Artificial afforestation area extraction method based on high-resolution remote sensing image
Through high-resolution remote sensing images combined with machine learning and Hough transformation algorithm, the problems of low efficiency, high cost and insufficient accuracy of artificial afforestation monitoring in the existing technology are solved, and efficient and accurate monitoring effects are achieved, suitable for large-scale forestry Investigation and ecological monitoring.
Patent Information
- Application Number
- CN202411083018.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-22
AI Technical Summary
The prior art is inefficient, cost-effective and insufficient in monitoring artificial afforestation. Especially through supervision classification and manual outline methods, it is difficult to accurately distinguish between artificial forests and natural forests, which affects the objectivity and large-scale application of monitoring.
Using high-resolution remote sensing images combined with machine learning and Hough transformation algorithm, the forest land intelligent extraction model constructed by Swin transformer and UperNet is used to extract and distinguish the area of artificial afforestation. Canny edge detection and Hough transformation linear detection are used to identify row and queue features and calculate area To achieve efficient and accurate monitoring.
It improves monitoring efficiency and accuracy, reduces labor and capital costs, and provides more objective large-scale forestry surveys and ecological monitoring support.
Smart Images

Figure CN119006574A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the fields of remote sensing image processing technology, image analysis and classification technology, and specifically relates to a method for extracting artificial afforestation area based on high-resolution remote sensing images. Background Art
[0002] Afforestation is an important measure to address global climate change and protect the ecological environment. Monitoring the area and effect of afforestation is of great significance to achieving carbon sequestration goals, protecting biodiversity, and formulating scientific forestry policies. In particular, artificial afforestation has clear layout rules, and monitoring its coverage area can better evaluate the implementation effect and ecological benefits of afforestation projects.
[0003] Currently, there are two most commonly used methods for monitoring artificial afforestation. One is to conduct supervised classification through remote sensing images to classify forest land; the other is to conduct manual delineation based on remote sensing images to determine the planting areas for artificial afforestation through manual means.
[0004] When using supervised classification technology to distinguish forest land, the model has weak generalization ability for different regions and vegetation types, and is prone to classification errors. At the same time, there are great problems in distinguishing between artificial forests and natural forests, and it is impossible to accurately extract the distribution and area of artificial afforestation.
[0005] Using the manual delineation method, experts manually mark the artificial tree planting areas through visual interpretation of remote sensing images, accurately extract the distribution and area of artificial tree planting, and produce good results. However, the manual delineation method also has the following disadvantages:
[0006] (1) Low efficiency: Manual labeling is slow and cannot meet the monitoring needs of large areas.
[0007] (2) High subjectivity: Differences in experience and judgment among operators may lead to inconsistent results and affect the objectivity of the measurement.
[0008] (3) High cost: Large-scale manual interpretation requires a lot of manpower and financial investment and is difficult to sustain in the long term. Summary of the invention
[0009] The present invention aims to provide a technical method that utilizes high-resolution remote sensing images combined with machine learning and Hough transform algorithm, which can efficiently and accurately identify and measure the area of artificial afforestation, effectively distinguish between natural forest land and artificial forest land, and improve the efficiency of monitoring, analysis and management.
[0010] The technical solution of the present invention is as follows:
[0011] A method for extracting artificial afforestation area based on high-resolution remote sensing images, the method comprising:
[0012] Acquire remote sensing images and preprocess the remote sensing images;
[0013] The pre-processed remote sensing images are input into the forest land intelligent extraction model to obtain the distribution map of all forest lands;
[0014] Perform edge detection and Hough transform line detection on the distribution map of all the forest lands to obtain artificial tree planting areas and natural forest areas;
[0015] The areas of artificial afforestation areas and natural forest areas are calculated to obtain the statistical results of artificial afforestation and natural forest areas.
[0016] Furthermore, the forestland intelligent extraction model is constructed by Swin transformer and UperNet.
[0017] Furthermore, the construction method of the forest land intelligent extraction model is:
[0018] Obtain remote sensing image samples and forest land pixel-by-pixel delineation samples;
[0019] Preprocess the obtained sample data, and divide the preprocessed sample data into a training set and a data set;
[0020] The Swin transforme and UperNet are iteratively trained through the training set to obtain the forest intelligent extraction model, and then the forest intelligent extraction model is verified through the verification set.
[0021] Furthermore, the preprocessing is:
[0022] First, the remote sensing image is geometrically corrected to remove image distortion caused by the sensor or satellite attitude, then radiation correction is performed to adjust the image brightness and contrast, and finally the image is enhanced using contrast stretching and sharpening techniques to improve image quality and texture features.
[0023] Furthermore, the distribution map of all the forest lands is subjected to edge detection and Hough transform line detection to obtain the artificial tree planting area and the natural forest area, which are specifically:
[0024] Grayscale processing is performed on the forest areas extracted from the distribution map of all forests, and the color images are converted into grayscale images;
[0025] Use the Canny edge detection algorithm to detect the edge of the grayscale image and extract the significant edges in the image;
[0026] Set the accumulator threshold and straight line length threshold of the Hough transform algorithm to detect straight lines in the image that meet the characteristics of artificial tree planting rows and columns;
[0027] According to the straight line distribution density and arrangement rules, the areas with obvious row and column characteristics are identified as artificial afforestation areas, and the rest are natural forest areas.
[0028] Furthermore, the areas of artificial afforestation and natural forests were calculated and the statistical results of artificial afforestation and natural forest areas were obtained as follows:
[0029] Count the number of pixels in the artificial tree planting area and the natural forest area respectively;
[0030] Based on the resolution of remote sensing images, the number of pixels is converted into actual area to obtain the statistical results of artificial afforestation and natural forest areas.
[0031] Furthermore, the remote sensing image accuracy is greater than or equal to 0.5 meters.
[0032] Compared with the prior art, the present invention has the following advantages:
[0033] The present invention can solve the shortcomings of traditional monitoring methods in terms of efficiency, cost and accuracy, and provide technical support for large-scale forestry surveys and ecological monitoring. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the embodiments of the invention. Where appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.
[0035] Figure 1 A schematic diagram of the method steps of the present invention is shown;
[0036] Figure 2 A schematic diagram showing the comparison of 0.5 meter and 1 meter remote sensing images of the present invention is shown;
[0037] Figure 3 The Hough transform principle diagram of the present invention is shown;
[0038] Figure 4 A schematic diagram of the division structure of natural forest land and artificial afforestation forest land of the present invention is shown. DETAILED DESCRIPTION
[0039] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0040] The overall idea of the present invention is: based on remote sensing images of 0.5 meters or better than 0.5 meters, remote sensing images of the research area are obtained after preprocessing. Then, typical areas are selected to make forest samples for training, and a semantic segmentation method based on deep learning technology is used to achieve pixel-level recognition of forests on 0.5m images to obtain forest identification results. Finally, the Hough transform algorithm is used to perform straight line detection to distinguish the artificial afforestation areas in the forest identification results.
[0041] 1. Remote sensing images
[0042] The remote sensing image accuracy used in the present invention needs to reach or exceed 0.5 meters. A large number of experiments have shown that remote sensing images with an accuracy lower than 0.5 meters cannot reflect the regularity that artificial afforestation is planted in rows and there is a clear distance between rows. When the Hough transform algorithm is used, the straight line detection effect is poor and the expected effect cannot be achieved.
[0043] 2. Deep Learning
[0044] Deep learning is a special type of machine learning. It is a model that draws on the characteristics of the human brain, which is composed of many neurons and completes thinking. It can be simply understood as a multi-layer neural network model. Each layer uses the output of the previous layer as input, and uses mathematical models to simulate neuronal activities and interactions to complete information processing.
[0045] Semantic segmentation is a very important direction in computer vision. Different from target detection and recognition, semantic segmentation realizes pixel-level classification of images, which can be used for object extraction and land use type classification.
[0046] The present invention adopts a large semantic segmentation model based on deep learning, uses Swin transformer as the backbone network, and uses UperNet as the downstream semantic segmentation algorithm to perform pixel-level segmentation of forest land and assign semantics.
[0047] Swin transformer is a new Transformer architecture that aims to address some limitations of traditional Transformer models when processing large-scale image data, such as fixed-size patch input and computational efficiency issues. It achieves efficient processing of image data with linear computational complexity by introducing hierarchical feature maps and shifted window self-attention mechanism. Swin Transformer can merge deep image patches to generate hierarchical feature maps, thereby adapting to image processing of different scales and high resolutions. In addition, the shifted window mechanism improves the performance of the model by bridging the windows of the previous layer, allowing the model to better capture local and global information. Swin Transformer demonstrates state-of-the-art performance on multiple visual tasks, including COCO object detection and ADE20K semantic segmentation. Due to its high efficiency and adaptability to different scales and resolutions, Swin Transformer has broad application prospects in the field of computer vision, and also provides new ideas and possibilities for other fields such as natural language processing. Different from ViT (Vision Transformer), whose low-resolution characteristics and complexity grow quadratically with the image size, Swin Transformer is suitable as a backbone network for high-resolution input images and uses the results of multiple downsampling as features. Therefore, it is suitable for the extraction of forest land in remote sensing images in the present invention.
[0048] (III) Hough transform algorithm, line detection
[0049] Hough Transform is a feature extraction technology in image processing. In this process, a set that conforms to the specific shape is obtained as the Hough Transform result by calculating the local maximum of the cumulative results in a parameter space.
[0050] The Hough transform was first proposed by Paul Hough in 1962. The original Hough transform was designed to detect straight lines and curves. The initial method required the knowledge of the analytical equation of the boundary line of the object, but did not require prior knowledge of the location of the region. One of the outstanding advantages of this method is the robustness of the segmentation result, that is, it is not very sensitive to incomplete data or noise. However, it is often impossible to obtain an analytical expression to describe the boundary. Later, it was promoted and used by Richard Duda & Peter Hart in 1972. The classic Hough transform is used to detect straight lines in the image. Later, the Hough transform was extended to the recognition of objects of arbitrary shapes, mostly circles and ellipses. The Hough transform uses the transformation between two coordinate spaces to map curves or lines of the same shape in one space to a point in another coordinate space to form a peak, thereby converting the problem of detecting arbitrary shapes into a statistical peak problem.
[0051] The principle of Hough transform to detect straight lines is based on the idea of point-line duality. In the image space XY, a straight line can be represented by the equation y=px+q, where p is the slope and q is the intercept. However, Hough transform transforms this problem into parameter space PQ, where p and q are variables, and x and y are fixed values. In this way, a straight line in the image space corresponds to a point in the parameter space, and a point in the image space corresponds to a straight line in the parameter space.
[0052] Specifically, for each point (x, y) in the image space, a straight line q = y-px can be found in the parameter space. Each point (p, q) on this line represents a possible straight line through the point (x, y) in the image space. Therefore, collinear points in the image space will form multiple straight lines in the parameter space that intersect at the same point, and this intersection corresponds to the straight line in the image space.
[0053] In practical applications, the lines corresponding to the collinear points can be found by accumulating or voting in the parameter space. Specifically, a two-dimensional array A(p,q) can be created. For each point (x,y) in the image space, the corresponding line equation q=y-px is calculated, and the value of the corresponding position in array A is added by 1. Finally, the peak points in array A are found. These peak points are the parameters of the lines corresponding to the collinear points in the image.
[0054] In artificial afforestation, trees have obvious row and column characteristics, so the Hough transform algorithm can be used to detect straight lines and distinguish between natural forests and artificial afforestation.
[0055] Example 1
[0056] 1. Data acquisition and preprocessing
[0057] Use drones or satellites to obtain high-resolution remote sensing images, ensuring that the image resolution is 0.5 meters or higher. Perform geometric correction on the images to remove image distortion caused by the sensor or platform posture, perform radiometric correction on the images, adjust image brightness and contrast, and ensure image consistency. Finally, perform image enhancement on the images, using contrast stretching and sharpening techniques to improve image quality and detail.
[0058] 2. Woodland Extraction
[0059] Sample collection: label representative forest samples and generate training data sets
[0060] Model training: Model training is performed based on the sample set, and accuracy evaluation is performed to obtain the forestland intelligent extraction model.
[0061] Intelligent forest land extraction: Apply the trained model to high-resolution remote sensing images to obtain the distribution map of all forest lands, including artificial tree plantings and natural forest lands.
[0062] 3. Line detection and plantation identification
[0063] Image preprocessing
[0064] The extracted forest area is grayscaled and the color image is converted into a grayscale image.
[0065] Edge Detection
[0066] The Canny edge detection algorithm is used to perform edge detection on the grayscale image and extract the significant edges in the image.
[0067] Hough transform line detection
[0068] (1) Set the accumulator threshold and straight line length threshold of the Hough transform algorithm to detect the straight lines in the image that meet the characteristics of artificial tree planting rows.
[0069] (2) Based on the linear distribution density and arrangement rules, areas with obvious row and column characteristics are identified as artificial afforestation areas.
[0070] 4. Area calculation
[0071] Area Marking
[0072] The identified artificial tree planting areas and natural forest areas are marked to generate a classification result map.
[0073] Area conversion
[0074] (1) Pixel statistics: Count the number of pixels in different types of areas in the classified image.
[0075] (2) Area calculation: Based on the resolution of remote sensing images, the number of pixels is converted into actual area to obtain the statistical results of artificial afforestation and natural forest areas.
[0076] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for extracting artificial afforestation area based on high-resolution remote sensing images, characterized in that: The method comprises: Acquire remote sensing images and preprocess the remote sensing images; The pre-processed remote sensing images are input into the forest land intelligent extraction model to obtain the distribution map of all forest lands; Perform edge detection and Hough transform line detection on the distribution map of all the forest lands to obtain artificial tree planting areas and natural forest areas; The areas of artificial afforestation areas and natural forest areas are calculated to obtain the statistical results of artificial afforestation and natural forest areas.
2. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1, characterized in that: The forestland intelligent extraction model is constructed by Swin transformer and UperNet.
3. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1, characterized in that: The construction method of the forest land intelligent extraction model is: Obtain remote sensing image samples and forest land pixel-by-pixel delineation samples; Preprocess the obtained sample data, and divide the preprocessed sample data into a training set and a data set; The Swin transforme and UperNet are iteratively trained through the training set to obtain the forest intelligent extraction model, and then the forest intelligent extraction model is verified through the verification set.
4. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1 or 3, characterized in that: The pre-processing is: First, the remote sensing image is geometrically corrected to remove image distortion caused by the sensor or satellite attitude, then radiation correction is performed to adjust the image brightness and contrast, and finally the image is enhanced using contrast stretching and sharpening techniques to improve image quality and texture features.
5. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1, characterized in that: The distribution map of all the forest lands is subjected to edge detection and Hough transform line detection to obtain the artificial tree planting area and natural forest area as follows: Grayscale processing is performed on the forest areas extracted from the distribution map of all forests, and the color images are converted into grayscale images; Use the Canny edge detection algorithm to detect the edge of the grayscale image and extract the significant edges in the image; Set the accumulator threshold and straight line length threshold of the Hough transform algorithm to detect straight lines in the image that meet the characteristics of artificial tree planting rows and columns; According to the straight line distribution density and arrangement rules, the areas with obvious row and column characteristics are identified as artificial afforestation areas, and the rest are natural forest areas.
6. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1, characterized in that: The areas of artificial tree planting and natural forest are calculated to obtain the statistical results of artificial tree planting and natural forest areas: Count the number of pixels in the artificial tree planting area and the natural forest area respectively; Based on the resolution of remote sensing images, the number of pixels is converted into actual area to obtain the statistical results of artificial afforestation and natural forest areas.
7. The method for extracting artificial afforestation area based on high-resolution remote sensing images according to claim 1, characterized in that: The remote sensing image accuracy is greater than or equal to 0.5 meters.
Citation Information
Patent Citations
High canopy density stand crown area acquiring method based on images acquired by unmanned aerial vehicle
CN107451982A
Forest land degradation area detection method based on visible light visual remote sensing
CN118097446A
Method for automatically generating crown contour label based on three-dimensional analogue simulation
CN118115864A
Natural resource change monitoring system
CN118129713A
