CN-121980040-A - Method and device for updating urban small micro space based on regional characteristics

CN121980040ACN 121980040 ACN121980040 ACN 121980040ACN-121980040-A

Abstract

The application relates to a method and a device for updating urban small spaces based on regional characteristics, and belongs to the technical field of urban planning. The method comprises the steps of collecting the updated small micro-space cases, constructing a multi-layer classification system, classifying and labeling each case, extracting a first feature vector of each case, associating the first feature vector with a corresponding image, text and classification labels, storing the first feature vector and the corresponding image, text and classification labels into a database, obtaining current situation data and region data of the small micro-space to be updated, extracting a second feature vector of the small micro-space to be updated according to the current situation data, identifying classification attributes, extracting region features according to the region data, taking the second feature vector and the classification attributes of the small micro-space to be updated as composite query conditions, retrieving the case matched with the small micro-space to be updated from the database as a reference case, and generating an updating scheme of the small micro-space to be updated based on the region features of the small micro-space to be updated and updated multi-mode solution data of the reference case.

Inventors

SHEN YITING
WANG SIQI
GUO LUNING
Gao Chaoxuan
LI MINGZE
JING KUN
CAO ZILIANG
QU YAO
AN YI
CHEN HUIYUN
Shan Yanming
TIAN JIAXING
WANG BEINI
LIU JUAN
TIAN JING
JIA MENG
ZHAN HONG
REN YUE

Assignees

北京建筑大学

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. A method for updating a small urban subspace based on regional characteristics, comprising: Collecting updated small microspace cases, each case comprising pre-update multimodal problem data and updated multimodal solution data, the problem data comprising a problem image and a problem text, the solution data comprising a result image and a solution text; Constructing a multi-layer classification system, classifying each case according to the multi-layer classification system, labeling labels, extracting first feature vectors of each case by using a multi-mode pre-training model, and storing the first feature vectors, corresponding images, texts and classification labels in a database in an associated manner; Acquiring current situation data and region data of a small micro space to be updated, extracting a second feature vector of the small micro space to be updated according to the current situation data, identifying classification attributes of the small micro space to be updated, and extracting region features according to the region data; Taking the second feature vector and the classification attribute of the small micro space to be updated as a composite query condition, and retrieving a case matched with the small micro space to be updated from the database as a reference case; and generating an updating scheme of the small micro space to be updated based on the regional characteristics of the small micro space to be updated and the updated multi-mode solution data of the reference case.
2. The method for updating urban small subspaces based on regional characteristics according to claim 1, wherein the multi-layer classification system at least comprises function type labels divided according to the functional attributes of the small subspaces, morphological characteristic labels divided according to the spatial morphology and texture characteristics of the small subspaces, and facility constitution labels divided according to the facilities and material types in the small subspaces.
3. The method for updating a small urban subspace based on regional characteristics of claim 1 wherein extracting the first feature vector of each case using a multimodal pre-training model comprises, for each case, encoding the problem image and the outcome image into a problem image feature vector and an outcome image feature vector by an image encoder in the multimodal pre-training model, and encoding the problem text and the solution text into a problem text feature vector and a solution text feature vector by a text encoder, the problem image feature vector, the problem text feature vector, the outcome image feature vector and the solution text feature vector constituting the first feature vector.
4. A method of updating urban subspaces based on regional characteristics according to claim 1, wherein the presence data comprises presence question images and/or presence question text; the extracting the second feature vector of the small micro space to be updated and identifying the classification attribute of the small micro space to be updated according to the current situation data comprises the following steps: An image encoder in the multi-mode pre-training model encodes the current problem image into a current problem image feature vector, a text encoder encodes the current problem text into a current problem text feature vector, and the current problem image feature vector and/or the current problem text feature vector form the second feature vector; and determining the classification attribute of the small micro space to be updated based on the multi-layer classification system.
5. The method for updating urban subspaces based on regional characteristics according to any one of claims 1-4, wherein the regional data comprises at least climate environment data, historical cultural attribute data and ambient build-up environment data; the step of extracting the region characteristics according to the region data comprises the steps of carrying out structural processing on the region data, and respectively extracting climate characteristics, cultural characteristics and built environment characteristics through statistical analysis, semantic analysis and space analysis to form the region characteristics of the small micro space to be updated.
6. A method for updating a small urban subspace based on regional characteristics according to any one of claims 1-4, wherein said retrieving from said database a case matching said small subspace to be updated as a reference case comprises: And calculating the similarity between the second feature vector and the first feature vector of the cases in the database, taking the cases meeting the similarity threshold as candidate cases, carrying out constraint screening by combining the classification attribute of the small micro space to be updated and the classification label of the candidate cases, and taking the candidate cases meeting the screening condition as the reference cases.
7. The method for updating a small urban subspace based on a regional signature of claim 6, wherein the generating an update scheme for the small subspace to be updated based on the regional signature of the small subspace to be updated and the updated multi-modal solution data of the reference case comprises: And taking the regional characteristics as constraint conditions or adjustment parameters, taking the scheme text and the result image of the reference case as adjustment objects, inputting the adjustment objects into a generated artificial intelligent model, driving the generated artificial intelligent model to adjust elements related to the regional characteristics in the scheme text and the result image by constructing a prompt word instruction, and taking the adjusted scheme text and result image as an updating scheme of the small micro space to be updated.
8. An apparatus for updating urban mini-spaces based on regional characteristics, the apparatus comprising: the system comprises a case collection module, a case analysis module and a case analysis module, wherein the case collection module is used for collecting updated cases of a small micro-space, each case comprises pre-update multi-modal problem data and updated multi-modal solution data, the problem data comprises a problem image and a problem text, and the solution data comprises a result image and a solution text; The data classification storage module is used for constructing a multi-layer classification system, classifying each case according to the multi-layer classification system, labeling labels, extracting first feature vectors of each case by using a multi-mode pre-training model, and storing the first feature vectors, corresponding images, texts and classification labels in a database in an associated manner; The data acquisition and processing module is used for acquiring current situation data and region data of the small micro space to be updated, extracting a second feature vector of the small micro space to be updated according to the current situation data, identifying classification attributes of the small micro space to be updated, and extracting region features according to the region data; The retrieval module is used for retrieving a case matched with the small micro space to be updated from the database as a reference case by taking the second feature vector and the classification attribute of the small micro space to be updated as a composite query condition; And the updating scheme generating module is used for generating an updating scheme of the small micro-space to be updated based on the regional characteristics of the small micro-space to be updated and the updated multi-mode solution data of the reference case.
9. An electronic device comprising a memory and a processor, the memory storing a computer program that when executed by the processor implements a method of updating urban subspaces based on geographic features as claimed in any one of claims 1-7.
10. A storage medium having stored thereon a computer program which, when executed by a processor, implements a method of updating urban subspaces based on geographical features as claimed in any one of claims 1-7.

Description

Method and device for updating urban small micro space based on regional characteristics Technical Field The application belongs to the technical field of urban planning, and particularly relates to a method and a device for updating urban small micro-space based on regional characteristics. Background In the process of urban treatment, public spaces and traditional courtyards face the problems of old facilities and the like, and transformation is realized by updating small micro-spaces. The patent application with publication number CN119938963A discloses a database management system based on building remote sensing images, which is used for identifying target images from the building remote sensing images, dividing the target images into image sets, correspondingly importing the image sets into a data layer of a library architecture, generating an image database, constructing an associated channel with a break-through point, configuring on-off conditions on the associated channel, analyzing inquiry requirements from request data, and selecting a corresponding inquiry mode. In addition, the patent application with the publication number of CN117436519A relates to a construction method of a construction engineering database supporting a knowledge graph, which is used for obtaining a construction engineering data source, constructing a construction engineering knowledge graph associated database frame by adopting the knowledge graph, constructing a construction engineering label pool in the construction engineering knowledge graph associated database frame, marking the construction engineering data source with labels according to the database label pool, and forming the construction engineering knowledge graph associated database according to label marking results. However, the two patent applications have a single data mode, the current problem of the small micro space cannot be deeply understood only by remote sensing images or engineering labels, the regional characteristics are not considered, the functions are only remained on the data query level, and the generation of a guiding scheme cannot be conducted. Disclosure of Invention In view of the above analysis, the embodiment of the invention aims to provide a method and a device for updating urban small micro-space based on regional characteristics. The purpose of the invention is realized in the following way: An embodiment of a first aspect of the present invention provides a method for updating a small urban subspace based on regional characteristics, including: Collecting updated small microspace cases, each case comprising pre-update multimodal problem data and updated multimodal solution data, the problem data comprising a problem image and a problem text, the solution data comprising a result image and a solution text; Constructing a multi-layer classification system, classifying each case according to the multi-layer classification system, labeling labels, extracting first feature vectors of each case by using a multi-mode pre-training model, and storing the first feature vectors, corresponding images, texts and classification labels in a database in an associated manner; Acquiring current situation data and region data of a small micro space to be updated, extracting a second feature vector of the small micro space to be updated according to the current situation data, identifying classification attributes of the small micro space to be updated, and extracting region features according to the region data; Taking the second feature vector and the classification attribute of the small micro space to be updated as a composite query condition, and retrieving a case matched with the small micro space to be updated from the database as a reference case; and generating an updating scheme of the small micro space to be updated based on the regional characteristics of the small micro space to be updated and the updated multi-mode solution data of the reference case. Further, the multi-layer classification system at least comprises a function type label divided according to the function attribute of the small micro space, a morphological characteristic label divided according to the space morphology and texture characteristics of the small micro space and a facility constitution label divided according to the facilities and material types in the small micro space. Further, the extracting the first feature vector of each case by using the multi-mode pre-training model comprises the steps that for each case, an image encoder in the multi-mode pre-training model encodes the problem image and the result image into a problem image feature vector and a result image feature vector, and a text encoder encodes the problem text and the scheme text into a problem text feature vector and a scheme text feature vector, wherein the problem image feature vector, the problem text feature vector, the result image feature vector and the scheme text feature vector form the