CN115858843A - City space map information platform in block form and construction method thereof - Google Patents
City space map information platform in block form and construction method thereof Download PDFInfo
- Publication number
- CN115858843A CN115858843A CN202211469465.2A CN202211469465A CN115858843A CN 115858843 A CN115858843 A CN 115858843A CN 202211469465 A CN202211469465 A CN 202211469465A CN 115858843 A CN115858843 A CN 115858843A
- Authority
- CN
- China
- Prior art keywords
- block
- data
- model
- entity
- case
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a city space map information platform in a block form and a construction method thereof, belonging to the field of city planning; the construction method of the information platform comprises the steps of block form data input, city space map information platform construction, case multi-mode data fusion, block form automatic clustering model training, intelligent link model cluster training and holographic sand table demonstration interaction; the method adopts three-dimensional fusion entity link and artificial intelligence technology to cluster the block forms, and then forms an updated working model of a target block by linking block information of the same case; the method can meet the construction requirements of old city updating, generate a multi-mode data-based street updating work model, expand the acquisition range of old city updating data and improve the working efficiency of old city updating.
Description
Technical Field
The invention belongs to the field of city planning, and particularly relates to a block-shaped city space map information platform and a construction method thereof.
Background
The block is one of the basic components of urban material space form, is the basic unit of urban planning and management, is the most direct environment support of urban buildings, presents various complex forms, and is the current work difficulty in classifying and developing block updating construction efficiently and scientifically. The existing block updating modes mainly include comprehensive improvement, reconstruction and addition, dismantling and reconstruction and the like, different updating modes are different in improvement strength, scale and leading mode, the improvement effect is quite different, the wrong block updating mode selection causes a series of problems of updating failure, economic loss, city development imbalance and the like, but the existing technology is difficult to accurately predict the applicable block updating and developing modes.
At present, the development of updating work of the street is mainly dependent on the subjective classification judgment of experienced designers on the form of the street, the judgment mode has great randomness, the classification conclusion obtained through different classification standards is different, and the classification efficiency is low.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a city space map information platform in a street block form and a construction method thereof.
The purpose of the invention can be realized by the following technical scheme:
a method for constructing a city space map information platform in a street form comprises the following steps:
s1, acquiring road information, building information and space form three-dimensional vector data of natural elements of a block in a target area, and performing unit splitting, block grouping and element coding;
s2, building a spatial form attribute algorithm rule, building a city spatial map information platform, inputting the processed data of the target block in the S1 into a map information platform, and calculating the spatial unit attribute and the spatial correlation attribute of the target block;
s3, obtaining images, characters and model data of the street updating design case, forming three-dimensional fusion entity link, inputting a platform case library, and calculating the spatial unit attribute and the spatial correlation attribute of the case street;
s4, taking the spatial unit attributes and the spatial correlation attributes of the block as machine learning labels, and performing clustering correlation machine learning training on the target block and the case block by adopting a supervised clustering learning algorithm to form a block form automatic clustering model and perform model optimization;
s5, taking the space unit keywords and the space unit attributes of the block as machine learning labels, adopting a supervised clustering learning algorithm to perform data transfer machine learning training on the three-dimensional fusion entity link of the case block and the three-dimensional model of the target block, forming an intelligent link model cluster of various entity data, and constructing the target block to update the multi-mode model;
and S6, demonstrating an updated multi-mode model of the target block on the holographic sand table, feeding back and optimizing an automatic clustering model of the block form and an intelligent link model cluster of various entity data through interactive selection and operation of a user to form an updated work model of the target block, outputting an updated work manual, connecting a 3D printer and outputting an entity model.
Further, in S1, the road information data refers to center line, width and intersection shape data of the street roads subjected to rasterization; the building information data refers to the coordinate position, building height and three-dimensional shape data after rasterization processing, and the natural element data refers to the natural element unit data which is distinguished by using an infrared remote sensing waveband dividing technology and is subjected to vectorization processing.
Further, in S1, the grouping of the blocks refers to grouping the road data expressed as a closed polygon as a block outline, together with the internal building data and natural element data, into block data;
the element coding refers to the ten-digit coding of the entity unit in each block, and the coding basis is that the first six digits are the block serial number, the seventh digit is the belonging entity type, and the last three digits are the entity unit serial number.
Further, in S2, the step of constructing the urban spatial map information platform is as follows:
1) Carrying out structuralization processing on urban space data, carrying out data duplication removal, feature calculation and supplement operation, and generating a space unit attribute and a space association attribute;
2) Digitally encoding the read structured data, packaging the data into entities according to the body, and establishing the relationship among the entities through an algorithm so as to construct an urban space map information platform.
Further, in S3, the step of forming a three-dimensional fusion entity link and entering the platform case library includes:
1) Extracting keywords from a case vocabulary library as an entity, extracting keywords from a case block updating rule and an image library as candidate entities, calculating the matching degree of the entities and the candidate entities by using a supervision method, and forming a two-dimensional fused entity link by using a link with the highest matching degree;
2) And taking the case vocabulary library key words contained in the obtained two-dimensional fusion entity link as an entity, taking the case three-dimensional model library extracted key words as a candidate entity, calculating the matching degree of the entity and the candidate entity by using a supervision method, forming a three-dimensional fusion entity link by using the link with the highest matching degree, and inputting the three-dimensional fusion entity link into a city space map information platform to form a multi-mode database of the block updating design case.
Further, in S4, the step of machine learning training is:
1) According to the following steps of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; a holographic sand table with the platform size not less than 200cm multiplied by 200cm is used for demonstrating a three-dimensional model of a block, an operator wears data gloves with the attitude calculation static precision Roll/pitch not more than 1.0deg to select a case block with reference value for the demonstrated block, and an eye tracker with the 0.4-degree sight tracking precision is used for assisting in acquiring the selection tendency of the operator;
2) And performing block association machine learning training through a 512GB video memory deep learning system, and selecting a machine learning model with strong generalization performance as a block form automatic clustering model through cross validation and generalization inspection.
Further, the model optimization means that whether n necessary incidence relations exist between the case block and the target block is judged; if the necessary incidence relation number is larger than n, outputting the obtained case block; and if the number of necessary incidence relations is less than n, returning to adjust and optimize the automatic clustering model of the block morphology.
Further, in S5, the machine learning training step is:
1) According to the following steps of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; demonstrating a three-dimensional model of a block by using a holographic sand table with a platform size not less than 200cm multiplied by 200cm, selecting entity data from multi-mode data linked with a case block output in the step S4-3 by wearing a data glove with a posture resolving static precision Roll/pitch not more than 1.0deg (RMS) by an operator, transferring the entity data to a space unit of the demonstrated block, and obtaining a selection tendency of the user in an auxiliary manner by using an eye tracker with 0.4-degree sight tracking precision;
2) And performing data transfer machine learning training through a 512GB video memory deep learning system, selecting an updating rule, an effect graph, an analysis graph and a data transfer machine learning model of the built photo with strong generalization performance as an entity data intelligent link model through cross validation and generalization inspection, and further combining the entity data intelligent link model with the entity data intelligent link model to form an entity data intelligent link model cluster.
Further, the step of establishing the block update multi-modal model comprises the following steps: and according to the three case blocks obtained by intelligent matching, performing fusion linking on the updated regulations, the effect graph, the analysis graph, the built photos and the corresponding spatial units of the target block by using the trained entity data intelligent link model cluster.
A city space map information platform in a street form is constructed by using the method.
The invention has the beneficial effects that:
compared with the existing classification method, the automatic clustering model of the street block form obtained through machine learning considers the complex and various forms of the city street block, and can obtain the street block clusters which are similar in shape, and also are similar in terms of spatial relationship, form characteristics and functional zone; the application of the model in the field of block classification allows multi-line parallel calculation of block classification, so that the block classification efficiency is doubled, internal multi-aspect factors except for form are considered, the classification accuracy is improved, and the block updating work of subsequent classification is facilitated;
according to the invention, the multi-mode database of the street updating design case is obtained by obtaining the image, the character and the model data of the street updating design case through linkage, and is intelligently matched with the street to be updated, so that the full automation of the whole process is realized;
the multi-mode street updating model obtained by the invention breaks through single street updating results such as traditional pictures, videos and the like, and provides three updating display modes of holographic sand table interaction, MR mixed display glasses interaction and 3D model printing besides the original traditional text and model display mode by aggregating multi-source information, so that the public can participate in the working process of street updating;
the invention utilizes the holographic sand table and the MR mixed display glasses, can correspond the street shape and the updated effect in real time, shortens the decision time, and realizes the immersive feeling and the real-time operation modification of the street updating work model.
Drawings
In order to more clearly illustrate the embodiments or technical solutions in the prior art of the present invention, the drawings used in the description of the embodiments or prior art will be briefly described below, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without creative efforts.
FIG. 1 is a flow chart of the construction of a city space map information platform in a neighborhood form according to the present invention;
FIG. 2 is a table of exemplary neighborhood clusters in an embodiment of the present invention;
FIG. 3 is a table of updated workbook contents in an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, a method for constructing a city space map information platform in a street form includes the following steps:
s1: acquiring road information, building information and space form three-dimensional vector data of natural elements of a block in a target area, and performing unit splitting, block grouping and element coding;
the road information data refers to the center line, width and intersection shape data of the street roads subjected to rasterization processing; building information data refers to rasterized coordinate position, building height and three-dimensional shape data, and natural element data refers to natural element unit data which is distinguished by using an infrared remote sensing waveband dividing technology and is subjected to vectorization processing;
grouping the blocks, namely grouping the road data expressed as a closed polygon as a block outline, and grouping the road data with the internal building data and the natural element data into block data;
element coding refers to ten-digit coding of the entity units in each block, wherein the coding basis is that the first six digits are block serial numbers, the seventh digit is the type of the entity to which the element belongs, and the last three digits are entity unit serial numbers.
S2: building a spatial form attribute algorithm rule, building a city spatial map information platform, inputting the processed data of the target block in the S1 into a map information platform, and calculating the spatial unit attribute and the spatial correlation attribute of the target block;
constructing a spatial form attribute algorithm rule, constructing an urban spatial map information platform, and performing structured processing on urban spatial data, performing operations such as data duplication removal, feature calculation and supplement, and generating spatial unit attributes and spatial correlation attributes; digitally encoding the read structured data, packaging the data into entities according to an ontology, and establishing a relation between the entities through an algorithm so as to construct an urban space map information platform;
the spatial unit attribute comprises spatial attributes of six units of a water system, a mountain, a road, a block, a land and a building; the spatial correlation attribute is composed of three main types including, adjacent and similar.
S3: acquiring images, characters and model data of a block update design case, forming a three-dimensional fusion entity link, inputting a platform case library, and calculating spatial unit attributes and spatial correlation attributes of a case block;
the method comprises the following steps that a workstation with a 64-core processor or more is used for word data, feature words in a text are counted by a semantic analysis method, a block feature index syntax tree is built, block feature parameters are subjected to priority sorting, keywords are classified, and a block case word library and block updating regulations are extracted;
the image data consists of a case effect image, a case analysis image and a built photo, and the case image is subjected to keyword labeling by using a workstation with more than 64-core processors according to the extraction result of key words in the text entity data through full-pixel semantic segmentation machine learning so as to extract and obtain a block case image library;
the model data consists of a case three-dimensional model and a field elevation model, and the three-dimensional model is organized and coded according to the block organizing rule and the element coding rule by using a workstation with more than 64 core processors; carrying out keyword labeling on the case three-dimensional model through three-dimensional scene online semantic segmentation according to the extraction result of key words in the text data, and extracting to obtain an updated design case three-dimensional model library;
the method comprises the following steps of forming a three-dimensional fusion entity link and inputting a platform case library:
1) Extracting keywords from a case vocabulary library as an entity, extracting keywords from a case block updating rule and an image library as candidate entities, calculating the matching degree of the entities and the candidate entities by using a supervision method, and forming a two-dimensional fused entity link by using a link with the highest matching degree;
2) And (3) taking the case vocabulary library key words contained in the obtained two-dimensional fusion entity link as an entity, taking the key words extracted from the case three-dimensional model library as candidate entities, calculating the matching degree of the entity and the candidate entities by using a supervision method, forming a three-dimensional fusion entity link by using the link with the highest matching degree, inputting the link into a city space map information platform, and forming a multi-mode database of the block updating design case.
S4: taking the spatial unit attributes and the spatial correlation attributes of the blocks as machine learning labels, adopting a supervised clustering learning algorithm to perform clustering correlation machine learning training on the target block 00 and the case blocks to form a block form automatic clustering model, recommending three case blocks with highest correlation degree with the spatial form of the target block in a platform case library, and performing model optimization;
the machine learning training refers to: according to the method of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; demonstrating a three-dimensional model of a block by using a holographic sand table with a platform size not less than 200cm multiplied by 200cm, selecting a case block with a reference value for the demonstrated block by an operator wearing a data glove with an attitude calculation static accuracy Roll/pitch not more than 1.0deg (RMS), and assisting in acquiring the selection tendency of the operator by an eye tracker with 0.4-degree sight tracking accuracy; and performing block association machine learning training through a 512GB video memory deep learning system, and selecting a machine learning model with strong generalization performance as a block form automatic clustering model through cross validation and generalization inspection.
Model optimization, namely judging whether n necessary incidence relations exist between the case block and the target block; if the situation meets the requirement (the number of necessary incidence relations is more than n), outputting the obtained case block; if not (the number of necessary incidence relations is less than n), returning to adjust the automatic clustering model for optimizing the street block shape.
S5: taking the space unit keywords and the space unit attributes of the block as machine learning labels, adopting a supervised clustering learning algorithm, performing data transfer machine learning training on the three-dimensional fusion entity link of the case block and the three-dimensional model of the target block to form an intelligent link model cluster of various entity data, and constructing the target block to update a multi-mode model;
the machine learning training refers to: according to the following steps of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; demonstrating a three-dimensional model of a block by using a holographic sand table with a platform size not less than 200cm multiplied by 200cm, selecting entity data from multi-mode data linked with the case block output in the step S4-3 by an operator wearing a data glove with a posture resolving static precision Roll/pitch not more than 1.0deg (RMS), transferring the entity data to a space unit of the demonstrated block, and assisting to acquire the selection tendency of the operator by an eye tracker with 0.4 degree of sight tracking precision; performing data transfer machine learning training through a 512GB video memory deep learning system, selecting an updating rule, an effect graph, an analysis graph and a data transfer machine learning model of a built photo with strong generalization performance as an entity data intelligent link model through cross validation and generalization inspection, and further combining the entity data intelligent link model with the entity data intelligent link model to form an entity data intelligent link model cluster;
wherein, the multi-modal model is updated in the block, which means that: according to the three case blocks obtained through intelligent matching, the trained entity data are used for intelligently linking the model clusters, updating regulations, effect graphs, analysis graphs and built photos of the case blocks are fused and linked with the corresponding spatial units of the target blocks, and therefore the block updating multi-mode model is obtained.
S6: demonstrating an updated multi-mode model of a target block on a holographic sand table, feeding back and optimizing an automatic clustering model of block form and an intelligent link model cluster of various entity data through interactive selection and operation of a user to form an updated working model of the target block, outputting an updated working manual, connecting a 3D printer and outputting an entity model;
wherein, user's interactive selection and operation indicate: a user wears a data glove with posture resolving static accuracy of Roll/pitch less than or equal to 1.0deg (RMS) to connect three-dimensional model processing software to interact with the holographic sand table, and modifies the model according to case entity data linked by the space unit;
updating the working manual, namely: combining the entity data in the case block, and outputting an updating work manual comprising an updating guide rule, an updating material table, a landscape figurine model schematic diagram and an updating intention diagram;
outputting a solid model, namely: connecting a 3D printer with a molding size of more than 255mm 300mm and supporting a PolyJet 3D printing technology to print a color entity updating model; wherein, the non-updated content is printed by white consumables, the updated content is divided into three categories, including deletion (building removal), modification (facade modification, material modification), addition (greening addition, landscape addition), and printing by color consumables with different colors; the user can wear the MR mixed display glasses to observe the entity updating model, and the updated regulations, the effect graph, the analysis graph and the built-up photo can be linked on the entity updating model to enhance the display.
Example (b):
the technical scheme of the invention is described in detail below by taking an urban area in a certain city as an example.
S1, acquiring three-dimensional vector data of a city by using a multi-rotor unmanned aerial vehicle carrying an oblique photographic camera with more than 4000 ten thousand pixels with the maximum load of more than 3kg and a remote sensing satellite resource III with the flying height of less than 800km, and acquiring city design scheme data and city design standard specification data from a city planning department; the method comprises the following specific steps:
s11, carrying an oblique photography camera with more than 4000 ten thousand pixels by using a multi-rotor unmanned aerial vehicle with the maximum load more than 3kg, acquiring an aerial image in a target area, acquiring spatial form three-dimensional vector data of roads and buildings in a block in the target area of a city center city at a fixed point by using oblique photography data, and acquiring spatial form three-dimensional vector data of city natural elements by using a remote sensing satellite resource III (the multispectral resolution is 5.8 meters) with the flying height of less than 800 km; the road information data refers to street center line, width and intersection shape data in a rasterized target area of a city center city area, the building information data refers to building coordinate positions, building heights (the building heights are calculated through building layer numbers under the condition of no height) and three-dimensional shape data in the rasterized target area, the natural element data refers to natural element unit data of the target area which is distinguished by an infrared remote sensing wave band dividing technology and subjected to vectorization, and the vector data can be integrated into a DWG or SHP format and needs to be added with geographic coordinate data;
s12, taking the city center city road data expressed as a complete closed polygon as a block outline, acquiring an independent block and block outline range, dividing the block and block outline range by using a row index through a DataFrame, grouping three-dimensional vector data in the blocks, importing the current situation as a CAD file of the closed block, internal building data and natural element data into geographic information system software, and exporting the closed multi-segment in-line data into an SHP format;
and S13, carrying out ten-digit coding on the entity units in each block of the city center city, wherein the coding basis is that the first six digits are block serial numbers, the seventh digit is the affiliated entity type, and the last three digits are entity unit serial numbers.
S2, constructing a city space map information platform, constructing a space form attribute algorithm rule, constructing the city space map information platform, inputting the processed data of the target block of the central city area into the platform, and calculating the space unit attribute and the space correlation attribute of the target block; the method comprises the following specific steps:
s21, carrying out structuralization processing on the acquired city spatial data of the target block of the city center city block, and carrying out operations such as urban spatial data deduplication, feature calculation, supplementation and the like by using duplicate and drop _ duplicates data processing methods provided by a python platform to generate spatial unit attributes and spatial correlation attributes;
s22, digitally encoding the read urban center urban structured data by using ProtoBuf, packaging the data according to a body, carrying out error correction and encryption processing on the data, encapsulating the data into an entity, establishing association between the encapsulated entities through a Visual Studio algorithm, and establishing an urban space map information platform according to an association attribute mapping relation, wherein a space unit attribute comprises space attributes of six units of a water system, a mountain body, a road, a block, a land and a building, and the space association attribute comprises three main types of inclusion, adjacency and similarity;
s23, establishing a data folder, establishing a connection with the folder by utilizing a data adding function of the geographic information platform, and inputting the acquired data in each closed block outline represented by the city center and the city into a city space map information platform;
s24, calculating the spatial unit attributes in the target blocks of the urban center and the urban area in the information platform by using workstations with more than 64 core processors; based on NEO4J, the target block form is associated to a feasible knowledge graph network, a corresponding relation in an ontology is established, for example, block Id in the field is a foreign key and is connected to an id field of a block, then, an affiliated relation between the corresponding block and the right of way is established, two attributes of r and theta are calculated for the relation between the blocks according to longitude and latitude information of the entities, semantic similarity between pictures is calculated through a deep learning BM25 model, the picture similarity between block entities is established, association between minimum space units in the same block is established according to the same, adjacent and similar relation between the minimum space unit attributes, and the spatial association attributes are calculated in an information platform.
S3, obtaining images, characters and model data of the street updating design case, forming three-dimensional fusion entity link, inputting a platform case library, and calculating the spatial unit attribute and the spatial correlation attribute of the case street; the method comprises the following specific steps:
s31, acquiring word entity information of an existing block updating design case, processing NLP (non-line-of-sight) by using a workstation with more than 64 core processors and natural language, counting the feature words in the text through semantic analysis, constructing a syntax tree of block feature index shell syntax, performing priority ordering on block feature parameters by using a PowerShell operator, classifying keywords, and extracting to obtain a block case word library and a block updating rule;
s32, acquiring entity data of the existing block updated design case image, wherein the entity data comprises a case effect image, a case analysis image and a built photo, extracting key words of the case image by using a workstation with more than 64-core processors according to key word extraction results obtained by sequencing the syntax trees in S31 and performing key word labeling on the case image through full-pixel semantic segmentation machine learning, and extracting to obtain a block case image library;
s33, acquiring entity data of the three-dimensional model of the existing block updating design case, wherein the entity data consists of the case three-dimensional model and a field elevation model, and grouping and encoding the three-dimensional model by using a workstation with more than 64 core processors; according to the key vocabulary extraction result obtained by sequencing the syntax tree in S31, carrying out on-line semantic segmentation on the three-dimensional scene by using ACM TOG based on hyper-voxel convolution, carrying out keyword labeling on the case three-dimensional model, and extracting to obtain an updated design case three-dimensional model library;
s34, extracting keywords from a case vocabulary library as an entity, extracting keywords from a case block updating regulation and an image library as candidate entities, calculating the matching degree of the entities and the candidate entities by using a supervision method, and forming a two-dimensional fusion entity link by using a link with the highest matching degree;
s35, taking the case vocabulary library keywords contained in the obtained two-dimensional fusion entity link as an entity, taking the case three-dimensional model library extracted keywords as a candidate entity, calculating the matching degree of the entity and the candidate entity by using a supervision method, forming a three-dimensional fusion entity link by using the link with the highest matching degree, inputting the three-dimensional fusion entity link into an urban spatial map information platform, and forming a multi-mode database of the block updating design case;
s36, using a workstation with more than 64 core processors, associating the form of a case block into a linkable knowledge Graph network based on a knowledge Graph Platform provided by Neo4j Graph Platform, establishing an ontology corresponding relation, for example, connecting an external key to an id field of the case block, establishing an affiliated relation between the case block and a land, calculating a Graph relation attribute according to the geographical position information of an entity, expressing a picture into a vector by a Cosin similarity research method, representing the semantic similarity of two pictures by calculating the cosine distance between the vectors, calculating the similarity between the pictures, establishing a picture similarity between the entities of the case block, establishing the association between minimum space units in the same block according to the same, adjacent and similar relations between the minimum space unit attributes, and calculating the unit attribute and the space association attribute of the case library space in the information Platform.
S4, performing clustering association machine learning training on the target block and the case blocks to form a block form automatic clustering model, and recommending three case blocks with highest association degree with the spatial form of the target block in a platform case library, wherein the method comprises the following specific steps:
s41, taking the calculated spatial unit attributes and spatial correlation attributes of the target block and the case block as machine learning labels, and performing the following steps according to the following steps of 6:2:2, dividing a target block and a case block into a training set, a verification set and a test set according to the proportion, demonstrating a three-dimensional model of the block by using a holographic sand table, selecting a case block with reference value for the demonstrated block by an operator wearing data gloves, and obtaining the selection tendency of the operator in an auxiliary way by an eye tracker;
s42, performing block association machine learning training by using a deep learning system, selecting a machine learning model with strong generalization performance as a block form automatic clustering model by a K-fold cross validation method to obtain 6 types of typical blocks, and outputting and judging n necessary association relations in clustering division as shown in FIG. 2;
s43, resolving and screening three case blocks with the highest association degree with the target block by using a block form automatic clustering model and taking the type A as an example, judging whether the case blocks and the target block have 5 necessary association relations, and outputting the obtained case blocks; and if not, returning to adjust the optimized block form automatic clustering model until the output condition is met.
S5, performing data transfer machine learning training on the three-dimensional fusion entity link of the case block and the three-dimensional model of the target block to form an intelligent link model cluster of various entity data, and constructing a target block updating multi-mode model; the method comprises the following specific steps:
s51, taking the space unit keywords and the space unit attributes of the block as machine learning labels, and performing the following steps according to the following steps of 6:2:2, dividing a target block and a case block into a training set, a verification set and a test set in proportion, demonstrating a three-dimensional model of the block by using a holographic sand table, selecting entity data from multi-mode data linked with the case block by an operator wearing data gloves, transferring the entity data to a spatial unit of the demonstrated block, and obtaining the selection tendency of the operator in an auxiliary manner by an eye tracker;
s52, performing data transfer machine learning training by using a deep learning system, selecting a data transfer machine learning model with strong generalization performance, an effect graph, an analysis graph and an established photo as an intelligent entity data link model by using a K-fold cross verification method, and further combining the data transfer machine learning model with the intelligent entity data link model to form an intelligent entity data link model cluster;
and S53, according to the three case blocks obtained through intelligent matching, the updated rules, the effect graphs, the analysis graphs and the built photos of the case blocks are fused and linked with the corresponding spatial units of the target blocks by using the trained intelligent link model clusters, so that an updated multi-mode model of the case blocks is constructed.
S6, demonstrating an updated multi-mode model of the target block on the holographic sand table, feeding back and optimizing an automatic clustering model of the block form and an intelligent link model cluster of various entity data through interactive selection and operation of a user to form an updated work model of the target block, outputting an updated work manual, connecting a 3D printer and outputting an entity model; the method comprises the following specific steps:
s61, demonstrating the constructed case block updating multi-mode model by using a holographic sand table, enabling a user to wear a data glove to connect three-dimensional model processing software to interact with the holographic sand table, and modifying the model according to case entity data linked with spatial units to form a target block updating work model;
s62, the system further performs networking feedback to optimize the street shape automatic clustering model and the intelligent link model cluster according to the selection tendency and the modification behavior of the user;
s63, according to the modified block updating model, combining the entity data in the case block, outputting an updating work manual, wherein the specific content of the manual is shown in FIG. 3; and a 3D printer is connected to print the color entity updated model, a user can wear MR mixed display glasses to observe the entity updated model, and updated regulations, effect diagrams, analysis diagrams and built photos in case data can be linked on the entity updated model to be displayed in an enhanced mode.
In the description herein, references to the description of "one embodiment," "an example," "a specific example," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The foregoing shows and describes the general principles, principal features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are given by way of illustration of the principles of the present invention, but that various changes and modifications may be made without departing from the spirit and scope of the invention, and such changes and modifications are within the scope of the invention as claimed.
Claims (10)
1. A method for constructing a city space map information platform in a street form is characterized by comprising the following steps:
s1, acquiring road information, building information and spatial form three-dimensional vector data of natural elements of a block in a target area, and performing unit splitting, block grouping and element coding;
s2, building a spatial form attribute algorithm rule, building a city spatial map information platform, and inputting the processed data of the target block in the S1 into a map messageInformation processing devicePlatformCalculating the spatial unit attribute and the spatial correlation attribute of the target block;
s3, obtaining images, characters and model data of the block update design case, forming a three-dimensional fusion entity link, inputting a platform case library, and calculating the spatial unit attribute and the spatial correlation attribute of the case block;
s4, taking the spatial unit attributes and the spatial correlation attributes of the block as machine learning labels, and performing clustering correlation machine learning training on the target block and the case block by adopting a supervised clustering learning algorithm to form a block form automatic clustering model and perform model optimization;
s5, taking the space unit keywords and the space unit attributes of the block as machine learning labels, adopting a supervised clustering learning algorithm to perform data transfer machine learning training on the three-dimensional fusion entity link of the case block and the three-dimensional model of the target block, forming an intelligent link model cluster of various entity data, and constructing the target block to update the multi-mode model;
and S6, demonstrating an updated multi-mode model of the target block on the holographic sand table, feeding back and optimizing an automatic clustering model of the block form and an intelligent link model cluster of various entity data through interactive selection and operation of a user to form an updated work model of the target block, outputting an updated work manual, connecting a 3D printer and outputting an entity model.
2. The method for constructing the city space map information platform in the form of the street as claimed in claim 1, wherein in the step S1, the road information data refers to the central line, width and intersection form data of the street after rasterization processing; the building information data refers to the rasterized coordinate position, building height and three-dimensional shape data, and the natural element data refers to the natural element unit data which is distinguished by using an infrared remote sensing waveband division technology and is subjected to vectorization processing.
3. The method as claimed in claim 2, wherein in S1, the grouping of the blocks is to group road data expressed as a closed polygon as a block profile with internal building data and natural element data as block data;
the element coding refers to the ten-digit coding of the entity unit in each block, and the coding basis is that the first six digits are the block serial number, the seventh digit is the belonging entity type, and the last three digits are the entity unit serial number.
4. The method for constructing a street-shaped city space map information platform according to claim 1, wherein in S2, the step of constructing the city space map information platform is as follows:
1) Carrying out structuralization processing on city space data, carrying out data duplicate removal, feature calculation and supplement operation, and generating a space unit attribute and a space association attribute;
2) Digitally encoding the read structured data, packaging the data into entities according to the body, and establishing the relationship among the entities through an algorithm so as to construct an urban space map information platform.
5. The method for constructing the city space map information platform in the form of the neighborhood according to claim 1, wherein in the step S3, the steps of forming a three-dimensional fusion entity link and entering a platform case base are as follows:
1) Extracting keywords from a case vocabulary library as an entity, extracting keywords from a case block updating rule and an image library as candidate entities, calculating the matching degree of the entities and the candidate entities by using a supervision method, and forming a two-dimensional fused entity link by using a link with the highest matching degree;
2) And taking the case vocabulary library key words contained in the obtained two-dimensional fusion entity link as an entity, taking the case three-dimensional model library extracted key words as a candidate entity, calculating the matching degree of the entity and the candidate entity by using a supervision method, forming a three-dimensional fusion entity link by using the link with the highest matching degree, and inputting the three-dimensional fusion entity link into a city space map information platform to form a multi-mode database of the block updating design case.
6. The method for constructing a street-shaped city space map information platform according to claim 1, wherein the step of machine learning training in S4 is as follows:
1) According to the following steps of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; a holographic sand table with the platform size not less than 200cm multiplied by 200cm is used for demonstrating a three-dimensional model of a block, an operator wears data gloves with the attitude calculation static precision Roll/pitch not more than 1.0deg to select a case block with reference value for the demonstrated block, and an eye tracker with the 0.4-degree sight tracking precision is used for assisting in acquiring the selection tendency of the operator;
2) And performing block association machine learning training through a 512GB video memory deep learning system, and selecting a machine learning model with strong generalization performance as an automatic block form clustering model through cross validation and generalization inspection.
7. The method for constructing the city space map information platform in the form of the neighborhood of claim 6, wherein the model optimization in S4 is to determine whether n necessary associations exist between case neighborhoods and target neighborhoods; if the necessary incidence relation number is larger than n, outputting the obtained case block; and if the number of necessary incidence relations is less than n, returning to adjust and optimize the automatic clustering model of the block morphology.
8. The method for constructing a street-shaped urban space atlas information platform according to claim 1, wherein the machine learning training step in S5 is:
1) According to the following steps of 6:2:2, dividing the target block and the case block into a training set, a verification set and a test set according to the proportion; demonstrating a three-dimensional model of a block by using a holographic sand table with a platform size not less than 200cm multiplied by 200cm, selecting entity data from the case block three-dimensional fusion entity link output in the step S3 by an operator wearing a data glove with an attitude calculation static accuracy Roll/pitch not more than 1.0deg (RMS), transferring the entity data to a space unit of the demonstrated block, and assisting to acquire the selection tendency of the operator by using an eye tracker with 0.4-degree sight tracking accuracy;
2) And performing data transfer machine learning training through a 512GB video memory deep learning system, selecting an updating rule, an effect graph, an analysis graph and a data transfer machine learning model of the built photo with strong generalization performance as an entity data intelligent link model through cross validation and generalization inspection, and further combining the entity data intelligent link model with the entity data intelligent link model to form an entity data intelligent link model cluster.
9. The method for constructing a street-shaped city space map information platform according to claim 8, wherein the step of building the street-updated multi-modal model in S5 is as follows: and according to the three case blocks obtained by intelligent matching, performing fusion linking on the updated regulations, the effect graph, the analysis graph, the built photos and the corresponding spatial units of the target block by using the trained entity data intelligent link model cluster.
10. A city space map information platform in a neighborhood form, constructed using the method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211469465.2A CN115858843A (en) | 2022-11-22 | 2022-11-22 | City space map information platform in block form and construction method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211469465.2A CN115858843A (en) | 2022-11-22 | 2022-11-22 | City space map information platform in block form and construction method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115858843A true CN115858843A (en) | 2023-03-28 |
Family
ID=85665113
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211469465.2A Pending CN115858843A (en) | 2022-11-22 | 2022-11-22 | City space map information platform in block form and construction method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115858843A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740506A (en) * | 2023-07-10 | 2023-09-12 | 东南大学 | Morphological feature element identification and extraction method for historical cultural neighborhood |
| CN117314198A (en) * | 2023-10-25 | 2023-12-29 | 北京华清安地建筑设计有限公司 | Comprehensive analysis method and system for historical cultural block function update |
-
2022
- 2022-11-22 CN CN202211469465.2A patent/CN115858843A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740506A (en) * | 2023-07-10 | 2023-09-12 | 东南大学 | Morphological feature element identification and extraction method for historical cultural neighborhood |
| CN117314198A (en) * | 2023-10-25 | 2023-12-29 | 北京华清安地建筑设计有限公司 | Comprehensive analysis method and system for historical cultural block function update |
| CN117314198B (en) * | 2023-10-25 | 2024-03-05 | 北京华清安地建筑设计有限公司 | Comprehensive analysis method and system for historical cultural block function update |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Reconstruction bias U-Net for road extraction from optical remote sensing images | |
| US9940749B2 (en) | Method and system for generating three-dimensional garment model | |
| CN115858843A (en) | City space map information platform in block form and construction method thereof | |
| CN105551084B (en) | A kind of outdoor three-dimensional scenic combination construction method of image content-based parsing | |
| Wu et al. | Automatic road extraction from high-resolution remote sensing images using a method based on densely connected spatial feature-enhanced pyramid | |
| CN112749339A (en) | Tourism knowledge graph-based tourism route recommendation method and system | |
| CN112784002B (en) | Virtual scene generation method, device, equipment and storage medium | |
| CN114758337B (en) | Semantic instance reconstruction method, device, equipment and medium | |
| CN113256649B (en) | Remote sensing image station selection and line selection semantic segmentation method based on deep learning | |
| CN104637090A (en) | Indoor scene modeling method based on single picture | |
| Jiang et al. | Local and global structure for urban ALS point cloud semantic segmentation with ground-aware attention | |
| CN115082254A (en) | Lean control digital twin system of transformer substation | |
| CN117033609A (en) | Text visual question-answering method, device, computer equipment and storage medium | |
| CN115374189A (en) | Block chain-based food safety tracing method, device and equipment | |
| Dehbi et al. | Robust and fast reconstruction of complex roofs with active sampling from 3D point clouds | |
| She et al. | 3D building model simplification method considering both model mesh and building structure | |
| CN116401736A (en) | Urban green space automatic layout method based on artificial intelligence | |
| CN113822232B (en) | Pyramid attention-based scene recognition method, training method and device | |
| CN110889718A (en) | Method and apparatus for screening program, medium, and electronic device | |
| CN111860517B (en) | Semantic segmentation method under small sample based on distraction network | |
| CN115905442A (en) | Method, system and medium for surveying landform of unmanned aerial vehicle based on cognitive map | |
| CN115617222A (en) | Virtual display system of literary composition product VR | |
| CN114840680A (en) | Entity relationship joint extraction method, device, storage medium and terminal | |
| Cui | Research on garden landscape reconstruction based on geographic information system under the background of deep learning | |
| Le Van et al. | An effective RGB color selection for complex 3D object structure in scene graph systems |
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 |