Unmanned aerial vehicle remote sensing image processing method for ecological resource monitoring

Abstract

The invention discloses an unmanned aerial vehicle remote sensing image processing method for ecological resource monitoring, and belongs to the technical field of unmanned aerial vehicle remote sensing, and the method comprises the steps: S10, carrying out the image collection of an ecological resource region, obtaining the region area of the ecological resource region before the image collection, and carrying out the image collection based on the region area; acquiring position positioning information of the unmanned aerial vehicle in the image acquisition process; step S20, preprocessing the collected image, wherein the preprocessing comprises image denoising, image enhancement and image splicing; wherein in the image splicing process, splicing is carried out by taking at least two images collected by the unmanned aerial vehicle as a splicing unit, and the two images are the images collected when the unmanned aerial vehicle flies in the same direction. The reliability of ecological resource monitoring is improved, the adaptability to the complex ecological environment and the sensitivity to ecological resource changes are high, and the requirement of modern society for ecological resource monitoring is met.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) remote sensing technology, and in particular to a UAV remote sensing image processing method for ecological resource monitoring. Background Technology

[0002] With increasing awareness of ecological and environmental protection, the demand for monitoring ecological resources is growing. Unmanned aerial vehicle (UAV) remote sensing technology, due to its advantages such as high flexibility, low cost, and wide coverage, has been widely used in ecological resource monitoring.

[0003] Regarding this research, application CN202111139251.4 provides a method for enhancing images from UAV remote sensing mapping. This technical solution includes a low-pass filter module, a high-pass filter module, and an image integration module. The low-pass filter module is used to block or reduce high-frequency signals exceeding a set threshold, and the high-pass filter module is used to block or reduce high-frequency signals below a set threshold. The steps are as follows: Step 1: Divide the acquired image into several equal intervals. In the above solution, the image is first split, and image enhancement is performed on each interval. Then, based on the image content, when low-pass filtering is needed, the low-pass filter module is used to perform low-pass filtering to remove edges and noise. This technical solution can enhance images for different situations, and the image enhancement effect is significant.

[0004] Another application, CN202510173668.4, provides a method for processing remote sensing images from unmanned aerial vehicles (UAVs). This method includes the following steps: S1: acquiring image data of the target area to obtain a set of images to be processed; S2: performing grayscale mapping on the set of images to be processed to obtain an initial grayscale image set, and then denoising the initial grayscale image set by introducing median filtering with a window weight factor; S3: calculating and processing the denoised grayscale image using inverse grayscale transformation and color channel restoration algorithms to obtain a color-enhanced image; S4: deblurring the color-enhanced image, and calculating the grayscale principal value region representation, a fixed segmentation threshold, and ambient light interference based on the denoised grayscale image and the color-enhanced image. This method uses the Harris matrix and FFT to form feature descriptors and performs a two-stage threshold determination, which is beneficial for accurate registration to obtain panoramic images.

[0005] However, the above-mentioned technical solutions are poorly adapted to complex ecological environments. In complex terrain or densely vegetated areas, these solutions are prone to errors during image stitching and analysis, which limits the accuracy and reliability of monitoring results and makes it difficult to meet the needs of ecological resource monitoring. Summary of the Invention

[0006] In view of the problems existing in the field of UAV remote sensing technology, the present invention is proposed.

[0007] Therefore, one of the objectives of this invention is to provide a UAV remote sensing image processing method for ecological resource monitoring. Through steps such as image stitching, feature extraction, and edge spacing calculation, it comprehensively analyzes ecological resource areas, achieves efficient and accurate dynamic monitoring of ecological resource areas, improves the reliability of ecological resource monitoring, has strong adaptability to complex ecological environments and strong sensitivity to changes in ecological resources, and meets the needs of modern society for ecological resource monitoring.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a UAV remote sensing image processing method for ecological resource monitoring, comprising the following steps: Step S10: Collect images of the ecological resource area. Before collecting the images, obtain the area of ​​the ecological resource area. Based on the area, obtain the location information of the UAV during the image collection process. Step S20: Preprocess the acquired images, including image denoising, image enhancement, and image stitching; wherein, in the image stitching, each stitching unit consists of at least two images acquired by the UAV, and the two images are those acquired by the UAV when flying in the same direction; Step S30: Based on the images collected by the UAV while flying in the same direction, obtain relevant information about the images, including corner points and edge points of the images; obtain the location information of the UAV when collecting the images based on the relevant information, and generate an image database based on the images; classify and identify the ecological resources in the ecological resource area in the image database; the ecological resources include forests, grasslands, water bodies and wetlands; Step S40: Divide the corner points into , ,..., , Indicates the first division The system identifies corner points and, based on these edge points, obtains an edge segment between any two corner points that is equidistant from the two corner points; and acquires the ecological resources of the edge segment. Step S50: Obtain the location corresponding to the location information in the ecological resource area. If the UAV collects images at the location in the future, compare the ecological resources of the collected images with the ecological resources of the edge segment. If the ecological resources of the collected images are the same as the ecological resources of the edge segment, it is determined that the ecology of the ecological resource area is in a normal state; otherwise, it is determined that the ecology of the ecological resource area has been damaged.

[0009] In a preferred embodiment of the present invention, step S20 further includes image segmentation and feature extraction of the acquired image, wherein... The image segmentation includes segmenting the image using a region growing algorithm and combining relevant features; the segmentation includes separating the ecological resource region from the background region; the relevant features include the spectral and texture features of the ecological resource region. The feature extraction includes extracting corresponding feature information for different ecological resource types in the ecological resource region; The types of ecological resources include forest resources and water resources; The forest resources include the extraction of vegetation indices and texture features; The water resources include the extraction of water indexes and spectral reflectance characteristics.

[0010] In a preferred embodiment of the present invention, in step S30, the image stitching method includes an optical flow-based stitching method, and the steps include optical flow estimation, image alignment, and image fusion. The optical flow estimation is to obtain the relative motion relationship between images by calculating the motion vectors of pixels in the image; The image alignment is achieved by translating and rotating the image based on the calculated motion vectors to align the images. The image fusion refers to merging the aligned images to generate a panoramic image; Image stitching methods also include stitching methods based on multi-scale fusion, the steps of which include multi-scale decomposition, sub-band image stitching and multi-scale fusion; The multi-scale decomposition refers to decomposing an image into sub-band images of multiple scales. The sub-band image stitching involves stitching each sub-band image separately to generate a multi-scale stitching result; The multi-scale fusion refers to fusing the stitching results of multiple scales to generate a panoramic image.

[0011] In a preferred embodiment of the present invention, in the optical flow estimation, the motion vectors of pixels in the image are calculated using the Horn-Schunck method. ; In the formula, Let be the brightness function of the image, representing the time... At time, position and Pixel values; The energy function represents the error in optical flow estimation. The weights represent the smoothing terms and are used to control the degree of smoothness of the optical flow. , and These represent the brightness function of the image at different positions. , and time The partial derivative at time .

[0012] In a preferred embodiment of the present invention, in step S40, the orientation of the edge segment in the image is obtained, and the image is marked as a reference image; at least four monitoring points are given in the edge segment, the image edges corresponding to the four monitoring points are obtained, the edge spacing of the image edges of adjacent monitoring points is calculated, time series analysis is performed based on the edge spacing to obtain the changes of the edge segment at different time points, and the calculated edge spacing is marked as a reference edge spacing; Calculate the edge distance between adjacent monitoring points in the image, including calculations based on the edge distance formula derived from image features: ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; and These represent the timestamps of any two adjacent monitoring points; This represents a weighting factor, which is used to adjust the impact of the timestamp on the edge spacing.

[0013] In a preferred embodiment of the present invention, the calculation of the edge spacing between adjacent monitoring points further includes calculating the edge spacing according to a time-series-based edge spacing formula. ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; and These represent the timestamps of any two adjacent monitoring points; This represents a weighting factor, which is used to adjust the impact of the timestamp on the edge spacing.

[0014] In a preferred embodiment of the present invention, when the UAV performs image acquisition at a position corresponding to the edge segment in the future, a monitoring point with the same reference coordinates is given at the edge segment of the acquired image, the image edge of the monitoring point is obtained, and the edge distance between the two image edges is calculated. If the calculated edge distance is less than the reference edge distance, it is determined that the ecological growth of the ecological resource area is in an abnormal state; otherwise, no determination is made.

[0015] In a preferred embodiment of the present invention, the acquired image is verified based on the determination result, and the verification steps include: Calculate the distance from the center of the reference image to the edge segment; Once a monitoring point with the same reference coordinates is given for the edge segment of the acquired image, the distance from the center of the image to the edge segment is calculated; If the distance is the same as the distance from the center to the edge of the reference image, then the verification result is determined to be correct. If the distance is not the same as the distance from the center to the edge of the reference image, the verification result is determined to be incorrect; at the same time, the determination of the ecological growth status of the ecological resource area is revoked.

[0016] A terminal includes a processor, an input interface, an output interface, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the method described above.

[0017] A computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described above.

[0018] Beneficial effects: 1. This invention, through steps such as image stitching, feature extraction, and edge spacing calculation, as well as through time series analysis of edge segment changes, can more accurately monitor the dynamic changes of ecological resources. Furthermore, in the image segmentation and feature extraction stages, by combining multiple feature information such as spectral features and texture features, it improves the ability to identify different types of ecological resources and enhances the reliability of monitoring results. 2. This invention verifies the acquired images by verifying the distance from the center to the edge of the image, thereby ensuring the accuracy of image acquisition and analysis and avoiding misjudgments caused by image acquisition errors. 3. This invention combines multiple image processing and analysis methods (such as region growing algorithms, optical flow estimation, multi-scale fusion, etc.), enabling it to better adapt to complex ecological environments. For example, it can analyze tree growth in forest areas with greater detail, and monitor water changes more accurately in water bodies. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the process structure of an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0021] Because existing technical solutions are poorly adapted to complex ecological environments, errors are easily introduced during image stitching and analysis in complex terrain or densely vegetated areas, which limits the accuracy and reliability of monitoring results and makes it difficult to meet the needs of ecological resource monitoring.

[0022] Based on this, the present invention proposes a UAV remote sensing image processing method for ecological resource monitoring. By comprehensively applying a variety of image processing and analysis methods, it achieves efficient, accurate and dynamic monitoring of ecological resource areas, improves the reliability of ecological resource monitoring, and enhances adaptability to complex ecological environments and sensitivity to changes in ecological resources.

[0023] The present solution will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0024] Reference Figures 1 to 2 This is one embodiment of the present invention, which provides a UAV remote sensing image processing method for ecological resource monitoring, including the following steps: Step S10: Collect images of the ecological resource area. Before collecting images, obtain the area of ​​the ecological resource area. Based on the area, obtain the location information of the UAV during the image collection process. In this embodiment, in one feasible implementation, a drone equipped with a high-resolution multispectral camera and a high-precision positioning device is used to collect images of the ecological resource area. The multispectral camera can acquire image data in multiple bands such as visible light and near infrared, providing rich information for subsequent ecological resource identification and analysis. At the same time, the high-precision positioning device can ensure that the collected images have accurate geographical location information. It should be noted that by obtaining the area of ​​the ecological resource area and the location information of the UAV, the comprehensiveness and accuracy of image acquisition can be ensured. This helps to avoid omissions or duplications in the image acquisition process and improve the efficiency of image acquisition. Furthermore, the location information provides an important reference for subsequent image processing and analysis, enabling close integration of image data with geographic location information, which facilitates geospatial analysis and precise monitoring of ecological resources. Step S20: Preprocess the acquired images, including image denoising, image enhancement and image stitching; wherein, in the image stitching, each stitching unit consists of at least two images acquired by the UAV, and the two images are those acquired by the UAV when flying in the same direction; In this embodiment, the image denoising uses a wavelet transform-based denoising algorithm to denoise the acquired image, remove noise interference from the image, and improve image quality. Image enhancement, through histogram equalization and contrast stretching techniques, enhances the contrast and detail of an image, making it clearer and easier for subsequent analysis and processing. Step S30: Based on the images collected by the UAV while flying in the same direction, obtain relevant information about the images, including corner points and edge points; obtain the location information of the UAV when collecting the images based on the relevant information, and generate an image database based on the images; classify and identify ecological resources in the image database; ecological resources include forests, grasslands, water bodies and wetlands; In this embodiment, by extracting corner and edge point information, rich feature data is provided for subsequent image analysis, which helps to more accurately identify and classify ecological resources; Classifying and identifying ecological resources can quickly distinguish different types of ecological resources, providing a scientific basis for ecological protection and management. Generate an image database to facilitate the management and querying of image data, and support long-term ecological monitoring and analysis; Step S40: Divide the corner points into , ,..., , Indicates the first division The system identifies corner points and, based on edge points, obtains edge segments between any two corner points that are equidistant from the two corner points; and acquires the ecological resources of these edge segments. In this embodiment, the division of edge segments helps to conduct more detailed analysis of local areas, and can promptly detect changes and anomalies in ecological resources. Step S50: Obtain the location corresponding to the location information in the ecological resource area. If the UAV collects images at the location in the future, compare the ecological resources of the collected images with the ecological resources of the edge segment. If the ecological resources of the collected images are the same as the ecological resources of the edge segment, it is determined that the ecology of the ecological resource area is in a normal state; otherwise, it is determined that the ecology of the ecological resource area has been damaged. In this embodiment, at a future time, the drone will collect images at the same location, compare the ecological resources of the collected images with the ecological resources of the edge segments, and determine whether the ecological status of the ecological resource area is normal. This enables dynamic monitoring of changes in ecological resources and timely detection of abnormal situations in ecological resources; It can provide early warning for the implementation of ecological protection measures, reduce the loss of ecological resources, and improve the timeliness and effectiveness of ecological protection; Step S20 further includes image segmentation and feature extraction of the acquired image, wherein, Image segmentation includes segmenting an image using a region growing algorithm combined with relevant features; segmentation includes separating ecological resource regions from background regions; relevant features include the spectral and textural features of the ecological resource regions; Feature extraction includes extracting relevant feature information for different types of ecological resources within an ecological resource region; Ecological resources include forest resources and water resources; In forest resources, this includes extracting vegetation indices and texture features; It should be noted that vegetation indices (such as the Normalized Difference Vegetation Index NDVI) and texture features (such as the second moment of the gray-level co-occurrence matrix and contrast) are also relevant. In water resources, this includes extracting water indices and spectral reflectance characteristics; It should be noted that water body indices (such as the Normalized Difference Water Index NDWI) are used. By extracting these features, a basis can be provided for subsequent classification and identification of ecological resources; In this embodiment, the region growing algorithm combines spectral and texture features to more accurately separate ecological resource areas from background areas, thereby improving the accuracy of image segmentation. Extracting corresponding feature information for different types of ecological resources enhances the ability to identify different ecological resources and improves the accuracy and reliability of monitoring. In step S30, the image stitching method includes an optical flow-based stitching method, and the steps include optical flow estimation, image alignment, and image fusion. Optical flow estimation is used to obtain the relative motion relationships between images by calculating the motion vectors of pixels in the image; Image alignment involves translating and rotating the image based on the calculated motion vectors to align the images. Image fusion is the process of merging aligned images to generate a panoramic image. It should be noted that it can process moving objects in images, making it suitable for images taken by drones during flight; Image stitching methods also include stitching methods based on multi-scale fusion, the steps of which include multi-scale decomposition, sub-band image stitching and multi-scale fusion; Multiscale decomposition involves decomposing an image into sub-band images of multiple scales. Subband image stitching involves stitching each subband image separately to generate a multi-scale stitched result; Multi-scale fusion is used to fuse the stitching results from multiple scales to generate a panoramic image. It should be noted that it can preserve high-frequency detail information of the image, improve the quality of the stitched image, and is suitable for images of different resolutions and different content; In this embodiment, the relative motion relationship between images is obtained through optical flow estimation, which enables more accurate image alignment and fusion, thereby improving the stitching accuracy. Multi-scale fusion technology can preserve high-frequency detail information of images, improve the quality of stitched images, and is suitable for images of different resolutions and complexities. In optical flow estimation, the motion vectors of pixels in the image are calculated using the Horn-Schunck method: ; In the formula, Let be the brightness function of the image, representing the time... At time, position and Pixel values; The energy function represents the error in optical flow estimation. The weights represent the smoothing terms and are used to control the degree of smoothness of the optical flow. , and These represent the brightness function of the image at different positions. , and time Partial derivatives at time; The Horn-Schunck method is a global optical flow estimation method that assumes smooth motion across the entire image. This method estimates optical flow by minimizing the following energy function; This method solves through iteration. and until the energy function Reaching the minimum value; In this embodiment, the smoothness of optical flow is controlled by the weight of the smoothing term, thereby reducing the impact of noise and improving the stability of optical flow estimation. In optical flow estimation, the motion vectors of pixels in an image are calculated. This also includes calculating the motion vectors of pixels in an image using a deep learning method, by training a convolutional neural network (CNN), as shown below: ; In the formula, and This represents two consecutive frames of images; This represents a Convolutional Neural Network (CNN), used to learn the optical flow relationships between images through training; and Indicates the position of the pixel and The speed of movement on the surface This formula learns an optical flow estimation model by training a large number of image pairs and directly outputs the optical flow vector at test time; In step S40, the orientation of the edge segment in the image is obtained, and the image is marked as a reference image; and given at least 4 monitoring points in the edge segment, the image edges corresponding to the 4 monitoring points are obtained, the edge spacing of the image edges of adjacent monitoring points is calculated, time series analysis is performed based on the edge spacing to obtain the changes of the edge segment at different time points; and the calculated edge spacing is marked as the reference edge spacing. Calculate the edge distance between adjacent monitoring points in the image, including calculations based on the edge distance formula derived from image features: ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; the two coordinates are then marked as reference coordinates. and These represent the image feature values ​​of any two adjacent monitoring points; It should be noted that image feature values ​​include texture and color; This represents the weighting factor, which is used to adjust the influence of image feature values ​​on edge spacing. In this embodiment, changes in edge segments are detected to promptly identify trends and anomalies in ecological resources; for example, monitoring dynamic changes such as forest growth and water level fluctuations. Calculating the edge distance between adjacent monitoring points in the image also includes calculating it based on the edge distance formula derived from the time series: ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; and These represent the timestamps of any two adjacent monitoring points; This represents the weighting factor, which is used to adjust the impact of the timestamp on the edge spacing. By combining image features and timestamps to calculate edge spacing, the accuracy and reliability of edge spacing calculation are improved, enabling dynamic monitoring of edge segment changes and timely detection of ecological resource change trends and anomalies. When the UAV collects images at the location corresponding to the edge segment in the future, a monitoring point with the same reference coordinates is given at the edge segment of the collected image. The image edge of the monitoring point is obtained, and the edge distance between the two image edges is calculated. If the calculated edge distance is less than the reference edge distance, it is determined that the ecological growth of the ecological resource area is in an abnormal state; otherwise, no judgment is made. It should be noted that if the edge spacing increases with time and the rate of increase conforms to the growth pattern of ecological resources, it indicates that the ecological growth is good. If the edge spacing remains relatively stable over time without significant fluctuations, it indicates stable ecological growth. If the edge spacing changes abnormally over time (such as a sudden increase or decrease), it may indicate that the ecological resource area has been affected in some way (such as by pests and diseases, environmental pollution, etc.). For example, in forest areas, the growth of trees can be monitored by calculating the spacing of the tree crown edges. If the spacing of the tree crown edges increases over time and the rate of increase is in line with the growth pattern of trees, it indicates that the trees are growing well. If the spacing of the tree crown edges changes abnormally over time, such as suddenly decreasing, it may indicate that the trees have been affected by pests or diseases or have been cut down. The acquired images are verified based on the judgment results. The verification steps include: Calculate the distance from the center of the reference image to the edge segment; Once a monitoring point with the same reference coordinates is given for the edge segment of the acquired image, the distance from the center of the image to the edge segment is calculated; If the distance is the same as the distance from the center to the edge of the reference image, then the verification result is considered correct. If the distance is not the same as the distance from the center to the edge of the reference image, the verification result is deemed incorrect; at the same time, the determination of the ecological growth status of the ecological resource area is revoked. In this embodiment, a verification mechanism is used to ensure the accuracy of image acquisition and analysis, avoid misjudgment caused by image acquisition errors, and improve the reliability of monitoring results. A terminal includes a processor, an input interface, an output interface, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the method described above.

[0025] A computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described above.

[0026] In summary, this application comprehensively analyzes ecological resource areas through steps such as image stitching, feature extraction, and edge spacing calculation, achieving efficient and accurate dynamic monitoring of ecological resource areas, improving the reliability of ecological resource monitoring, and demonstrating strong adaptability to complex ecological environments and sensitivity to changes in ecological resources, thus meeting the needs of modern society for ecological resource monitoring.

[0027] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for processing unmanned aerial vehicle (UAV) remote sensing images for ecological resource monitoring, characterized in that, Includes the following steps: step S10: Image acquisition is performed on the ecological resource area. Before the image acquisition, the area of ​​the ecological resource area is obtained. Based on the area, the location information of the UAV is obtained during the image acquisition process. Step S20: Preprocess the acquired images, including image denoising, image enhancement, and image stitching; wherein, in the image stitching, each stitching unit consists of at least two images acquired by the UAV, and the two images are those acquired by the UAV when flying in the same direction; Step S30: Based on the images collected by the UAV while flying in the same direction, obtain relevant information about the images, including corner points and edge points of the images; obtain the location information of the UAV when collecting the images based on the relevant information, and generate an image database based on the images; classify and identify the ecological resources in the ecological resource area in the image database; the ecological resources include forests, grasslands, water bodies and wetlands; Step S40: Divide the corner points into , ,..., , Indicates the first division The system identifies corner points and, based on these edge points, obtains an edge segment between any two corner points that is equidistant from the two corner points; and acquires the ecological resources of the edge segment. Step S50: Obtain the location corresponding to the location information in the ecological resource area. If the UAV performs image acquisition at the location in the future, compare the ecological resources of the acquired image with the ecological resources of the edge segment. If the ecological resources of the acquired image are the same as the ecological resources of the edge segment, it is determined that the ecology of the ecological resource area is in a normal state. Conversely, it is determined that the ecology of the ecological resource area has been damaged.

2. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 1, characterized in that, In step S20, the method further includes image segmentation and feature extraction of the acquired image, wherein, The image segmentation includes segmenting the image using a region growing algorithm and combining relevant features; the segmentation includes separating the ecological resource region from the background region; the relevant features include the spectral and texture features of the ecological resource region. The feature extraction includes extracting corresponding feature information for different ecological resource types in the ecological resource region; The types of ecological resources include forest resources and water resources; The forest resources include the extraction of vegetation indices and texture features; The water resources include the extraction of water indexes and spectral reflectance characteristics.

3. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 1, characterized in that, In step S30, the image stitching method includes an optical flow-based stitching method, and the steps include optical flow estimation, image alignment, and image fusion. The optical flow estimation is to obtain the relative motion relationship between images by calculating the motion vectors of pixels in the image; The image alignment is achieved by translating and rotating the image based on the calculated motion vectors to align the images. The image fusion refers to merging the aligned images to generate a panoramic image; Image stitching methods also include stitching methods based on multi-scale fusion, the steps of which include multi-scale decomposition, sub-band image stitching and multi-scale fusion; The multi-scale decomposition refers to decomposing an image into sub-band images of multiple scales. The sub-band image stitching involves stitching each sub-band image separately to generate a multi-scale stitching result; The multi-scale fusion refers to fusing the stitching results of multiple scales to generate a panoramic image.

4. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 3, characterized in that, In the optical flow estimation, the motion vectors of pixels in the image are calculated using the Horn-Schunck method: ； In the formula, Let be the brightness function of the image, representing the time... At time, position and Pixel values; The energy function represents the error in optical flow estimation. The weights represent the smoothing terms and are used to control the degree of smoothness of the optical flow. , and These represent the brightness function of the image at different positions. , and time The partial derivative at time .

5. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 1, characterized in that, In step S40, the orientation of the edge segment in the image is obtained, and the image is marked as a reference image; and at least 4 monitoring points are given in the edge segment, the image edges corresponding to the 4 monitoring points are obtained, the edge spacing of the image edges of adjacent monitoring points is calculated, and time series analysis is performed based on the edge spacing to obtain the changes of the edge segment at different time points; The calculated edge spacing is then marked as the reference edge spacing; Calculate the edge distance between adjacent monitoring points in the image, including calculations based on the edge distance formula derived from image features: ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; the two coordinates are marked as reference coordinates. and These represent the image feature values ​​of any two adjacent monitoring points; This represents a weighting factor, which is used to adjust the influence of the image feature values ​​on the edge spacing.

6. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 5, characterized in that, Calculating the edge distance between adjacent monitoring points in the image also includes calculating it based on the edge distance formula derived from the time series: ;in, This represents the edge distance between any two adjacent monitoring points; In the formula, and These represent the coordinates of any two adjacent monitoring points; and These represent the timestamps of any two adjacent monitoring points; This represents a weighting factor, which is used to adjust the impact of the timestamp on the edge spacing.

7. A method for processing unmanned aerial vehicle (UAV) remote sensing images for ecological resource monitoring as described in any one of claims 5 to 6, characterized in that, When the UAV collects images at the location of the image corresponding to the edge segment in the future, a monitoring point with the same reference coordinates is given at the edge segment of the collected image, the image edge of the monitoring point is obtained, and the edge distance between the two image edges is calculated. When the calculated edge distance is less than the reference edge distance, it is determined that the ecological growth of the ecological resource area is in an abnormal state. Conversely, no judgment is made.

8. The UAV remote sensing image processing method for ecological resource monitoring as described in claim 7, characterized in that, The acquired images are verified based on the judgment results. The verification steps include: Calculate the distance from the center of the reference image to the edge segment; Once a monitoring point with the same reference coordinates is given for the edge segment of the acquired image, the distance from the center of the image to the edge segment is calculated; If the distance is the same as the distance from the center to the edge of the reference image, then the verification result is determined to be correct. If the distance is not the same as the distance from the center to the edge of the reference image, the verification result is determined to be incorrect; at the same time, the determination of the ecological growth status of the ecological resource area is revoked.

9. A terminal, characterized in that, The system includes a processor, an input interface, an output interface, and a memory, which are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described in any one of claims 1 to 8.

Images