CN115170968A - Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium - Google Patents
Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium Download PDFInfo
- Publication number
- CN115170968A CN115170968A CN202210881475.0A CN202210881475A CN115170968A CN 115170968 A CN115170968 A CN 115170968A CN 202210881475 A CN202210881475 A CN 202210881475A CN 115170968 A CN115170968 A CN 115170968A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- training
- complex scene
- data
- scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a remote sensing extraction method, a device, computer equipment and a storage medium for a complex scene plot, wherein the method comprises the steps of acquiring high-resolution remote sensing data, processing the data, performing parallel block division on a remote sensing image according to target element features to be extracted in the plot, extracting and drawing samples, and generating corresponding label images; training the drawn sample and the label image, and forming a plurality of training models of different ground feature types through deep learning; and forming a complete target element vector result of the plot spots by adopting a parallel computing mode according to the plurality of training models. According to the method, the complex scene task areas are divided into the simple scene blocks, so that mutual interference among different types of blocks is avoided, the difficulty in sample preparation and model extraction training is reduced, and the extraction missing rate and the extraction error rate are effectively reduced.
Description
Technical Field
The invention relates to the field of remote sensing data processing, in particular to a remote sensing extraction method and device for a complex scene plot.
Background
In recent years, a deep learning technology is rapidly developed in the field of extracting high-resolution remote sensing image information, an intelligent extraction technology for agricultural plots is gradually developed, a research area is more than that of a plain area, the area of the plain area in China only occupies 12%, and the research on an effective intelligent extraction method under a complex scene is less at present. In the existing agricultural land parcel intelligent extraction method based on the deep learning technology, classification and extraction of all agricultural land parcels in a high-resolution remote sensing image are always attempted to be realized by using a deep learning model, and the effect of artificial visual interpretation is not achieved. The method is characterized in that the method comprises the following steps of dividing a plurality of agricultural land types into a plurality of regions, dividing the regions into a plurality of regions, and classifying the regions into a plurality of regions.
Therefore, a high-resolution remote sensing extraction technology for a complex scene such as an agricultural plot in a mountain area is needed to solve the above technical problems.
Disclosure of Invention
The invention aims to provide a method, a device, computer equipment and a storage medium for extracting a remote sensing plot of a complex scene, wherein a task area is divided into a plurality of parallel computing modules according to the characteristics of the complex scene, and different models are adopted in each computing module for extracting different elements to be extracted.
In order to achieve the purpose, the invention provides a remote sensing extraction method of a complex scene plot, which comprises the following steps:
acquiring high-resolution remote sensing data and processing the data;
performing parallel block division on the remote sensing image according to the target element characteristics to be extracted in the block, and dividing a complex scene into a plurality of relatively independent simple scenes;
extracting and drawing a sample, and generating a corresponding label image;
training the drawn sample and the label image, and forming a plurality of training models of different ground feature types through deep learning; and
and forming a complete target element vector result of the plot pattern by adopting a parallel computing mode according to the plurality of training models.
The invention also provides a remote sensing extraction device for the complex scene plots, which comprises the following components:
the remote sensing data acquisition and processing module is used for acquiring high-resolution remote sensing data and processing the data;
the scene simplifying module is used for carrying out parallel block division on the remote sensing image according to the acquired remote sensing data and the target element characteristics to be extracted in the block, and dividing the complex scene into a plurality of relatively independent simple scenes;
the tag image generation module is used for extracting and drawing samples, forming sample sets aiming at different ground feature types and generating corresponding tag images;
the training model generation module is used for training the drawn sample and the label image and forming a plurality of training models of different ground feature types through deep learning; and
and the vector result generation module is used for forming a complete target element vector result of the plot spot by adopting a parallel computing mode according to the plurality of training models.
The present invention further provides a computer device, including:
a processor, a memory for storing processor-executable instructions, and a computer program stored on the memory and executable on the processor, the processor implementing the above method when executing the program.
The invention also proposes a computer-readable storage medium, on which a computer program is stored, characterized in that the computer program implements the above method when executed by a processor.
The method adopts a parallel computing mode, divides the complex scene into a plurality of simple scene blocks according to the feature of the target element to be extracted, ensures that the feature of the agricultural land element in each block is relatively consistent, changes the complex scene into a plurality of relatively independent single simple scenes, and improves the precision and efficiency of the extraction required by the agricultural land object.
Meanwhile, the invention adopts a distributed terminal editing mode, and distributes the tasks needing manual editing to a plurality of distributed terminals through the Internet for parallelization operation, thereby improving the working efficiency and shortening the project operation period of the task area.
The method divides the agricultural land parcel into five levels to extract aiming at the difference of the edge and the textural feature of the target element to be extracted, fully considers the characteristics of each type of agricultural land parcel under a complex scene, and avoids the mutual interference among land parcels of different types.
The beneficial effects of the invention are: according to the remote sensing extraction method of the complex scene plot, disclosed by the invention, through the idea of simplifying the complex scene plot into the simple scene plot, the complex scene task area is divided into the simple scene plots with relatively consistent internal conditions, the influence of terrain background noise in the complex scene is effectively inhibited, and the integrity of the boundary of the spot plot is ensured to the greatest extent.
Meanwhile, the method for extracting the samples by the step-by-step parallel computing fully considers the characteristics of various types of the plots in the complex scene, avoids mutual interference among the plots in different types, greatly reduces the difficulty of sample making and model extracting training, and effectively reduces the extraction missing rate and the extraction error rate.
Through practice verification, compared with the traditional human-computer interaction visual interpretation extraction method, the method adopted by the application takes a block of 1175km in area 2 For example, the extraction efficiency of the human visual interpretation is about 3km 2 The total time for completing the extraction is 408 days; by adopting the computer, the intelligent automatic extraction only needs 2 hours, and the operation efficiency is greatly improved. And meanwhile, distributed personnel editing is adopted for sample drawing and manual editing, so that the project period of a task area is greatly shortened.
Drawings
FIG. 1 is a flow chart diagram of the remote sensing extraction method of a complex scene plot of the present invention;
FIG. 2 is a schematic flow chart of the remote sensing extraction method of complex scene plots of the present invention;
FIG. 3 is a schematic diagram of a parallelized block partitioning process of the present invention;
FIG. 4 is a schematic diagram of a mountain region parallelization block partitioning according to the present invention;
FIG. 5 is a sample schematic view of a tillable area of the present invention;
FIG. 6a is a schematic view of the superposition of the regular farmland sample vector and the image according to the present invention;
FIG. 6b is a schematic view of the label image of FIG. 6 a;
FIG. 7a is a schematic view of the superimposition of the fuzzy farmland sample vector and the image according to the present invention;
FIG. 7b is a schematic view of the label image of FIG. 7 a;
FIG. 8a is a schematic view of an edge extraction image of a regular farmland according to the present invention;
FIG. 8b is a schematic illustration of the predicted gray scale of FIG. 8 a;
FIG. 8c is a schematic view of the line vectors of FIG. 8 a;
FIG. 8d is a schematic view of the face vector of FIG. 8 a;
FIG. 8e is a schematic diagram of the extraction results of FIG. 8 a;
fig. 9 is a block diagram of the remote sensing extraction device for a complex scene plot of the present invention.
Detailed Description
The technical solution of the embodiment of the present invention will be clearly and completely described below with reference to the accompanying drawings of the present invention.
The embodiment of the invention discloses a remote sensing extraction method of a complex scene plot, which divides a complex scene in a test area into a plurality of relatively independent single simple scenes according to information such as terrain, topography, landform and the like, divides the complex scene into different types of plots according to visual characteristics presented by agricultural plots on the basis of each type of simple scenes, and designs different deep learning models for the different types of plots to extract layer by layer. Specifically, referring to fig. 1 and fig. 2, the method for extracting a parcel includes:
s1, acquiring high-resolution remote sensing data and processing the data
The method comprises the following steps of preparing data, mainly obtaining high-resolution remote sensing data and processing the data, wherein the high-resolution remote sensing data comprises high-resolution remote sensing image data, DEM data and navigation data. High-resolution remote sensing is preferably remote sensing images with the resolution better than 1 meter, such as high-resolution second-order, high-resolution seventh-order and Beijing second-order, and topographic data is a grid spacing digital elevation model DEM which covers a task area and meets the requirements of orthorectification precision, such as 30 meters and 12 meters. The data processing comprises processing the remote sensing image into an orthoscopic remote sensing image, and forming ridges, valley lines and the like through image analysis.
S2, according to the obtained and processed remote sensing data, dividing parallelization blocks on the remote sensing image according to the features of the target elements to be extracted in the blocks
Specifically, the complex scene is divided into a plurality of simple scenes on the basis of a road network, a water system network and terrain analysis and extraction of terrain lines such as ridge lines/valley lines and the like through DEM data, and the simple scenes are divided into a plurality of simple scenes on the basis of not damaging map spots of the terrain blocks, wherein the simple scenes are relatively independent, and the features of target elements to be extracted under a single scene are as consistent as possible. And a control unit for queuing and sequencing is provided for implementing parallelization calculation of large-scale production tasks while providing boundary constraint for next layer-by-layer extraction. In the embodiment, the agricultural land parcel under the complex scene is divided into 5 target elements to be extracted, namely a regular farmland, a terrace, a fuzzy farmland, a water area and a garden land, the characteristics of the land parcel in the same area are consistent, and the deep learning calculation can be carried out by adopting the same model.
S3, extracting and drawing the sample and generating a corresponding label image
The method comprises the steps of selecting samples according to target element features to be extracted in each parallelization block, taking a single block or a plurality of blocks with consistent target element features to be extracted as the same sample set, finally forming a plurality of sample sets by the single target element to be extracted, and finally forming a plurality of sample sets covering five types of elements and different types, such as a sample set 1, a sample set 2 \8230anda sample set n. And manually drawing the selected sample set to form a sample vector file with the vector and the image superposed.
And generating a corresponding label image according to a sample vector file formed by superposing the drawn sample vector and the image, wherein the label image correspondingly adopts different forms according to different extraction modes of 5 different elements.
S4, training the drawn sample and the label image, and forming a plurality of training models of different ground feature types through deep learning
And training the sample image and the label image, wherein the training is to finally form a surface/line model aiming at a certain ground object type by utilizing a deep learning technology according to the drawn sample. The model formed as in the present embodiment is a regular arable land line model, terrace line model, water surface model, garden surface model, and fuzzy arable land surface model. The boundary of the regular farmland is clear, and boundary information is extracted by adopting a deep learning edge model (line model); the texture in the fuzzy cultivated land, the garden land and the water area is uniform, and the texture information is extracted by adopting a texture model (surface model) of deep learning.
And S5, forming a complete target element vector result of the plot spots by adopting a parallel computing mode according to the training models.
Specifically, by utilizing various trained target element models and adopting a parallel computing mode, regular farmland, terraced fields, water, garden land and fuzzy farmland are respectively extracted according to the visual attention sequence under a single simple scene module. The method comprises the following steps:
and S51, predicting the image by using the model formed by training to generate predicted gray grid data.
And S52, grid vectorization, wherein the predicted gray grid data are converted into vector data.
And S53, performing a series of post-processing such as boundary smoothing and beautifying on the generated vector data.
And finally, summarizing the complex scene task area spot elements by a plurality of simple scene modules.
S6, distributing the vector result to a plurality of distributed editing terminals for manual editing in a distributed terminal mode
The complex scene task area is huge, a large amount of manual achievements are needed for sample drawing and intelligent extraction achievement editing, a plurality of simple scene blocks are respectively distributed to a plurality of distributed terminals for synchronous manual editing in a distributed terminal mode through the Internet, the interpretation efficiency of the target elements of the large complex scene is improved, and the engineering period is shortened.
The following specific examples take Chongqing city (diverse topography and landform, complex meteorological conditions) as a research area, and the specific implementation scheme is as follows:
first phase, data preparation
1) The sub-meter high-resolution remote sensing images such as high-resolution second and high-resolution seventh images are used as data sources and processed to form an orthographic remote sensing image which is better than 1 meter.
2) And (3) adopting a 30-meter equal grid distance DEM which covers a task area and meets the requirement of orthorectification precision for topographic data, and analyzing to form ridge lines and valley lines.
3) Other data
Navigation data, road and river basic data and the like in Chongqing city areas.
Second stage, parallelize the computation
1) Based on data such as a road network, a water system network, a valley/ridge line and the like, the division of the parallelization blocks is carried out on the remote sensing image according to the following steps: the characteristics of target elements to be extracted in the same block are ensured to be consistent, as shown in fig. 3 and 4, in the parallel computing block example in the plain area, according to different texture characteristics of cultivated land presented by different types of crops, a task area is divided into 5 blocks capable of being computed in parallel, characteristics of cultivated land in the same block are consistent, and deep learning computation can be performed by adopting the same model.
2) And (3) cutting the parallelization block image, wherein the task area image is cut based on the drawn parallelization calculation block vector so as to adapt to deep learning of a machine, and the Chongqing city is divided into 458 parallelization calculation modules.
3) Selecting a target element sample to be extracted: and selecting samples according to the features of the target elements to be extracted in each parallelization block, wherein a single block or a plurality of blocks with consistent features of the target elements to be extracted serve as the same sample set, and finally, the single target element to be extracted can form a plurality of sample sets. Fig. 5 is a sample schematic diagram of farmland.
4) Preparing a target element sample to be extracted: the sample form uses vector samples for rendering edges and textures. The method and the device have the advantages that the range of the task area is large, and the sample drawing is completed in a distributed editing mode.
5) Sample label image preparation: and identifying the field of the ground object type of the target in the sample vector file to generate a single-waveband gray image with the gray value of the edge pixel of the target ground object of 255 and the gray value of the rest pixels of 0. For the difference of the 5 level element extraction modes, the label image correspondingly adopts different forms, and the label making schematic diagrams of the regular farmland and the fuzzy farmland are combined with the diagrams of fig. 6 and 7, and are respectively as follows:
for regular farmland, the regular farmland is mainly positioned in regions with gentle topography, has clear boundaries, uniform internal textures, relatively regular spatial distribution and regular forms, and can better extract boundary information by using an edge model for deep learning to construct land blocks. Therefore, an edge label image is created for the regular farmland sample, and the label creation result is as shown in fig. 6 as an example of the regular farmland label creation.
For terraced fields, because the boundaries of terraced field plots are clear, the textures are uniform, and compared with regular farmland plots, the terraced field plots are fine, long and narrow in shape, a more focused deep learning edge model needs to be designed, extraction of finer boundaries is achieved, and the tag manufacturing process is consistent with regular farmland.
For fuzzy farmland, the edge characteristics of the fuzzy farmland are very fuzzy or basically nonexistent, and the shapes of the fuzzy farmland are different, and the fuzzy farmland is mainly distinguished by the texture structure of bare soil or crop canopies, so that the identification needs to be carried out through a deep learning texture segmentation model. Therefore, a texture label image is created for the blurred farmland sample, and the label creation result is shown in fig. 7 as an example of blurred farmland label creation.
For the water area, the water area plots have the characteristics of disordered and irregular edges and uniform internal textures due to the influence of water level, soil quality and the like, so that the extraction of the water area surface is realized by using a deep learning texture model, and the label manufacturing process is consistent with fuzzy arable land.
For the garden, perennial woody and herbaceous plants mainly comprising fruits and leaves are planted in the garden plots, the distribution in the garden plots is relatively regular, the textures are relatively uniform, the extraction of the garden plots is realized by using a texture model for deep learning, and the label making process is consistent with that of fuzzy arable land.
Third-stage sample model training and target element vector achievement generation
The training is to finally form a surface/line model aiming at a certain ground object type by utilizing a deep learning technology according to the drawn samples. In the embodiment, agricultural land parcel models in Chongqing are divided into the following 5 types: a regular arable land line model, a terrace line model, a water surface model, a garden surface model, and a fuzzy arable land surface model.
Based on a plurality of training models of 5 types of elements, 458 parallel computing blocks of the task area are sequentially and respectively subjected to parallel computing and extraction according to 5 types of regular arable land, terrace, water area, garden land and fuzzy arable land, and a relatively independent target element vector result with complete plot and pattern spots is formed by combining a schematic diagram of an edge extraction process and a result of the regular arable land shown in fig. 8.
Fourth phase distributed terminal processing
The application relates to a manual editing part, which is carried out in a mode of parallel development of distributed terminals, wherein the distributed editing relates to two stages:
1) Sample drawing: and respectively distributing the five types of target element samples to be extracted to a plurality of distributed terminals for manual drawing.
2) Manual editing of parallel computing results: the five types of achievements extracted by parallel computing still have certain errors in form and topological relation, and the method adopts a distributed manual editing mode to distribute the achievements to a plurality of distributed editing terminals for manual editing.
Various target extraction results are formed through the process flow, and finally, the task area results with full coverage and fine morphology are formed through layer-by-layer combination to obtain the combined results.
With reference to fig. 9, the present invention further provides a remote sensing extraction device for complex scene plots, including:
the remote sensing data acquisition and processing module is used for acquiring high-resolution remote sensing data and processing the data;
the scene simplification module is used for carrying out parallel block division on the remote sensing image according to the acquired remote sensing data and target element features to be extracted in the land parcel, and dividing the complex scene into a plurality of relatively independent simple scenes;
the tag image generation module is used for extracting and drawing samples, forming sample sets aiming at different ground feature types and generating corresponding tag images;
the training model generation module is used for training the drawn sample and the label image and forming a plurality of training models of different ground feature types through deep learning;
and the vector result generation module is used for forming a complete target element vector result of the plot spot by adopting a parallel computing mode according to the plurality of training models. And
and the distributed terminal editing module is used for distributing the vector result to a plurality of distributed editing terminals for manual editing and interpretation in a distributed terminal mode, and combining layer by layer to obtain a combined result.
The detailed functions of the modules are described above, and are not described in detail here.
The invention also discloses a computer device which can realize the remote sensing extraction method of the complex scene land parcel, and the electronic device comprises at least one memory, at least one processor and a computer program, wherein the at least one memory is coupled to the at least one processor, and the computer program is stored in the memory and can run in the processor. The computer program herein may be divided into one or more units, which are stored in and executed by the memory, to accomplish the present invention. One or more of the units may be a series of computer program instruction segments for describing the execution of the computer program in the electronic device, which can implement specific functions.
The invention also discloses a computer readable storage medium, wherein a computer program is stored in the computer readable storage medium, and when the computer program is executed by a processor, the remote sensing extraction method of the complex scene plot can be realized. Wherein the computer program includes computer program code, which may be in source code form, executable file or some intermediate form, etc., and the computer readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a usb disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a Random Access Memory (RAM), etc.
Therefore, the scope of the present invention should not be limited to the disclosure of the embodiments, but includes various alternatives and modifications without departing from the scope of the present invention, which is defined by the claims of the present patent application.
Claims (10)
1. A remote sensing extraction method of a complex scene plot is characterized by comprising the following steps:
acquiring high-resolution remote sensing data and processing the data;
performing parallel block division on the remote sensing image according to the target element features to be extracted in the block;
extracting and drawing a sample, and generating a corresponding label image;
training the drawn sample and the label image, and forming a plurality of training models of different ground feature types through deep learning; and
and forming a complete target element vector result of the plot pattern by adopting a parallel computing mode according to the plurality of training models.
2. The remote sensing extraction method of complex scene plots according to claim 1, characterized by further comprising: and distributing the vector result to a plurality of distributed editing terminals for manual editing and interpretation in a distributed terminal mode, and combining layer by layer to obtain a combined result.
3. The remote sensing extraction method of the complex scene land parcel according to claim 1, characterized in that, the parallel block division comprises dividing the complex scene into a plurality of simple scenes, each simple scene is relatively independent, and the features of the target elements to be extracted of the single scene are consistent.
4. The remote sensing extraction method of complex scene plots according to claim 1, wherein a plurality of blocks with consistent target element features are used as the same sample set, and manual drawing is respectively performed in the selected sample set to form a sample vector file with a vector and an image superposed.
5. The remote sensing extraction method of complex scene plots according to claim 1, wherein the parallel computing mode comprises:
predicting the image by using a model formed by training to generate predicted gray grid data;
converting the predicted gray grid data into vector data;
and performing post-processing of boundary smoothing and beautification on the generated vector data.
6. The remote sensing extraction method of complex scene plots according to claim 1, wherein for agricultural plots in the complex scene, the types of plots comprise regular farmland, terrace, water, garden land and fuzzy farmland, and the training model comprises an edge model and a texture model.
7. A remote sensing extraction element of complicated scene plot characterized in that includes:
the remote sensing data acquisition and processing module is used for acquiring high-resolution remote sensing data and processing the data;
the scene simplifying module is used for carrying out parallel block division on the remote sensing image according to the acquired remote sensing data and the target element characteristics to be extracted in the block;
the label image generation module is used for extracting and drawing the sample and generating a corresponding label image;
the training model generation module is used for training the drawn sample and the label image and forming a plurality of training models of different ground feature types through deep learning; and
and the vector result generation module is used for forming a complete target element vector result of the plot spot by adopting a parallel computing mode according to the plurality of training models.
8. The utility model provides a remote sensing extraction element of complicated scene plot which characterized in that still includes:
and the vector result editing module is used for distributing the vector result to a plurality of distributed editing terminals for manual editing and interpretation in a distributed terminal mode, and combining layer by layer to obtain a combined result.
9. A computer device, comprising:
a processor, a memory for storing processor-executable instructions, and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of claims 1 to 6 when executing the program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210881475.0A CN115170968A (en) | 2022-07-26 | 2022-07-26 | Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210881475.0A CN115170968A (en) | 2022-07-26 | 2022-07-26 | Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115170968A true CN115170968A (en) | 2022-10-11 |
Family
ID=83496674
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210881475.0A Pending CN115170968A (en) | 2022-07-26 | 2022-07-26 | Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115170968A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117173353A (en) * | 2023-09-04 | 2023-12-05 | 广东省核工业地质局测绘院 | Geological mapping system based on remote sensing image |
-
2022
- 2022-07-26 CN CN202210881475.0A patent/CN115170968A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117173353A (en) * | 2023-09-04 | 2023-12-05 | 广东省核工业地质局测绘院 | Geological mapping system based on remote sensing image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2020103026A4 (en) | A Single Tree Crown Segmentation Algorithm Based on Super-pixels and Topological Features in Aerial Images | |
| CN111598045B (en) | Remote sensing farmland change detection method based on object spectrum and mixed spectrum | |
| Lu et al. | How to automate timely large-scale mangrove mapping with remote sensing | |
| CN101488226A (en) | Tree measurement and reconstruction method based on single three-dimensional laser scanning | |
| CN113221806B (en) | Cloud platform fusion multi-source satellite image and tea tree phenological period based automatic tea garden identification method | |
| CN111489092B (en) | Method and system for evaluating suitable growing area of plant cultivation and planting environment | |
| CN103745497A (en) | Plant growth modeling method and system | |
| CN111091079A (en) | TLS-based method for measuring dominant single plant structural parameters of vegetation in alpine and fragile regions | |
| Risse et al. | Software to convert terrestrial LiDAR scans of natural environments into photorealistic meshes | |
| CN103824324A (en) | Fruit tree crown layer leaf and fruit three-dimensional reconstruction method and system | |
| CN115170968A (en) | Remote sensing extraction method and device for complex scene plot, computer equipment and storage medium | |
| CN115761486A (en) | Rice planting area judgment method and system based on multi-stage rice field image characteristics | |
| Zhu et al. | A calculation method of phenotypic traits based on three-dimensional reconstruction of tomato canopy | |
| CN115468917A (en) | Method and system for extracting crop information of farmland plot based on high-resolution remote sensing | |
| CN112509142B (en) | Bean strain rapid three-dimensional reconstruction method based on phenotype-oriented precise identification | |
| CN117409322A (en) | Regional forest natural resource ecological estimation method and system based on air-ground remote sensing combination | |
| CN115968760B (en) | On-site irrigation method based on potato root space-time distribution | |
| CN106980701A (en) | Statistical system and statistical method for geography influence on residential building form | |
| CN114842356B (en) | High-resolution earth surface type sample automatic generation method, system and equipment | |
| CN112132066B (en) | Weak transition zone extraction method based on fuzzy logic | |
| CN113421255B (en) | Grid-based farmland cropping index extraction method and system | |
| CN113052433B (en) | Crop unit yield estimation method based on key time phase and farmland landscape characteristic parameters | |
| Xing et al. | The Application of Augmented Reality Technology in Urban Greening Plant Growth State Detection | |
| CN118031912B (en) | Mapping method for natural resource research | |
| Zhang et al. | A stratified temporal spectral mixture analysis model for mapping cropland distribution through MODIS time-series data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |