application effect of remote sensing images, but also provides more reliable data support and decision-making basis for the mapping and analysis of mountainous and hilly terrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The present invention is a schematic diagram of the steps of an image enhancement processing method for UAV remote sensing mapping. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Example 1

[0059] The present invention provides a method for enhancing the image processing of UAV remote sensing and mapping, please refer to Figure 1 , including the following steps:

[0060] Step 1: Use GNSS and IMU devices to record flight attitude and position information, control the flight path of the drone, adjust the altitude and angle, and ensure coverage of the target area to collect high-precision flight status, image sampling, lighting conditions, and terrain feature data;

[0061] Step 2: Perform preliminary processing on the collected original images, remove interference information, preliminarily mark the peaks and gullies in the images, and calculate and obtain: flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf;

[0062] Step 3: Use the non-local means (NLM) algorithm to identify high-frequency noise points and remove random noise from remote sensing images, while retaining terrain information, optimizing edge areas, and avoiding boundary blur.

[0063] Step 4: Use the Laplace pyramid algorithm to improve the edge clarity and texture details of the terrain image and enhance the features of mountainous and hilly landforms;

[0064] Step 5: Correct the color distortion caused by the complex lighting conditions in the mountains. Adjust the color balance of the image based on the grayscale world hypothesis and tone mapping, and adjust the brightness of the shadow and highlight areas so that the image color can truly restore the terrain characteristics and avoid the loss of important landform information due to excessive brightness or darkness.

[0065] Step 6: Optimize terrain feature extraction and generate the final index Z final ,Quality assessment of enhanced and rectified images.

[0066] In this embodiment, the flight attitude and position information are recorded by GNSS and IMU devices, the flight path of the UAV is controlled, the altitude and angle are adjusted to ensure coverage of the target area, high-precision flight status, image sampling, lighting conditions and terrain feature data are collected, the collected original images are preliminarily processed, interference information is eliminated, and the peaks and gullies in the image are preliminarily marked. The flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf are calculated and obtained, and the non-local mean NLM algorithm is used to identify high-frequency noise points, remove random noise in the remote sensing image, and retain terrain information. The edge area is optimized to avoid boundary blur. The edge clarity and texture details of the terrain image are improved by the Laplace pyramid algorithm, and the mountain and hilly landform features are enhanced. The color distortion caused by the complex mountain lighting conditions is corrected. The color balance of the image is adjusted based on the gray world hypothesis and tone mapping. The brightness of the shadow and highlight areas is adjusted so that the image color truly restores the terrain features, avoids the loss of important landform information due to excessive brightness or darkness, optimizes the terrain feature extraction, and generates the final index Z final ,Quality assessment of enhanced and rectified images.

[0067] Example 2

[0068] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: The real-time position and attitude information of the UAV is collected through GNSS and IMU devices, and the flight path planning is generated in combination with the coordinate points of the terrain target area. The flight path planning includes horizontal position calibration, target area coverage analysis and path overlap rate adjustment to ensure that the UAV covers the target area with high precision. The control system dynamically adjusts the flight altitude and angle according to the real-time environmental parameters to avoid interference from obstacles;

[0069] The flight status acquisition is based on the data fusion of the barometer, laser rangefinder and environmental sensor. The flight stability status is obtained through multi-dimensional data analysis, and the key parameters of the flight status are recorded, including the flight altitude Hf, wind speed Ws, wind direction Wd and flight inclination angle Ap;

[0070] The multi-spectral camera, light sensor and lidar system are used to collect terrain feature data, lighting condition data and image data. The data collected includes the light intensity Li, light uniformity Lu, slope change gradient Sg and elevation difference Hd of the target area. After basic verification and formatting, the collected data is passed to the preprocessing module as the system initial data.

[0071] The collected image frame data is analyzed by algorithms to determine the quality of images caused by flight altitude fluctuations, wind speed influence or abnormal lighting. Image frames with quality below the standard threshold will be marked as abnormal data frames and removed from the data set. The image data retained after interference removal is used for subsequent terrain feature analysis and processing;

[0072] The edge detection algorithm is used to preliminarily mark the peaks and gullies of the significant terrain features in the image. The terrain feature marking is automatically completed by calculating the local brightness variance Vl and the image gradient Eg. The marking results are stored in vector form. The marked feature information is used to assist in terrain feature retention analysis and detail enhancement.

[0073] Based on the flight status and image data, the flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf are calculated.

[0074] In this embodiment, through the implementation of this method, the accuracy and adaptability of UAV remote sensing mapping are significantly improved. First, the flight path planning generated by GNSS and IMU equipment combined with the coordinate points of the terrain target area ensures that the UAV can achieve high-precision coverage in complex terrain. The calibration of the horizontal position, the target area coverage analysis and the adjustment of the path overlap rate in the path planning effectively avoid the blind spots or redundancy of data collection, and significantly improve the efficiency of the mapping task. Secondly, the flight status acquisition is based on the data fusion analysis of the barometer, laser rangefinder and environmental sensor, which helps to monitor the key parameters such as flight altitude, wind speed, wind direction and inclination in real time, dynamically adjust the flight status, and ensure flight stability, thereby improving the acquisition quality and spatial consistency of image data. In addition, through the collaborative work of multi-spectral cameras, light sensors and laser radar systems, the system can comprehensively collect the terrain features, lighting conditions and elevation information of the target area, and through basic verification and formatting processing, ensure the accuracy and consistency of the data, and provide high-quality input for subsequent image preprocessing. On the whole, this method effectively solves the problems of unstable data acquisition and insufficient precision in complex terrain environments, and lays a solid foundation for high-precision mapping and subsequent image analysis.

[0075] Example 3

[0076] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: the flight stability coefficient Fs is calculated by the following formula:

[0077]

[0078] In the formula, H f Indicates the flight altitude, ΔH f It represents the fluctuation range of flight altitude, Havg represents the average flight altitude, Ws represents the wind speed, Wmax represents the maximum allowable wind speed, Ap represents the shooting inclination angle, k1 and k2 represent the weight coefficients of each parameter, respectively, which adjust the influence of wind speed and inclination angle on flight stability.

[0079] The image acquisition quality coefficient Iq is calculated by the following formula:

[0080]

[0081] Where Pr represents the resolution accuracy, P ideal Indicates the ideal resolution accuracy, L u Indicates the uniformity of illumination, L ideal Ideal illumination uniformity, L i Indicates light intensity, L minrepresents the minimum received light intensity, k3 and k4 represent the weight coefficients of each parameter, which are used to adjust the contribution of resolution and illumination to image acquisition quality.

[0082] The terrain feature retention coefficient Tf is calculated by the following formula:

[0083]

[0084] In the formula, S g Indicates the slope change gradient, S max Indicates the maximum slope gradient, V r represents the vegetation coverage, V max represents the maximum vegetation coverage, T c Indicates the texture complexity, T max represents the maximum texture complexity, H d Indicates the elevation difference, H max represents the maximum elevation difference, k5, k6, k7 and k8 represent the weight coefficients of each parameter, which are used to adjust the contribution of each parameter to the retention of terrain features.

[0085] In this embodiment, through the calculation and application of the flight stability coefficient Fs, the image acquisition quality coefficient Iq and the terrain feature retention coefficient Tf, the method realizes the multi-dimensional quantitative analysis of the UAV flight state, image acquisition quality and terrain feature retention ability, and significantly improves the comprehensive performance of the mapping task. First, the flight stability coefficient Fs ensures that the UAV maintains stable flight in a complex environment through multi-factor analysis such as flight altitude Hf, wind speed Ws, wind direction Wd and flight inclination Ap, which greatly reduces the image blur and acquisition deviation caused by unstable flight; secondly, the image acquisition quality coefficient Iq combines key parameters such as resolution accuracy Pr, illumination uniformity Lu and illumination intensity Li, ensures the clarity and illumination adaptability of the remote sensing image, and effectively avoids data distortion caused by insufficient lighting or resolution deviation; finally, the terrain feature retention coefficient Tf significantly enhances the ability of terrain mapping images to express complex terrain details through dynamic weighted analysis of multi-dimensional features such as slope gradient Sg, vegetation coverage Vr, texture complexity Tc and elevation difference Hd. Through the application of the above three comprehensive coefficients, this method can comprehensively improve the image quality, feature restoration accuracy and processing stability of UAV remote sensing mapping, providing a reliable guarantee for high-precision mountain and hill mapping.

[0086] Example 4

[0087] This embodiment is explained in Example 1, please refer to Figure 1Specifically: Use the non-local means NLM algorithm to detect high-frequency noise points in the image and identify random noise introduced by the sensor or external environment. Noise detection is based on pixel similarity weights combined with the original image gradient E g and resolution accuracy P r , mark the noise area;

[0088] For the identified high-frequency noise area, a smoothing algorithm is used to remove noise while retaining the edge and texture details of the image. Noise removal is achieved through the following formula:

[0089]

[0090] The weight w(p, q) is used for dynamic adjustment to ensure the adaptability of noise removal;

[0091] The image after noise removal is edge optimized to prevent blurred boundaries and ensure the integrity of the terrain feature contours of peaks and gullies. Edge optimization is combined with local contrast adjustment.

[0092] The Laplace pyramid algorithm is used to perform multi-scale enhancement on the edge of the image. The edge enhancement is achieved through the following formula:

[0093] I enhanced =I d +k*(I d -Gaussian(I d ));

[0094] I enhanced represents the enhanced image, I d Represents the denoised image, Gaussian(I d ) indicates that I d The image after Gaussian blur, I d -Gaussian(I d ) represents the high-frequency information of the image, k represents the enhancement coefficient, and Tf represents the terrain feature retention coefficient;

[0095] Among them, the enhancement coefficient k is determined by the terrain feature retention coefficient Tf. The Tf in the calculation formula integrates slope gradient, vegetation coverage, texture complexity and elevation difference.

[0096] The grayscale world hypothesis is used to adjust the color deviation of the image, correct the color distortion caused by the complex lighting conditions in the mountains, optimize the brightness of the shadow and highlight areas, avoid information loss through dynamic adjustment, and adjust the transition effect of light and shadow in combination with the illumination uniformity Lu parameter to make the image more natural. The overall image hue and brightness are fine-tuned to conform to the visual perception habits of the human eye and enhance the image's ornamental value and practicality.

[0097] In this embodiment, the non-local mean NLM algorithm is combined with the comprehensive analysis of pixel similarity weight, image gradient Eg and resolution accuracy Pr to achieve accurate detection and efficient denoising of high-frequency noise areas, retain the edge and texture details of the image, and significantly improve the image quality. On the basis of denoising, the Laplace pyramid algorithm is used to perform multi-scale enhancement on the edge of the image, extract and strengthen the high-frequency information of the image, and the enhancement coefficient k is dynamically adjusted according to the terrain feature retention coefficient Tf, so as to ensure that the edge clarity and detail expression of complex terrain (such as peaks and gullies) are comprehensively improved. At the same time, the grayscale world hypothesis is used to correct the image color, and the brightness transition of the shadow and highlight areas is optimized by combining the illumination uniformity Lu, which solves the color distortion problem under complex illumination conditions, making the image more natural and realistic. On the whole, this method has achieved remarkable results in improving image quality, strengthening terrain feature expression, and optimizing viewing and practicality, providing reliable technical support and high-quality data guarantee for remote sensing mapping of mountainous and hilly terrain.

[0098] Example 5

[0099] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: the image quality is quantified by the clarity coefficient, color balance coefficient and terrain detail retention coefficient, and the comprehensive evaluation index Z final Calculated by the following formula:

[0100] Z final =α*Fs+β*Iq+γ*Tf;

[0101] Where, Fs represents the flight stability coefficient, Iq represents the image acquisition quality coefficient, Tf represents the terrain feature retention coefficient, α, β and γ represent the weight values ​​of the flight stability coefficient Fs, the image acquisition quality coefficient Iq and the terrain feature retention coefficient Tf respectively;

[0102] The enhanced and corrected terrain features, including slope distribution, vegetation coverage and texture feature information, are extracted and evaluated by the comprehensive evaluation index Z final Compare with the preset standard threshold Q to determine whether the image processing quality meets expectations, and formulate targeted improvement measures to improve the overall image quality and the accuracy of terrain mapping results, and obtain the comparison strategy:

[0103] When the comprehensive evaluation index Z final ≥Preset standard threshold Q, indicating that the image quality meets the standard, the processing effect is good, no further optimization is required, and the output results include image data and terrain feature extraction analysis report;

[0104] When the comprehensive evaluation index Z final<Preset standard threshold Q, the image quality is lower than the standard, and it is necessary to further analyze the cause of the problem and take improvement measures;

[0105] The problems to be further analyzed include:

[0106] The image edges are blurred or details are missing. Reason: The flight stability is unqualified, resulting in blurred captured images, and the noise suppression or detail enhancement links are not fully optimized;

[0107] Improvement measures: Increase the stability of the flight path, reduce the flight height fluctuation ΔHf and the shooting inclination angle Ap, optimize the flight speed, and avoid blurred images caused by too fast flight;

[0108] The image has obvious color deviation or uneven light and shadow. Reason: The color correction process does not accurately adjust the light or color temperature deviation, and the light and shadow balance does not reach the expected value;

[0109] Improvement measures: Enhance the accuracy of the color temperature sensor, record the actual color temperature value Tc, and use it as a reference for the color correction link;

[0110] The slope, texture, and vegetation in the terrain features are insufficient. Reason: The Laplacian pyramid enhancement fails to highlight complex terrain features, and the parameter adjustment does not fully consider the terrain complexity;

[0111] Improvement measures: For high-slope terrains, adjust the Laplacian pyramid enhancement coefficient k to highlight the terrain feature edges and texture details.

[0112] In this embodiment, through the comprehensive evaluation index Z final By comparing the quantitative calculation of and the preset standard threshold Q, this method realizes the comprehensive evaluation and dynamic optimization of the image processing quality. First, based on the weight integration of the flight stability coefficient Fs, the image acquisition quality coefficient Iq, and the terrain feature retention coefficient Tf, this method can quantify the clarity, color balance, and terrain detail performance of the image, ensuring that the comprehensive quality of the mapping image meets the standard in multiple dimensions. Second, through the determination mechanism of Z final ≥Q, when the image quality meets the expectation, directly output high-quality image data and terrain feature extraction analysis reports, significantly improving the efficiency of the mapping task; while when Z final <Q, by analyzing the sources of problems such as blurred image edges, color deviation, uneven light and shadow, and insufficient terrain detail performance, dynamically adjust the flight path, optimize the noise suppression and detail enhancement algorithms, and improve the color correction and light and shadow balance strategies, effectively improving the overall quality of the image and the expression accuracy of terrain features. Overall, this method not only overcomes the limitations of single-dimensional optimization in traditional technical means but also provides full-process quality assurance and continuous optimization capabilities for high-precision mapping of mountainous and hilly terrains

[0113] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for enhancing image processing of UAV remote sensing mapping, characterized in that: The following steps are involved: Step 1: Use GNSS and IMU devices to record flight attitude and position information, control the flight path of the drone, adjust the altitude and angle, and ensure coverage of the target area to collect high-precision flight status, image sampling, lighting conditions, and terrain feature data; Step 2: Perform preliminary processing on the collected original images, remove interference information, preliminarily mark the peaks and gullies in the images, and calculate and obtain: flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf; The flight stability coefficient Fs is calculated by the following formula: ； In the formula, H f Indicates the flight altitude, ΔH f Indicates the flight altitude fluctuation range, Havg indicates the average flight altitude, Ws indicates the wind speed, Wmax indicates the maximum allowable wind speed, Ap indicates the shooting inclination angle, k1 and k2 indicate the weight coefficients of each parameter, respectively, to adjust the effect of wind speed and inclination angle on flight stability; The image acquisition quality coefficient Iq is calculated by the following formula: ； Where Pr represents the resolution accuracy, P ideal Indicates the ideal resolution accuracy, L u Indicates the uniformity of illumination, L ideal Ideal illumination uniformity, L i Indicates light intensity, L min represents the minimum received light intensity, k3 and k4 represent the weight coefficients of each parameter, which are used to adjust the contribution of resolution and light to image acquisition quality; The terrain feature retention coefficient Tf is calculated by the following formula: ； In the formula, S g Indicates the slope change gradient, S max Indicates the maximum slope gradient, V r represents the vegetation coverage, V max represents the maximum vegetation coverage, T c Indicates the texture complexity, T max represents the maximum texture complexity, H d Indicates the elevation difference, H max represents the maximum elevation difference, k5, k6, k7 and k8 represent the weight coefficients of each parameter, which are used to adjust the contribution of each parameter to the retention of terrain features; Step 3: Use the non-local means (NLM) algorithm to identify high-frequency noise points and remove random noise from remote sensing images, while retaining terrain information, optimizing edge areas, and avoiding boundary blur. Step 4: Use the Laplace pyramid algorithm to improve the edge clarity and texture details of the terrain image and enhance the features of mountainous and hilly landforms; Step 5: Correct the color distortion caused by the complex lighting conditions in the mountains. Adjust the color balance of the image based on the grayscale world hypothesis and tone mapping, and adjust the brightness of the shadow and highlight areas so that the image color can truly restore the terrain characteristics and avoid the loss of important landform information due to excessive brightness or darkness. Step 6: Optimize terrain feature extraction and generate the final index Z final ,quality assessment of enhanced and rectified images; The image quality is quantified by the clarity coefficient, color balance coefficient and terrain detail retention coefficient, and the comprehensive evaluation index Z is used. final Calculated by the following formula: ； In the formula, Fs represents the flight stability coefficient, Iq represents the image acquisition quality coefficient, and Tf represents the terrain feature retention coefficient. They represent the weight values ​​of the flight stability coefficient Fs, the image acquisition quality coefficient Iq and the terrain feature retention coefficient Tf respectively; The enhanced and corrected terrain features, including slope distribution, vegetation coverage and texture feature information, are extracted and evaluated by the comprehensive evaluation index Z final Compare with the preset standard threshold Q to determine whether the image processing quality meets expectations, and formulate targeted improvement measures to improve the overall quality of the image and the accuracy of the terrain mapping results, and obtain the comparison strategy: When the comprehensive evaluation index Z final ≥Preset standard threshold Q, indicating that the image quality meets the standard, the processing effect is good, no further optimization is required, and the output results include image data and terrain feature extraction analysis report; When the comprehensive evaluation index Z final <Preset standard threshold Q, the image quality is below the standard, and further analysis of the cause of the problem is required, and corresponding improvement measures should be taken.

2. The method for enhancing image processing of UAV remote sensing mapping according to claim 1, characterized in that: The real-time position and attitude information of the UAV is collected through GNSS and IMU devices, and the flight path planning is generated in combination with the coordinate points of the terrain target area. The flight path planning includes horizontal position calibration, target area coverage analysis and path overlap rate adjustment to ensure that the UAV covers the target area with high precision. The control system dynamically adjusts the flight altitude and angle according to the real-time environmental parameters to avoid interference from obstacles. The flight status acquisition is based on the data fusion of the barometer, laser rangefinder and environmental sensor. The flight stability status is obtained through multi-dimensional data analysis, and the key parameters of the flight status are recorded, including the flight altitude Hf, wind speed Ws, wind direction Wd and flight inclination angle Ap; The multi-spectral camera, light sensor and lidar system are used to collect terrain feature data, lighting condition data and image data. The data collected includes the light intensity Li, light uniformity Lu, slope change gradient Sg and elevation difference Hd of the target area. After basic verification and formatting, the collected data is passed to the preprocessing module as the system initial data.

3. The method for enhancing image processing of UAV remote sensing mapping according to claim 1, characterized in that: The collected image frame data is analyzed by algorithms to determine the quality of images caused by flight altitude fluctuations, wind speed influence or abnormal lighting. Image frames with quality below the standard threshold will be marked as abnormal data frames and removed from the data set. The image data retained after interference removal is used for subsequent terrain feature analysis and processing; The edge detection algorithm is used to preliminarily mark the peaks and gullies of the significant terrain features in the image. The terrain feature marking is automatically completed by calculating the local brightness variance Vl and the image gradient Eg. The marking results are stored in vector form. The marked feature information is used to assist in terrain feature retention analysis and detail enhancement. Based on the flight status and image data, the flight stability coefficient Fs, image acquisition quality coefficient Iq and terrain feature retention coefficient Tf are calculated.

4. The method for enhancing image processing of UAV remote sensing mapping according to claim 1, characterized in that: The non-local means NLM algorithm is used to detect high-frequency noise points in the image and identify random noise introduced by the sensor or the external environment. Noise detection is based on pixel similarity weights combined with the original image gradient E g and resolution accuracy P r , mark the noise area; For the identified high-frequency noise area, a smoothing algorithm is used to remove noise while retaining the edge and texture details of the image. Noise removal is achieved through the following formula: ； The weight w(p, q) is used for dynamic adjustment to ensure the adaptability of noise removal; The image after noise removal is edge optimized to prevent blurred boundaries and ensure the integrity of the terrain feature contours of peaks and gullies. Edge optimization is combined with local contrast adjustment.

5. The method for enhancing image processing of UAV remote sensing mapping according to claim 1, characterized in that: The Laplace pyramid algorithm is used to perform multi-scale enhancement on the edge of the image. The edge enhancement is achieved through the following formula: ； I enhanced represents the enhanced image, I d Represents the denoised image, Gaussian(I d ) indicates that I d The image after Gaussian blur, I d −Gaussian(I d ) represents the high-frequency information of the image, k represents the enhancement coefficient, and Tf represents the terrain feature retention coefficient; Among them, the enhancement coefficient k is determined by the terrain feature retention coefficient Tf. The Tf in the calculation formula integrates slope gradient, vegetation coverage, texture complexity and elevation difference.

6. The method for enhancing image processing of UAV remote sensing mapping according to claim 1, characterized in that: The grayscale world hypothesis is used to adjust the color deviation of the image, correct the color distortion caused by the complex lighting conditions in the mountains, optimize the brightness of the shadow and highlight areas, avoid information loss through dynamic adjustment, and adjust the transition effect of light and shadow in combination with the illumination uniformity Lu parameter to make the image more natural. The overall image hue and brightness are fine-tuned to conform to the visual perception habits of the human eye and enhance the image's ornamental value and practicality.

Images