CN112767244B - High-resolution seamless sensing method and system for earth surface elements - Google Patents
High-resolution seamless sensing method and system for earth surface elements Download PDFInfo
- Publication number
- CN112767244B CN112767244B CN202011634478.1A CN202011634478A CN112767244B CN 112767244 B CN112767244 B CN 112767244B CN 202011634478 A CN202011634478 A CN 202011634478A CN 112767244 B CN112767244 B CN 112767244B
- Authority
- CN
- China
- Prior art keywords
- data
- cloud
- sensing
- perception
- seamless
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000008447 perception Effects 0.000 claims abstract description 66
- 238000000605 extraction Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000012795 verification Methods 0.000 claims abstract description 10
- 230000004927 fusion Effects 0.000 claims description 9
- 238000012937 correction Methods 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 8
- 230000005855 radiation Effects 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 7
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 238000003786 synthesis reaction Methods 0.000 claims description 4
- 230000002194 synthesizing effect Effects 0.000 claims description 4
- 230000000295 complement effect Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 2
- 230000002123 temporal effect Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 7
- 239000002689 soil Substances 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 229910001385 heavy metal Inorganic materials 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000013618 particulate matter Substances 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 238000004579 scanning voltage microscopy Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a high-resolution seamless sensing method and system for earth surface elements, which comprises the following steps: decomposing the earth surface element sensing task, and determining a specific sensing task according to earth surface element task requirements given by a user; then, data related to surface elements are obtained through sensing nodes combined by virtuality and reality, and sensing data resources meeting task indexes are screened and accessed; detecting cloud, shadow and space observation blind areas by the accessed image data; generating high-resolution seamless perception data; according to the obtained high-resolution seamless sensing data, a training sample set obtained by sensing nodes combined with virtuality and reality is utilized to construct a ground surface element information and knowledge extraction model, and a high-resolution ground surface element seamless sensing result is generated; and finally, calculating a precision index by utilizing a verification sample set obtained by the sensing nodes combined with the virtual and the real parts according to the obtained high-resolution surface element seamless sensing result.
Description
Technical Field
The invention relates to a seamless sensing method for surface elements, in particular to a seamless sensing method and system for high-resolution surface elements, and belongs to the field of geospatial information and knowledge extraction.
Background
The sensing of the surface elements refers to acquiring relevant data of substances or energy in the earth surface layer through technologies such as earth observation, remote sensing, internet of things, artificial intelligence and the like, and extracting information and knowledge of the surface elements. The perception of high-resolution surface elements is the key basic work of high-precision monitoring and fine modeling of the surface environment, and is beneficial to decision making to relieve environmental ecological disasters and problems. From the angle of the numerical expression type of the surface elements, the surface elements can be divided into two categories of qualitative type and quantitative type, the qualitative type surface elements comprise water impervious surfaces, water bodies, land utilization types, soil types and other classified type elements, and the quantitative type surface elements comprise parameters such as soil temperature and humidity, heavy metal content, air particulate matter content and the like.
The surface element sensing includes two stages, namely "sensing" of data acquisition and "learning" of information extraction. Due to the wide satellite observation coverage range, the high aerial observation resolution and the high ground station observation precision, the earth observation and remote sensing technology combined with the space-sky-earth sensing resources is an effective means for acquiring earth surface element data. However, in the existing high-resolution observation, due to the problems of cloud coverage, a spatial observation blind area, time discontinuity, insufficient spatial resolution and the like, the high-resolution perception data is insufficient, and the high-resolution earth surface element information is incomplete, discontinuous and has gaps. In addition, in the current technical solutions and systems, attention is usually focused on one stage of "feeling" and "knowing", for example, attention is focused on a ground observation data system of "feeling", attention is focused on a remote sensing or geographic information processing system of "knowing", a user needs to obtain a surface element product, and usually, a part of the "feeling" and "knowing" process can be realized by means of existing software or systems, but needs to realize another stage by himself, and there is no technology and system for realizing "feeling" and "knowing" integrally in a manner of facing to a whole flow of a perception task, generating surface element information knowledge from user requirements, and so on.
Disclosure of Invention
Aiming at the problems of perception gaps and separation of 'perception' and 'awareness' in the perception of the high-resolution surface elements, the invention provides a high-resolution surface element seamless perception method and system.
The invention discloses a high-resolution seamless perception method for surface elements, which is characterized by comprising the following steps:
step 1, decomposing a ground surface element sensing task, and determining detailed task indexes of the sensing task, such as time range, space coverage, ground surface element type, spatial resolution and the like according to the ground surface element task requirement given by a user;
and 2, accessing the sensing data resource. And (3) acquiring data related to the earth surface elements through sensing nodes combined by virtuality and reality, and screening and accessing sensing data resources meeting the task indexes in the step (1).
And 3, detecting the quality of the perception data, namely detecting cloud, shadow and space observation blind areas of the image data accessed in the step 2.
And 4, improving the quality of the perception data and generating high-resolution seamless perception data.
And 5, extracting the earth surface element information and knowledge, constructing an extraction model of the earth surface element information and knowledge by utilizing the training sample set obtained by the sensing nodes combined with the reality and the virtuality in the step 2 according to the high-resolution seamless sensing data obtained in the step 4, and generating a high-resolution earth surface element seamless sensing result.
And 6, verifying the precision of the seamless sensing result of the high-resolution surface element. And (4) according to the high-resolution seamless sensing result of the surface element obtained in the step (5), calculating a corresponding precision index according to the surface element belonging to a fixed type or a quantitative type by utilizing a verification sample set obtained by the sensing node combined with the virtual and real in the step (2).
Based on the same idea, the invention also designs a system for realizing the high-resolution seamless perception method of the surface elements, which comprises the following steps: the data access module is used for acquiring data related to the surface elements through sensing nodes combined by virtuality and reality, and screening and accessing sensing data resources meeting the task indexes in the step 1;
the virtual nodes and the real nodes respectively refer to sensing nodes in virtual spaces such as social media data, volunteer geographic information data, historical archived information, public surface element products, data acquired by users and the like; sensing nodes existing in a physical space represented by observation platforms and sensors of a space base, a space base and a foundation;
the perception data quality detection module is used for detecting cloud, shadow and space observation blind areas of the accessed image data;
the perception data quality improving module generates high-resolution seamless perception data;
the earth surface element information and knowledge extraction module is used for constructing an earth surface element information and knowledge extraction model according to the high-resolution seamless perception data and a training sample set obtained by using a perception node combined with a virtual node and a real node and generating a high-resolution earth surface element seamless perception result;
and the precision verification module is used for verifying the seamless perception result of the high-resolution earth surface element. And calculating the precision of the seamless sensing result of the high-resolution surface element by utilizing a verification sample set obtained by the sensing nodes combined by the virtual and the real.
The invention has the advantages that:
1. the technology and the system can realize perception integrated service, have low use threshold for users, only need the users to put forward a specific earth surface element perception task requirement, can automatically carry out a perception process of acquiring perception data resources and a perception process of extracting earth surface element information, and output high-resolution earth surface element perception results and precision thereof through all the technical steps and corresponding modules of the invention without additional operation of the users.
2. In the process of 'sensing' of data acquisition, the invention proposes to acquire data related to surface elements through sensing nodes combined by virtual and real, and the conventional data acquisition only focuses on sensing data resources existing in a physical space represented by observation platforms and sensors of a space base, a space base and a foundation, but can also meet the shortage of available resources (such as observation blind areas) or need additional auxiliary data (such as training samples and test samples) to assist in completing the subsequent 'sensing' process. According to the invention, the sensing nodes combined by the virtual and the real are used for acquiring data related to the surface elements, so that not only are the sensing resources of a physical space concerned, but also the sensing data resources in a virtual space such as social media data, volunteer geographic information data, historical archive information, public surface element products, data acquired by a user and the like are concerned, the available data resources in the 'sensing' process can be greatly enriched, a complete data base is provided for the subsequent steps 3-6, and the data resources required by the steps 3-6 are all from the sensing nodes combined by the virtual and the real.
3. The invention can realize seamless perception of the high-resolution surface elements, and aims at solving the problems that the high-resolution perception data is insufficient and the high-resolution surface element information is incomplete, discontinuous and gapped due to factors such as cloud coverage, a space observation blind area, time discontinuity and insufficient space resolution in the existing high-resolution observation.
4. In the process of improving the quality of the perception data and generating high-resolution seamless perception data in the step 4, processing is respectively carried out according to the light and heavy degrees of the cloud and shadow degrees of the image, and for the areas with light cloud and shadow degrees and small occupation ratio, the cloud and shadow are directly recovered based on the corresponding image; and regarding the areas with heavy cloud and shadow degrees and large occupation ratio, regarding the areas as space observation blind areas. And firstly carrying out cloud-free data synthesis and then rebuilding an observation blind area.
5. According to the numerical expression type of the earth surface elements, different earth surface element information extraction methods and different accuracy verification modes are respectively adopted for the earth surface elements according to qualitative and quantitative two categories in the steps 5 and 6.
Drawings
Fig. 1 is a general technical flowchart of a high-resolution surface element seamless perception method and system in the invention.
Detailed Description
In specific implementation, the technical method and process provided by the invention can be realized by a computer programming technology, and in order to make the technology of the invention easier to understand and master, the invention is further described by the following specific implementation in combination with the accompanying drawings and examples:
the invention discloses a seamless perception method for high-resolution surface elements, which adopts the following technical scheme:
step 1, decomposing a ground surface element sensing task, and determining detailed task indexes of the sensing task, such as time range, space coverage, ground surface element type, spatial resolution and the like according to the ground surface element task requirement given by a user;
and extracting the detailed perception task index requirements through technologies such as word segmentation extraction, semantic reasoning, geocoding and the like. For example, the sensing time range is 1 month and 1 day in 2019 to 12 months and 31 days in 2019, the spatial coverage range is a vector boundary range corresponding to ten thousand square kilometers in the Wuhan city circle, the surface element type is the impervious surface and belongs to qualitative surface elements, and the spatial resolution is better than or equal to 2 meters.
And 2, accessing the sensing data resource. And (3) acquiring data related to the earth surface elements through sensing nodes combined by virtuality and reality, and screening and accessing sensing data resources meeting the task indexes in the step (1).
And (2) accessing data resources meeting the task indexes in the step (1) through a sensing node combined by virtual and real, specifically, accessing data information comprising metadata and a data body, wherein the metadata is descriptive data of the data body, firstly accessing the metadata, screening out a unique identifier corresponding to the data body meeting the requirements according to the sensing task indexes in the step (1), and then screening and accessing the corresponding data body.
The sample data obtained from the sensing nodes combined by the virtual and the real are divided into two parts, wherein one part is used as a training sample set, and the other part is used as a verification sample set.
The access mode of the sensing data can comprise two modes, one mode is automatic access through an open interface of the sensing node, and the other mode is manual access of the sensing data through manual copying or manual input.
The sensing nodes combined by the virtual sensing and the real sensing are sensing antennae of surface element data and information, and comprise physical sensing nodes and virtual sensing nodes, specifically, the physical sensing nodes comprise sensing nodes which exist in a physical space represented by observation platforms and sensors of a space base, a space base and a foundation, and the virtual sensing nodes comprise sensing nodes in a virtual space represented by social media data, volunteer geographic information data, historical archive information, public surface element products, data acquired by a user and the like. The physical sensing node is a key basis for realizing the current earth surface element sensing task and acquires point source or surface source sensing data with strong timeliness, high resolution and high coverage rate. The virtual sensing nodes are enhancement means for assisting in realizing the sensing of the earth surface elements, and are used for acquiring auxiliary information related to an earth surface element sensing task from multiple sources and multiple angles so as to enhance the capability of the earth surface element sensing task in constructing a model and extracting earth surface element information in a 'learning' stage. For example, the data of interest points extracted from social media data, the surface classification sample data extracted from volunteer geographic information data, the surface classification sample labeled by a user, and the like can be used as a sample truth value to assist the qualitative surface element perception task to construct a model in the 'learning' stage.
And 3, detecting the quality of the perception data, namely detecting cloud, shadow and space observation blind areas of the image data accessed in the step 2.
And 3.1, detecting the clouds and the shadows in the image data bodies accessed in the step 2 one by one, and generating cloud shadow masks corresponding to the images one by one, wherein the cloud shadow masks represent the positions of the clouds and the shadows in the corresponding images, and the total amount of the clouds and the shadows can be calculated on the basis. Cloud and shadow detection can be performed by methods such as reflectivity threshold, spectral index and machine learning. Specifically, cloud and shadow information of the image can be extracted by setting a threshold value based on the spectral band value and the calculated cloud/shadow index, and a cloud and shadow mask is generated; or selecting clouds (such as thick clouds and thin clouds), shades (such as thick shades and thin shades) with different contents and clear cloud-free and shadow-free objects (such as buildings, roads, vegetation, farmlands, bare soil, water bodies and the like) from a sample set obtained from the sensing nodes combined in the step 2 to construct training samples, then training a classifier by using a machine learning method (such as SVM, random forest and neural network), and extracting cloud and shadow information (including the cloud and shadow conditions with different degrees) in each image by using the trained classifier to obtain the cloud and shadow masks with different degrees. And counting the number of the pixels covered by the cloud and the shadow and the proportion of the pixels in the image on the basis of the cloud and shadow checking results.
And 3.2, detecting that the image data accessed in the step 2 cannot cover a space observation blind area caused by the perception space-time range in the step 1. Firstly, extracting an effective coverage boundary bi of each available image data volume, merging the coverage boundaries of all the single images to obtain an image data global coverage boundary B accessed in the step 2, and erasing the image global coverage boundary B accessed in the step 2 on the basis of the perception task vector range A in the step 1 to obtain a spatial observation blind area M, as shown in (1).
M＝A-A∩B (1)
Wherein M is the image space observation blind area, a is the sensing task space range of step 1, and B is the image data global coverage boundary accessed in step 2.
B＝b1∪b2…∪bn (2)
Wherein b is1,b2…bnAnd n is the number of the image data accessed in the step 2 for the coverage boundary of each available image data volume.
And 4, improving the quality of the perception data and generating high-resolution cloud-free seamless perception data.
The steps can be processed according to the cloud degree and the shadow degree of the image respectively. How to judge the degree of cloud and shadow needs to be according to the cloud and shadow detection method and result selected in step 3.1, which specifically includes two ways:
one is that if the method adopted in step 3.1 makes the cloud or shadow of different degrees as a condition, and does not distinguish thick cloud, thin cloud, thick shadow, thin shadow, then the result of step 3.1 is only cloud shadow mask and cloud shadow content, and through selecting a threshold value to judge, the threshold value selected by the invention is 10%, namely the image with cloud and shadow content within 10% is regarded as the condition with lighter degree, and the image with cloud and shadow content exceeding 10% is regarded as the condition with heavier degree;
the other is that if the method adopted in step 3.1 regards the clouds or shadows with different degrees as different situations and distinguishes thick clouds, thin clouds, thick shadows and thin shadows, the result of step 3.1 will include cloud shadow information with different degrees, if the result of cloud shadow detection shows that the thin clouds and thin shadows are present, the cloud shadow degree is determined to be light, and if the cloud shadow degree is thick clouds and thick shadows, the cloud shadow degree is determined to be heavy.
For the areas with light cloud and shadow degrees and small occupation ratio, the cloud and shadow are restored directly based on the corresponding images; regarding the areas with heavy cloud and shadow degrees and large proportion, regarding the areas as space observation blind areas; after the cloud-free data synthesis is carried out, a space observation blind area which is effectively covered by observation is still lacked, and the space observation blind area is reconstructed by combining with multi-stage image data, so that high-resolution seamless perception data required by the perception task in the step 1 are obtained;
and 4.1, recovering the cloud and the shadow by utilizing the global observation feature of the image of the cloud and the shadow region with light degree and small occupation ratio, and performing the cloud shadow detection again in the step 3.1 on the image subjected to the step to update the corresponding cloud shadow mask and the cloud shadow occupation ratio. Cloud and shadow recovery can be performed by adopting a homomorphic filtering method and a Retinex method, or a sample set consisting of cloud, cloud-free shadow and cloud-free shadow data can be selected from a sample set obtained from sensing nodes combined in the step 2, and training is performed through models such as a deep neural network (such as U-Net) or a generated countermeasure network (such as DCGAN), so that cloud and shadow recovery is realized;
and 4.2, regarding the cloud and shadow areas with heavier degree and large proportion, because the corresponding effective information in the whole image is less, the difficulty of directly recovering the data is high, and the effect is poor, and the areas are regarded as space observation blind areas.
In addition, the specific processing operation of step 4.1 can be skipped by considering both cases of steps 4.1 and 4.2 as the observation blind areas.
And 4.3, synthesizing the cloud-free data. And removing cloud shadow parts of all the access images one by utilizing a cloud shadow mask, and synthesizing the residual cloud-free parts. The target area T of this step is the spatial observation blind area M obtained in step 3.2 removed from the spatial range a specified in step 1, as shown in (3). Specifically, whether or not there is any cloud data is checked for each pixel position in the target region T, and if there is any cloud data in a plurality of time phases at the position, the pixel value corresponding to the median of the plurality of time phases is taken as the output value of the position. Gaps may exist in the cloud-free data obtained in the step, and the gaps include the spatial observation blind area obtained in the step 3.2 and the cloud and shadow which are not recovered in the step 4.2.
T＝A-M (3)
And 4.4, observing the gap fusion reconstruction and generating seamless data. Firstly, judging whether observation gap fusion reconstruction is needed, and skipping the step 4.4 if the result of the step 4.3 obtains high-resolution seamless perception data meeting the perception task requirement of the step 1. If gaps still exist in the cloudless data synthesized in the step 4.3, due to the lack of effective observation in the space-time range pointed by the step 1, the time range needs to be expanded, and the cloudless observation data covering the observation gaps in the previous and later time periods are combined to be used as auxiliary data to perform fusion reconstruction on the gap region, so that high-resolution cloudless seamless perception data covering the space range designated by the step 1 is finally generated.
The fusion reconstruction method of the observation gap can adopt a multi-temporal spatial perception gap reconstruction method based on gradual radiation adjustment and residual error correction, and the perception gap to be filled is regarded as target data, and the adjacent temporal data of the target data is regarded as auxiliary data. First, the boundary of the target mask is optimized according to the result of superpixel segmentation to ensure that it passes through a homogeneous region in the target image and to avoid spatial discontinuity of the boundary of the perceptual gap reconstruction region. Then, the complementary regions of the auxiliary image are linearly transformed pixel by pixel and used to fill the perceived gap in the target data, which can be achieved by performing a gradual local radiance adjustment based on the same non-cloud regions in the target image and auxiliary image local windows. And finally, residual error correction is carried out on the filled area so as to further eliminate the radiation difference between the reconstruction area and the non-cloud area and realize seamless reconstruction of the perception gap.
The method of time domain filling and residual error correction can also be adopted, cloud-free data available for front and rear time phases are searched pixel by pixel in a gap area, the front and rear time phase data closest to the current time are used as the data of the current time, the radiation difference between the gap area after the time domain filling and the cloud-free area of the current time phase is compared, the residual error correction is carried out on the filling area, and the seamless reconstruction of the perceived gap is realized.
And 5, extracting the earth surface element information and knowledge, constructing an extraction model of the earth surface element information and knowledge by utilizing the training sample set obtained by the sensing nodes combined with the reality and the virtuality in the step 2 according to the high-resolution seamless sensing data obtained in the step 4, and generating a high-resolution earth surface element seamless sensing result. And according to the fact that the earth surface elements belong to a fixed type or a quantitative type, using corresponding type methods including remote sensing indexes, machine learning, physical models and the like.
And for qualitative earth surface elements, adopting a remote sensing earth surface classification method. Taking the impervious surface as an example, the impervious surface is extracted by methods such as impervious surface indexes, object-oriented classification, convolutional neural network and the like.
And for the quantitative earth surface elements, constructing an inversion model of the earth surface elements by adopting an earth surface parameter inversion method of point-surface fusion and combining ground stations and remote sensing images. Taking soil moisture as an example, the soil moisture is inverted by using methods such as a vertical drought index, an L-MEB model, an SCA model, machine learning and the like.
And 6, verifying the precision of the seamless sensing result of the high-resolution surface element. And (4) according to the high-resolution seamless sensing result of the surface element obtained in the step (5), calculating a corresponding precision index according to the surface element belonging to a fixed type or a quantitative type by utilizing a verification sample set obtained by the sensing node combined with the virtual and real in the step (2).
And calculating classification precision indexes such as overall precision, producer precision, user precision and Kappa coefficient for the qualitative type surface elements.
For quantitative surface elements, fit indices, such as root mean square error, correlation coefficients, decision coefficients, are calculated.
The above description is only a preferred embodiment of the present invention, and the specific implementation of the present invention is not to be considered as limited to the above embodiment, and the present invention is not limited to the above embodiment, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present invention should be considered to be within the protection scope of the present invention.
Claims (9)
1. A seamless perception method for high-resolution surface elements is characterized by comprising the following steps:
step 1, decomposing a ground surface element sensing task, and determining a specific sensing task according to the ground surface element task requirement given by a user;
step 2, accessing perception data resources, acquiring data related to surface elements through perception nodes combined by virtuality and reality, and screening and accessing perception data resources meeting the task indexes of the step 1;
the virtual sensing nodes are nodes in a virtual space and comprise social media data, volunteer geographic information data, historical archive information, public surface element products and data acquired by a user; the real sensing nodes comprise sensing nodes existing in a physical space represented by observation platforms and sensors of a space base, a space base and a foundation;
step 3, sensing data quality detection, namely detecting cloud, shadow and space observation blind areas of the image data accessed in the step 2;
step 4, improving the quality of the perception data and generating high-resolution seamless perception data;
step 5, extracting surface element information and knowledge, constructing an extraction model of the surface element information and the knowledge by utilizing a training sample set obtained by the sensing nodes combined with the virtual and the real in the step 2 according to the high-resolution seamless sensing data obtained in the step 4, and generating a high-resolution surface element seamless sensing result;
step 6, verifying the precision of the seamless sensing result of the high-resolution surface element; and (4) according to the high-resolution seamless sensing result of the surface element obtained in the step (5), calculating a corresponding precision index according to the surface element belonging to a fixed type or a quantitative type by utilizing a verification sample set obtained by the sensing node combined with the virtual and real in the step (2).
2. The method of claim 1, wherein: and 2, constructing a model by taking the data accessed by the physical sensing node as a main part and the data of the virtual sensing node as an auxiliary part.
3. The method of claim 1, wherein: the specific process of step 3 is as follows:
step 3.1, detecting the cloud and the shadow in the image data body accessed in the step 2 one by one, and generating a cloud shadow mask corresponding to each image one by one, wherein the cloud shadow mask represents the position of the cloud and the shadow in the corresponding image, and the total amount of the cloud and the shadow can be calculated on the basis;
and 3.2, detecting that the data body accessed in the step 2 can not cover a space observation blind area caused by the perception space-time range in the step 1.
4. The method of claim 1, wherein: the specific process of step 4 is as follows:
step 4.1, recovering the cloud and the shadow by utilizing the global observation feature of the image of the cloud and the shadow region with light degree and small occupation ratio, and performing the step 3.1 cloud shadow detection on the image subjected to the step again to update the corresponding cloud shadow mask and cloud shadow occupation ratio;
step 4.2, regarding the cloud and shadow areas with heavier degree and large proportion, because the corresponding effective information in the whole image is less, the difficulty of directly recovering the data is high and the effect is not good, and regarding the areas as space observation blind areas;
step 4.3, synthesizing the cloud-free data, removing cloud shadow parts of all the access images by using a cloud shadow mask, and synthesizing the remaining cloud-free parts;
and 4.4, observing the gap fusion reconstruction and generating seamless data.
5. The method of claim 1, wherein: the specific process of step 4 is as follows:
regarding the cloud and shadow areas as space observation blind areas;
cloud-free data synthesis, wherein cloud shadow parts of all access images are removed by using a cloud shadow mask, and the rest cloud-free parts are synthesized;
and observing the gap fusion reconstruction and generating seamless data.
6. The method according to claim 4 or 5, characterized in that: the cloud-free data synthesis specifically comprises the following steps:
in the target region T, whether or not there is any cloud data is checked for each pixel position, and if there is any cloud data in a plurality of time phases at the position, the pixel value corresponding to the median of the plurality of time phases is taken as the output value of the position.
7. The method according to claim 4 or 5, characterized in that: the gap fusion reconstruction specifically comprises the following steps:
adopting a multi-temporal spatial perception gap reconstruction method based on gradual radiation adjustment and residual error correction, regarding a perception gap needing to be filled as target data, and regarding adjacent temporal data of the target data as auxiliary data; firstly, optimizing the boundary of a target mask according to the result of superpixel segmentation so as to ensure that the boundary passes through a homogeneous region in a target image and avoid spatial discontinuity of the boundary of a perception gap reconstruction region; then, the complementary area of the auxiliary image is linearly transformed pixel by pixel and used for filling a sensing gap in the target data, and the complementary area can be realized by gradually performing local radiation adjustment based on the same cloud-free area in the local windows of the target image and the auxiliary image; and finally, residual error correction is carried out on the filled area so as to further eliminate the radiation difference between the reconstruction area and the non-cloud area and realize seamless reconstruction of the perception gap.
8. The method according to claim 4 or 5, characterized in that: the gap fusion reconstruction specifically comprises the following steps:
the method of time domain filling and residual error correction is adopted, cloud-free data available for front and rear time phases are searched pixel by pixel in a gap area, the front and rear time phase data closest to the current time are used as the data of the current time, the radiation difference between the gap area after the time domain filling and the cloud-free area of the current time phase is compared, the residual error correction is carried out on the filling area, and the seamless reconstruction of the perceived gap is realized.
9. A system for implementing a high-resolution seamless perception method of surface elements, comprising:
the data access module is used for acquiring data related to the surface elements through sensing nodes combined by virtuality and reality, and screening and accessing sensing data resources meeting the task indexes in the step 1;
the perception data quality detection module is used for detecting cloud, shadow and space observation blind areas of the accessed image data;
the perception data quality improving module generates high-resolution seamless perception data;
the earth surface element information and knowledge extraction module is used for constructing an earth surface element information and knowledge extraction model according to the high-resolution seamless perception data and a training sample set obtained by using a perception node combined with a virtual node and a real node and generating a high-resolution earth surface element seamless perception result;
the precision verification module is used for verifying the seamless perception result of the high-resolution earth surface element; and calculating a corresponding precision index according to the fact that the earth surface element belongs to a fixed type or a quantitative type by utilizing a verification sample set obtained by the sensing node combining the virtuality and the reality according to the seamless sensing result of the high-resolution earth surface element.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011634478.1A CN112767244B (en) | 2020-12-31 | 2020-12-31 | High-resolution seamless sensing method and system for earth surface elements |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011634478.1A CN112767244B (en) | 2020-12-31 | 2020-12-31 | High-resolution seamless sensing method and system for earth surface elements |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112767244A CN112767244A (en) | 2021-05-07 |
| CN112767244B true CN112767244B (en) | 2022-04-01 |
Family
ID=75697982
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011634478.1A Active CN112767244B (en) | 2020-12-31 | 2020-12-31 | High-resolution seamless sensing method and system for earth surface elements |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112767244B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114639027B (en) * | 2022-05-20 | 2023-02-03 | 山东省地质科学研究院 | System and method for identifying low-efficiency town land use according to land use classification data |
| CN114842356B (en) * | 2022-07-01 | 2022-10-04 | 江西师范大学 | High-resolution earth surface type sample automatic generation method, system and equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521598A (en) * | 2011-12-16 | 2012-06-27 | 北京市城市规划设计研究院 | Identification method based on remote-sensing image |
| KR20160048255A (en) * | 2014-10-23 | 2016-05-04 | 인하대학교 산학협력단 | A gis-based thematic mapping method to use in estimating the annual amount of non-point source pollutants using partially updated land-cover map |
| CN108648200A (en) * | 2018-05-10 | 2018-10-12 | 武汉大学 | A kind of indirect city high-resolution impervious surface extracting method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9796400B2 (en) * | 2013-11-27 | 2017-10-24 | Solfice Research, Inc. | Real time machine vision and point-cloud analysis for remote sensing and vehicle control |
| CN111899194A (en) * | 2020-07-30 | 2020-11-06 | 青海省地理空间和自然资源大数据中心 | Method for removing cloud and cloud shadow in remote sensing image |
-
2020
- 2020-12-31 CN CN202011634478.1A patent/CN112767244B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102521598A (en) * | 2011-12-16 | 2012-06-27 | 北京市城市规划设计研究院 | Identification method based on remote-sensing image |
| KR20160048255A (en) * | 2014-10-23 | 2016-05-04 | 인하대학교 산학협력단 | A gis-based thematic mapping method to use in estimating the annual amount of non-point source pollutants using partially updated land-cover map |
| CN108648200A (en) * | 2018-05-10 | 2018-10-12 | 武汉大学 | A kind of indirect city high-resolution impervious surface extracting method |
Non-Patent Citations (2)
| Title |
|---|
| Integrated open geospatial web service enabled cyber-physical information infrastructure for precision agriculture monitoring;Nengcheng Chen 等;《Computers and Electronics in Agriculture》;20141213;第78-91页 * |
| 智慧城市综合感知与智能决策的进展及应用;龚健雅 等;《测绘学报》;20191215(第12期);第1482-1497页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112767244A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huang et al. | A near real-time flood-mapping approach by integrating social media and post-event satellite imagery | |
| Abdollahi et al. | Improving road semantic segmentation using generative adversarial network | |
| De Colstoun et al. | Documentation for the global man-made impervious surface (GMIS) dataset from landsat | |
| Djan'na et al. | Impact of the accuracy of land cover data sets on the accuracy of land cover change scenarios in the Mono River Basin, Togo, West Africa | |
| CN112767244B (en) | High-resolution seamless sensing method and system for earth surface elements | |
| Zhou | Urban High-Resolution Remote Sensing: Algorithms and Modeling | |
| CN116343103B (en) | Natural resource supervision method based on three-dimensional GIS scene and video fusion | |
| Singh et al. | Detection of 2011 Sikkim earthquake-induced landslides using neuro-fuzzy classifier and digital elevation model | |
| CN114943902A (en) | Urban vegetation unmanned aerial vehicle remote sensing classification method based on multi-scale feature perception network | |
| Tran et al. | Improving hydrologic modeling using cloud-free modis flood maps | |
| Mohan et al. | A brief review of recent developments in the integration of deep learning with GIS | |
| Mohamed et al. | Change detection techniques using optical remote sensing: a survey | |
| CN114166842A (en) | Town forest monitoring method based on cooperation of high-resolution remote sensing data and ground survey data | |
| Xie et al. | Land use classification using improved U-Net in remote sensing images of urban and rural planning monitoring | |
| Kumar et al. | Constrained linear spectral unmixing technique for regional land cover mapping using MODIS data | |
| Friehat et al. | Modeling urban sprawls in Northeastern Illinois | |
| Boonpook et al. | PCL–PTD Net: Parallel cross-learning-based pixel transferred deconvolutional network for building extraction in dense building areas with shadow | |
| Liu et al. | A Novel Impervious Surface Extraction Method Based on Automatically Generating Training Samples From Multisource Remote Sensing Products: A Case Study of Wuhan City, China | |
| Tamta et al. | Object-Oriented Approach of Landsat Imagery for Flood Mapping | |
| Sun et al. | Check dam extraction from remote sensing images using deep learning and geospatial analysis: A case study in the Yanhe River Basin of the Loess Plateau, China | |
| LOTFIAN | Urban climate modeling: case study of Milan city | |
| Idris et al. | Application of artificial neural network for building feature extraction in Abuja | |
| Schneiderbauer et al. | Population density estimations for disaster management: Case study rural Zimbabwe | |
| Ankayarkanni et al. | Object based segmentation techniques for classification of satellite image | |
| Nunes et al. | Semi-automatic road network extraction from digital images using object-based classification and morphological operators |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |