Street tree realistic style design method based on artificial intelligence
Data is collected through GAN architecture and drone technology to generate high-precision street tree design drawings, solving the problem that traditional design relies on manual experience and achieving efficient, accurate and consistent urban greening design.
Patent Information
- Application Number
- CN202510425199.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-07
- Publication Date
- 2025-08-01
AI Technical Summary
Traditional street tree design relies on manual experience and lacks objective basis and quantitative indicators, which leads to strong subjectivity, time-consuming, high cost and difficulty in achieving high efficiency and consistency in design results.
Generative adversarial network (GAN) architecture is adopted, data is collected in combination with drones and ground measurement technology, and high-precision street tree design drawings are generated. The design is optimized through quantitative evaluation system and virtual simulation tools to achieve automated drawing generation.
The generated drawings are highly consistent with the real scene, reducing labor costs, improving design efficiency and accuracy, adapting to different geographical conditions, quickly responding to different urban needs, reducing duplicate labor, and meeting personalized and customized needs.
Smart Images
Figure CN120409199A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a design method, and particularly to a realistic style design method for roadside trees based on artificial intelligence. Background Art
[0002] With the rapid development of China's economy and the continuous improvement of people's living standards, people's requirements for the living environment are also getting higher and higher. The construction of landscaping has become an important part of urban construction. As an important form of greening on both sides of urban roads, roadside trees play an important role in beautifying the city appearance, improving the ecological environment, and enhancing the urban image.
[0003] However, there are some problems in the current traditional design methods for roadside trees. Traditional designs mainly rely on the experience and intuition of designers, lacking objective basis and quantitative indicators, resulting in strong subjectivity of design results and being easily affected by personal factors. Due to the need for a large amount of manpower and material resources, traditional designs often take a long time and cost a lot. Traditional designs are usually implemented according to fixed templates or general standards, and cannot be flexibly adjusted and customized according to specific situations. In addition, the design drawings manually drawn often lack consistency and accuracy, making it difficult to achieve large-scale and high-efficiency urban greening design. Summary of the Invention
[0004] Object of the Invention: The object of the present invention is to provide a realistic style design method for roadside trees to generate high-precision and highly consistent design drawings, while enhancing the urban aesthetics, ecological functions, and the scientific nature and adaptability of greening design.
[0005] Technical Solution: The design method described in the present invention includes the following steps:
[0006] S1. Collect the basic data and environmental parameters of roadside trees;
[0007] S2. Input design parameters according to the requirements of the realistic style, and the design parameters include tree parameters, road parameters, and seasonal landscape parameters;
[0008] S3. Analyze and simulate the design parameters based on the generative adversarial network (GAN) architecture to generate realistic style design drawings of roadside trees.
[0009] Preferably, in S1, drones and ground measurement technologies are used to collect data.
[0010] Preferably, the basic data of the roadside trees in S1 includes the tree species distribution, tree species quantity, tree specifications, tree growth conditions, and the current situation of the road green belt.
[0011] Preferably, the environmental parameters in S1 include climate conditions, soil conditions, hydrological conditions, vegetation conditions, and topography and geomorphology.
[0012] Preferably, the tree parameters in S2 include the main tree species to be used, the tree species ratio, the tree species distribution, the tree specifications, the tree shapes, the crown shapes, the growth rate of the trees, the adaptability of the trees to environmental conditions, and the planting spacing between street trees.
[0013] Preferably, the road parameters in S2 include the road type and the road pattern. The road type includes expressways, arterial roads, sub-arterial roads, and branch roads. The road pattern includes one-way two-lane, two-way three-lane, three-way four-lane, and four-way five-lane.
[0014] Preferably, the design drawings in S3 include two-dimensional floor plans and three-dimensional renderings.
[0015] Preferably, it also includes verifying the design drawings through simulation software.
[0016] Advantages: Compared with the prior art, the present invention has the following remarkable advantages: 1. Through the dual-module collaborative mechanism of GAN, iterative training is carried out to optimize the detail restoration degree. The generated drawings have high consistency with the real scene. A quantitative evaluation system is used to conduct multi-dimensional verification of the generated drawings, improving the scientificity and accuracy. Combining virtual simulation tools to dynamically verify the influence of environmental factors such as light and wind fields on the tree morphology, and avoiding design defects in advance, thereby improving the quality and accuracy of the design drawings; 2. Through the automated drawing generation technology of the generative adversarial network GAN, it replaces the design process relying on manual experience, greatly reducing the time-consuming and improving the output speed. Combining drone aerial photography and ground sensing equipment to efficiently collect the basic information of street trees, compressing the data collection cycle, reducing the labor cost, and realizing rapid iteration of the scheme through modular parameter configuration, flexibly adapting to diverse needs, and avoiding repetitive labor, thereby comprehensively improving the efficiency, shortening the cycle, and controlling the cost; 3. Through GAN to batch generate design schemes adapted to different road types and seasonal characteristics, quickly responding to the differentiated needs of urban areas. By integrating the standardized framework and dynamic parameters, it takes into account both efficiency and customization needs, and promotes the efficient implementation of large-scale greening projects; 4. Through the dynamic balance training strategy of GAN, it adapts to different geographical conditions, generates design schemes that scientifically match the regional characteristics, supports users to customize key parameters, meets personalized scenario needs, and improves the urban aesthetics, ecological functions, and the scientificity and adaptability of greening design. Description of the Drawings
[0017] Figure 1 It is a flowchart of the present invention;
[0018] Figure 2 It is a schematic diagram of the algorithm framework of the present invention;
[0019] Figure 3 It is a schematic diagram of the process of designing the parameters of Paulownia fortunei in the present invention;
[0020] Figure 4This is a schematic diagram of the two-dimensional plane drawing of plane trees in early autumn output by the present invention;
[0021] Figure 5 This is a schematic diagram of the three-dimensional effect drawing of plane trees in early autumn output by the present invention;
[0022] Figure 6 This is a schematic diagram of the process of inputting the design parameters of camphor trees and ginkgo trees by the present invention;
[0023] Figure 7 This is a schematic diagram of the two-dimensional plane drawing of camphor trees and ginkgo trees in late autumn output by the present invention;
[0024] Figure 8 This is a schematic diagram of the three-dimensional effect drawing of camphor trees and ginkgo trees in late autumn output by the present invention. Detailed implementation manners
[0025] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.
[0026] Example 1
[0027] Drawing the design drawings with the "century-old plane trees" in Nanjing as the main street tree species, including the following steps:
[0028] S1. Use unmanned aerial vehicles and ground measurement technologies to collect the basic data and environmental parameters of the "plane trees" on the streets in Nanjing.
[0029] The basic data includes the distribution and quantity of street tree species. There are about 90,000 plane trees on both sides of the main roads in Nanjing; the tree specifications, the average tree height of adult plane trees is 25m, the diameter at breast height is 0.5m, the crown width is 15m, and the height of the branches below is 5m; the growth status of the trees, the plane trees grow healthily, without obvious aging and diseases and pests, and the growth trend is good; the current situation of the road green belt, the width of the street tree green belt is 3m, and the configuration mode is permeable, mainly with plane trees.
[0030] The environmental parameters include climatic conditions. The average temperature in early autumn is between 20-27°C, the precipitation is 50-100mm, the air humidity is 60-80%, and the dominant wind direction is northeast wind or northerly wind; soil conditions, the soil type is yellow-brown soil, the pH value is 6.5, the humidity is moderate, and the nutrient status is good; hydrological conditions, the groundwater level is 1.5 meters; vegetation conditions, the area where the plane trees are located is an urban green belt, the vegetation coverage is 80%, and the biodiversity index is medium; topographic and geomorphic conditions, the growth area of the plane trees is a plain, and the slope is gentle.
[0031] Unify the format of the collected data. For example, convert the image data into a matrix form that can be processed by the model, and perform standardization or normalization processing on the parameter data to eliminate the dimension difference, which is convenient for model calculation and learning.
[0032] S2. Input the design parameters.
[0033] Such as Figure 3As shown, input the design parameters according to the realistic style requirements, including tree parameters, road parameters, and seasonal landscape parameters.
[0034] The tree parameters include determining the tree species as "Platanus acerifolia (Ait.) Willd."; the tree specifications, with a tree height of 25m, a diameter at breast height of 0.5m, a crown width of 15m, and a height of the lowest branch of 5m; the tree shape and crown form, with the crown being bell-shaped, having good fullness, and looking majestic; the growth rate of the tree, medium; the adaptability of the tree to environmental conditions, with medium cold tolerance, good heat tolerance, medium drought tolerance, and strong pollution resistance; the planting spacing between street trees, 8 meters.
[0035] The road parameters include the road type being the arterial road; the road pattern being the three-board and four-belt type.
[0036] The seasonal landscape parameters are set as early autumn.
[0037] S3. Artificial intelligence design.
[0038] Initialize the parameters of the generator and discriminator based on the Generative Adversarial Network (GAN) architecture, select an appropriate network depth and width to ensure that the model has sufficient expressive power and learning ability, simulate the realistic style of Platanus acerifolia in early autumn, and automatically generate the design drawings of street trees, including the two-dimensional plan view and three-dimensional rendering of Platanus acerifolia.
[0039] Training process:
[0040] (1) Generator training: Use random noise and partial real sample data as inputs to generate preliminary design drawings, and adjust the parameters of the generator through the backpropagation algorithm to minimize the difference between the generated drawings and the real drawings. Adopt the generative loss function in the generative adversarial network, such as the least squares loss function, etc., to guide the generator to generate more realistic design drawings.
[0041] (2) Discriminator training: Input the drawings generated by the generator and the real design drawings into the discriminator together, train the discriminator to accurately distinguish between the two, and improve its discrimination ability by optimizing the loss function of the discriminator, such as the binary cross-entropy loss function, to provide more accurate feedback for the generator and promote the generator to continuously improve the output results.
[0042] (3) Alternating training: Cycle through the training of the generator and the discriminator to maintain the dynamic balance between the two, avoid premature convergence of one network or over-suppression of the other network, and gradually improve the quality and realism of the generated drawings.
[0043] (4) Use multi-dimensional evaluation metrics to comprehensively measure the performance of the model, including the visual quality of the generated drawings (such as clarity, color accuracy, etc.), the compliance with the actual design requirements (such as the accuracy of tree specifications, road parameters, etc.), and the stability of the model (such as the consistency and repeatability of the generated results).
[0044] (5) Optimize the model structure or training parameters according to the evaluation results. For example, if the generated drawings perform poorly in some details, the parameters of the corresponding feature extraction layer in the generator can be adjusted or specific loss function terms can be added to strengthen the model's attention and learning of these details; if the model has overfitting or underfitting phenomena, optimization can be carried out by adjusting the number of network layers, the number of neurons, regularization parameters, etc., to improve the generalization ability and adaptability of the model.
[0045] S4. Output of design drawings.
[0046] Such as Figure 4 (a), Figure 4 (b) and Figure 5 (a), Figure 4 As shown in (a) and (b), select appropriate two-dimensional floor plans and three-dimensional renderings to display the detailed layout and expected effects of plane trees in early autumn. Perform necessary post-processing operations on the design drawings generated by the model, such as image enhancement and detail modification, to further improve the visual effect and practicality of the drawings, making them more in line with professional design standards and actual construction requirements.
[0047] Example 2
[0048] Drawing of design drawings with "camphor trees and ginkgo trees" as the main street tree species in Nanjing, including the following steps:
[0049] S1. Use unmanned aerial vehicles and ground measurement technologies to collect basic data of camphor trees and ginkgo trees as street trees in the Nanjing area.
[0050] The basic data includes the distribution and quantity of street tree species, with camphor trees and ginkgo trees planted at intervals; tree specifications, the average tree height of camphor trees is 15m, the diameter at breast height is 0.4m, the crown width is 12m, and the height below branches is 4m; the average tree height of ginkgo trees is 20m, the diameter at breast height is 0.5m, the crown width is 10m, and the height below branches is 3m; the growth status of the trees, camphor trees and ginkgo trees grow healthily, without obvious senescence and pests and diseases, and have good growth; the current situation of the road green belt, the width of the street tree green belt is 3m, the configuration mode is interval planting, and camphor trees and ginkgo trees are arranged alternately.
[0051] The environmental parameters include climate conditions, the average temperature in late autumn is 18°C, the precipitation is 150mm, the air humidity is 60%, and the dominant wind direction is northeast; soil conditions, the soil type is yellow brown soil, the pH value is 6.5, the soil humidity is moderate, and the soil nutrient status is good; hydrological conditions, the groundwater level is 1.5m, and the Qinhuai River and the Yangtze River have a positive impact on the growth of camphor trees and ginkgo trees; vegetation conditions, the area where camphor trees and ginkgo trees are located is an urban green belt, the vegetation coverage is 80%, and the biodiversity index is medium; topography and geomorphology, the growth area of camphor trees and ginkgo trees is a plain, with a gentle slope and an altitude of 20 meters.
[0052] S2. Input of design parameters.
[0053] As Figure 6 shown, input design parameters according to the realistic style requirements, including tree parameters, road parameters, and seasonal landscape parameters.
[0054] Tree parameters include determining the tree species as camphor trees and ginkgo trees; tree specifications, the height of the ginkgo tree is 20m, the diameter at breast height is 0.5m, the crown width is 10m, and the height of the branches below is 3m; the height of the camphor tree is 15m, the diameter at breast height is 0.4m, the crown width is 12m, and the height of the branches below is 4m; the tree shape and crown shape, the ginkgo tree crown is conical and looks straight; the camphor tree crown is oval, with good fullness and beautiful appearance; the growth rate of the trees, the ginkgo is slow; the camphor is medium; the adaptability of the trees to environmental conditions, the ginkgo has good cold resistance, medium heat resistance, good drought resistance, and medium pollution resistance; the camphor has medium cold resistance, good heat resistance, medium drought resistance, and strong pollution resistance; the planting spacing between street trees, the ginkgo is 10m, and the camphor is 8m.
[0055] Road parameters include the road type as the main road; the road pattern as the three-board and four-belt type.
[0056] The seasonal landscape parameters are set as late autumn, with the camphor tree leaves being dark green and the ginkgo tree leaves being golden yellow.
[0057] S3, Artificial intelligence design.
[0058] Analyze the above parameters based on the generative adversarial network (GAN) architecture, simulate the realistic style of camphor trees and ginkgo trees in the late autumn season of Nanjing, and automatically generate street tree design drawings, including two-dimensional floor plans and three-dimensional renderings of camphor trees and ginkgo trees.
[0059] S4, Design drawing output.
[0060] As Figure 7 (a), Figure 7 (b) and Figure 8 shown, select suitable two-dimensional floor plans and three-dimensional renderings to display the detailed layout and expected renderings of camphor trees and ginkgo trees in the late autumn of Nanjing.
Claims
1. A realistic style design method for street trees based on artificial intelligence, characterized in that, It includes the following steps: S1. Collect the basic data and environmental parameters of street trees; S2. Input design parameters according to the requirements of realistic style, and the design parameters include tree parameters, road parameters and seasonal landscape parameters; S3. Analyze and simulate the design parameters based on the generative adversarial network (GAN) architecture to generate realistic style design drawings of street trees.
2. The artificial intelligence design method according to claim 1, wherein In S1, drones and ground measurement technologies are used to collect data.
3. The artificial intelligence design method according to claim 1, wherein The basic data of the street trees in S1 includes the tree species distribution, tree species quantity, tree specifications, tree growth conditions and the current situation of road green belts.
4. The design method according to claim 1, wherein The environmental parameters in S1 include climate conditions, soil conditions, hydrological conditions, vegetation conditions, topography and landforms.
5. The design method according to claim 1, characterized in that The tree parameters in S2 include the main tree species to be used, tree species proportion, tree species distribution, tree specifications, tree shapes, crown shapes, tree growth rates, the adaptability of trees to environmental conditions, and the planting spacing between street trees.
6. The design method according to claim 1, characterized in that The road parameters in S2 include road types and road patterns. The road types include expressways, arterial roads, sub-arterial roads and branch roads, and the road patterns include one-way two-lane, two-way three-lane, three-way four-lane and four-way five-lane.
7. The design method according to claim 1, characterized in that, The design drawings in S3 include two-dimensional plan views and three-dimensional effect drawings.
8. The design method according to claim 1, wherein It also includes verifying the design drawings through simulation software.
9. An electronic device, comprising: A processor; And a memory arranged to store computer-executable instructions, and the executable instructions, when executed, cause the processor to execute the method according to any one of claims 1 to 8.
10. A computer-readable storage medium storing one or more programs, and when the one or more programs are executed by an electronic device including a plurality of application programs, the electronic device is caused to execute the method according to any one of claims 1 to 8.