Urban design generation auxiliary method and device

Abstract

The invention provides a city design generation assisting method and device. The city design generation assisting method comprises the following steps that S1, an existing design scheme of a designated designer is collected to generate an original image-planning image pair; s2, calculating multiple urban design morphological parameters of the regional planning image, and classifying the regional original image to obtain a classification result; s3, clustering is carried out according to the urban design morphological parameters and the classification result, and a plurality of design style preferences are obtained; s4, constructing a training data set of each design style preference; s5, training according to the training data set to obtain a style design model; step S6, selecting an optimal design style preference; s7, inputting the image of the to-be-planned area into the style design model corresponding to the optimal design style preference to obtain an initial design drawing; and S8, performing multi-objective optimization on the initial design drawing to obtain an auxiliary design drawing. In a word, the method can generate the regional planning design drawing with the style preference of a specified designer.

Description

Technical Field

[0001] The invention relates to a city layout design method, and in particular to a city design generation auxiliary method and device. Background Art

[0002] In traditional block design, designers artificially create by imagining a future scene. With the development of digital modeling software, there are more studies on using models to create an urban block that meets the designer's expectations. Existing studies have fully explored the data analysis methods used to solve specific problems in different design links, including classic quantitative analysis models, big data analysis methods, and diagnostic methods that use AI to mine the laws of urban development in recent years, as well as system dynamics models, cellular automaton models, agent model game deduction, prediction models, etc., to obtain prediction methods based on objective conditions to describe the future. Parametric design methods and some software tools that use AI to generate design solutions solve the problem of generating design solutions based on the process of generating planning and design solutions based on knowledge.

[0003] However, the application of diversified auxiliary tools has increased the complexity of the design process. Designers often need to spend hours setting up, running and testing the model to complete an evaluation, which has a negative impact on design efficiency. In the traditional computer-aided design process, designers have two tendencies. One is to convert the design intent into specific generation conditions and their parameters, and then generate them through computer software. This strategy separates the intent from the design process and is usually called "pre-setting"; the other tendency is to use software to conduct multi-dimensional evaluation after design. This strategy separates the design process from performance optimization and is usually called "post-evaluation". It has been proven that both of the above paradigms may lead to insufficient optimization of the design, which will only bring small benefits and huge costs.

[0004] As a result, derivative research began to focus on strategies for design interaction. On the one hand, interactivity was improved during the design process. Some researchers established a visual interface between generative tools and 3D modeling platforms. These software and plug-ins made it easier for generative tools to provide assistance during the design process. On the other hand, the time-consuming problem of simulation engines hindered the rapid iteration of the design process, so scholars began to study some technologies to accelerate the generation or evaluation of design solutions. The typical approach is to simplify the underlying algorithm to achieve model acceleration.

[0005] However, the basic principles of the above methods are similar to those of mainstream methods, the improvement effect is very limited, and it is still difficult to obtain real-time feedback during the design process. Designers cannot effectively create ideal designs directly through the application of these tools. Summary of the invention

[0006] The present invention is made to solve the above-mentioned problems, and its purpose is to provide an auxiliary method and device for urban design generation.

[0007] The present invention provides an auxiliary method for generating an urban design, which is used for generating an auxiliary design drawing with a designated designer style corresponding to a to-be-planned area, and has the following characteristics, comprising the following steps: step S1, collecting a regional planning image of an existing design scheme of a designated designer and a corresponding original image of an unplanned area, and generating a plurality of original image-planning image pairs; step S2, for each original image-planning image pair, calculating a plurality of urban design morphological parameters of the regional planning image, and classifying the regional original image to obtain a classification result; step S3, clustering the urban design morphological parameters and classification results corresponding to each original image-planning image pair, and obtaining a plurality of classes as design style preferences; step S4, for each design style preference, using all corresponding original image-planning image pairs as corresponding training data sets; step S5, for each design style preference, constructing a corresponding initial model and training according to the corresponding training data set to obtain a corresponding style design model; step S6, selecting a corresponding design style preference as an optimal design style preference according to the planning and design parameters of the to-be-planned area; step S7, inputting the image of the to-be-planned area into the style design model corresponding to the optimal design style preference, and obtaining an initial design drawing; step S8, performing multi-objective optimization on the initial design drawing, and obtaining an auxiliary design drawing.

[0008] The urban design generation auxiliary method provided by the present invention may also have the following features: wherein the urban design morphological parameters include building density, volume ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio.

[0009] The urban design generation auxiliary method provided by the present invention may also have the following features: wherein the classification results include unbuilt type and built type, when the construction proportion in the original image of the region is less than a preset value, the original image of the region is an unbuilt type, and when the construction proportion in the original image of the region is greater than a preset value, the original image of the region is a built type.

[0010] The urban design generation auxiliary method provided by the present invention may also have the following features: wherein, in step S5, for each design style preference, generating a corresponding style design model includes the following steps: step T1, constructing a generator and a discriminator as a generative adversarial network; step T2, training the generative adversarial network according to the training data set to obtain a trained generator; step T3, inputting the original images of each area in the training data set into the trained generator respectively, and obtaining corresponding predicted planning images respectively; step T4, for each predicted planning image, calculating the similarity between the predicted planning image and the corresponding regional planning image, and obtaining a similarity value; step T5, judging whether each similarity value is greater than or equal to a preset similarity value threshold, if so, using the trained generator as the style design model, if not, removing the original images of the area whose similarity values ​​are less than the similarity value threshold from the training data set, and executing step T2.

[0011] The urban design generation auxiliary method provided by the present invention may also have the following feature: in step S6, the planning design parameters are compared with the urban design form parameters and classification results corresponding to each design style preference for similarity, and the design style preference with the highest similarity is selected as the optimal design style preference.

[0012] The urban design generation auxiliary method provided by the present invention may also have the following features: in step S8, the multi-objective optimization is to adjust the building combination in the initial design drawing until the parameters of the initial design drawing meet the planning design parameters, and then an auxiliary design drawing is obtained.

[0013] The present invention also provides an urban design generation auxiliary device, which is used to generate an auxiliary design drawing with a specified designer style corresponding to the area to be planned, and has the following characteristics: a data extraction module, which is used to collect the regional planning image of the existing design scheme of the specified designer and the corresponding unplanned regional original image, and generate multiple original image-planning image pairs; a feature calculation module, which is used to calculate multiple urban design morphological parameters of the regional planning image for each original image-planning image pair, and classify the regional original image to obtain a classification result; a classification module, which is used to cluster the urban design morphological parameters corresponding to each original image-planning image pair and the classification result, and obtain multiple classes as design The module comprises a design style preference; a training set construction module, which is used to use all the corresponding original image-planning image pairs as the corresponding training data set for each design style preference; a model construction module, which is used to construct the corresponding initial model for each design style preference and train it according to the corresponding training data set to obtain the corresponding style design model; a style selection module, which is used to select the corresponding design style preference as the optimal design style preference according to the planning and design parameters of the area to be planned; an initial picture generation module, which is used to input the image of the area to be planned into the style design model corresponding to the optimal design style preference to obtain the initial design picture; an optimization module, which is used to perform multi-objective optimization on the initial design picture to obtain the auxiliary design picture.

[0014] Functions and Effects of the Invention

[0015] According to the urban design generation auxiliary method and device involved in the present invention, first, the design style preference of the works of existing designers is identified through the data extraction module, the feature calculation module and the classification module; secondly, the corresponding style design model is trained according to each design style preference through the training set construction module and the model construction module; then, the appropriate design style preference is selected according to the design requirements of the area to be planned through the style selection module and the initial map generation module, and the initial design map containing the design style preference is generated; finally, the initial design map is optimized and adjusted through the optimization module, so as to obtain an auxiliary design map that better meets the design requirements. Therefore, the urban design generation auxiliary method and device of the present invention can generate a regional planning design map with a specified designer's style preference, providing a reference for regional planning and design. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a block diagram of an urban design generation auxiliary device in an embodiment of the present invention;

[0017] Figure 2 is a schematic diagram of a process of generating a style design model corresponding to a design style preference in an embodiment of the present invention;

[0018] Figure 31 is a schematic diagram showing the comparison before and after multi-objective optimization in an embodiment of the present invention;

[0019] Figure 4 It is a flow chart of the city design generation auxiliary method in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the following embodiments and the accompanying drawings specifically illustrate the urban design generation auxiliary method and device of the present invention.

[0021] In this embodiment, an urban design generation auxiliary device is provided, which is used to generate an auxiliary design drawing with a specified designer style corresponding to the area to be planned.

[0022] Figure 1 4 is a block diagram of an urban design generation auxiliary device in an embodiment of the present invention.

[0023] like Figure 1 As shown, the urban design generation auxiliary device 100 includes a data extraction module 10, a feature calculation module 20, a classification module 30, a training set construction module 40, a model construction module 50, a style selection module 60, an initial image generation module 70, an optimization module 80 and a control module 90 for controlling the operation of the above modules.

[0024] The data extraction module 10 is used to collect the regional planning images of the existing design scheme of the designated designer and the corresponding original images of the unplanned area, and generate multiple original image-planning image pairs. 2 Up to 200km 2 The regional planning image is a planar image of a fixed-size part of the design plan, while the regional original image is a vector map obtained by surveying the site status when the corresponding regional planning image is not planned, and the image is processed by formatting. The fixed size is 800 meters by 800 meters.

[0025] The feature calculation module 20 is used to calculate multiple urban design morphological parameters of the regional planning image for each original image-planning image pair, and classify the regional original image to obtain a classification result.

[0026] Among them, the urban design morphological parameters include building density, floor area ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio.

[0027] Building density is used to quantify the degree of building coverage on a site, and its calculation expression is:

[0028]

[0029] The floor area ratio is used to quantify the intensity of construction in a site, and its calculation expression is:

[0030]

[0031] Among them, for a single building, the corresponding building area is the sum of the areas of all floors of the building.

[0032] The average height is used for the overall situation of building heights in the site, and its calculation expression is:

[0033]

[0034] Building height variance is used to quantify the richness of building height variations in a site, and its calculation expression is:

[0035]

[0036] The building configuration variance is used to quantify the richness of the shape changes of each building in the site, and its calculation expression is:

[0037]

[0038] The green space ratio is used to quantify the proportion of green space in the site, and its calculation expression is:

[0039]

[0040] The water surface ratio is used to quantify the proportion of water surface in the site, and its calculation expression is:

[0041]

[0042] The hard square ratio is used to quantify the proportion of hard squares in the venue, and its calculation expression is:

[0043]

[0044] In this embodiment, the urban design morphology parameters of the regional planning image include not only the building density, volume ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio of the regional planning image, but also the mean values ​​of the building density, volume ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio of the regional planning images corresponding to the plots around the regional planning image. The characteristics of the regional planning image are further supplemented and improved by the mean values ​​of the urban design morphology parameters of the surrounding regional planning images.

[0045] In this embodiment, the urban design morphological parameters are used to reflect the roads, water bodies, building combinations and morphological forms in the area, thereby parameterizing the design style of the planning image.

[0046] The classification results include unbuilt and built types, which are used to distinguish the initial construction status. When the construction ratio in the original image of the region is less than the preset value, the original image of the region is unbuilt. When the construction ratio in the original image of the region is greater than the preset value, the original image of the region is built.

[0047] The classification module 30 is used to cluster the corresponding urban design form parameters and classification results according to each original image-planning image, and obtain multiple classes as design style preferences.

[0048] In this embodiment, four classes are generated based on the existing works of a certain designer, namely, the greening class where the unbuilt area is planned as greening facilities, the industrial class where the unbuilt area is planned as an industrial park, the residential class where the built area is planned as a community, and the financial class where the unbuilt area is planned as a financial business district. According to different designer works, a variety of different classes can be generated through the feature calculation module 20 and the classification module 30, and then according to the specific urban design morphological parameters and classification results of these classes, the actual meaning of the design style preference corresponding to the class, that is, the design focus, is obtained.

[0049] The training set construction module 40 is used to use all corresponding original image-planning image pairs as corresponding training data sets for each design style preference.

[0050] The model building module 50 is used to build a corresponding initial model for each design style preference and train it according to the corresponding training data set to obtain a corresponding style design model.

[0051] Figure 2 It is a schematic diagram of a process of generating a style design model corresponding to a design style preference in an embodiment of the present invention.

[0052] like Figure 2 As shown, in the model building module 50, the process of generating a style design model corresponding to the design style preference includes the following steps:

[0053] Step T1, construct the generator and discriminator as a generative adversarial network.

[0054] Step T2: train the generative adversarial network according to the training data set to obtain a trained generator.

[0055] In step T2 of this embodiment, the original image in the original image-planning image pair is selected from the training data set and input into the generator to obtain the corresponding generated image, and then the generated image is input into the discriminator to obtain the discrimination result. Finally, the parameters of the generator and the discriminator are iteratively optimized according to the discrimination result to obtain a trained generator.

[0056] Step T3, input the original images of each region in the training data set into the trained generator respectively to obtain the corresponding prediction planning images.

[0057] Step T4: for each predicted planning image, calculate the similarity between the predicted planning image and the corresponding regional planning image to obtain a similarity value.

[0058] The calculation expression of the similarity value in this embodiment is:

[0059]

[0060] Wherein the prediction image index is the morphological index corresponding to the prediction planning image, and the target image index is the morphological index corresponding to the regional planning image. That is, the similarity between the prediction planning image and the corresponding regional planning image is quantified by the morphological index in the architectural field.

[0061] Step T5, determine whether each similarity value is greater than or equal to a preset similarity value threshold. If so, use the trained generator as a style design model. If not, remove the original image in the area with a similarity value less than the similarity value threshold from the training data set and execute step T2.

[0062] In this embodiment, the training effect is optimized by combining the last elimination method. Specifically, for the trained generator, the similarity values ​​corresponding to the original images of each region in the training data set are sorted from large to small. If the similarity value at the end of the sorting queue is less than the similarity value threshold, it means that the corresponding original image of the region in the training data set - the regional planning image is not significant for the feature, that is, the trained generator cannot accurately imitate the regional planning image of the region. Therefore, the original image-planning image pair is removed from the training data set, and the generator is trained using the updated training data set until the last similarity value is greater than the similarity value threshold, thereby improving the generator's ability to generate corresponding style design preferences for features in the regional original image.

[0063] The style selection module 60 is used to select a corresponding design style preference as an optimal design style preference according to the planning and design parameters of the area to be planned.

[0064] Among them, the planning and design parameters are compared with the urban design morphological parameters and classification results corresponding to each design style preference, and the design style preference with the highest similarity is selected as the optimal design style preference. This ensures that the design style is consistent with the training sample, and also ensures that the generated results are consistent in the eight morphological parameters of building density, volume ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio, thus ensuring the reliability of the generation.

[0065] The initial image generation module 70 is used to input the image of the area to be planned into the style design model corresponding to the optimal design style preference to obtain an initial design image.

[0066] The optimization module 80 is used to perform multi-objective optimization on the initial design drawing to obtain an auxiliary design drawing.

[0067] Among them, in step S8, the multi-objective optimization is to adjust the building combination in the initial design drawing until the parameters of the initial design drawing meet the planning design parameters, and then obtain the auxiliary design drawing.

[0068] In this embodiment, a genetic algorithm is used to achieve multi-objective optimization. The specific method is as follows: the parameters and basic positions of each building are used as locally adjustable parameters, and the activity level is determined by using the benchmark parameter values, i.e., multiple parameter target values, on the urban morphological parameters. Subsequently, the urban indicators are calculated on this basis and compared with the indicators of the benchmark sample. Through the above process, a combination of morphological parameters that is very close to the benchmark sample is identified under the constraints of limited optimization adjustments and various local building micro-adjustments.

[0069] Figure 3 3 is a comparative schematic diagram before and after multi-objective optimization in an embodiment of the present invention.

[0070] like Figure 3 As shown in the figure, (a) is the initial design drawing, and (b) is the auxiliary design drawing generated after multi-objective optimization, where the parameter target value conditions are: height diversity +10%, configuration diversity -10%, average building height <35, volume ratio >1.5, building density +30%. It can be seen that multi-objective optimization can generate a design drawing that meets the requirements better.

[0071] The control module 90 stores a control program for controlling the operation of each module.

[0072] The following describes the process of using the urban design generation assistance device 100 to perform an urban design generation assistance method in conjunction with the accompanying drawings.

[0073] Figure 4 It is a flow chart of the city design generation auxiliary method in an embodiment of the present invention.

[0074] like Figure 4 As shown, the urban design generation auxiliary method includes the following steps:

[0075] Step S1 , using the data extraction module 10 to collect the regional planning images of the existing design solutions of the designated designer and the corresponding original images of the unplanned regions, and generate a plurality of original image-planning image pairs.

[0076] Step S2, using the feature calculation module 20 to calculate multiple urban design morphological parameters of the regional planning image for each original image-planning image pair, and classifying the regional original image to obtain a classification result.

[0077] Step S3, using the classification module 30 to cluster the urban design form parameters and classification results corresponding to each original image-planning image pair, and obtain multiple classes as design style preferences.

[0078] Step S4, using the training set construction module 40 to construct all corresponding original image-planning image pairs for each design style preference as corresponding training data sets.

[0079] Step S5, using the model building module 50 to build a corresponding initial model for each design style preference and train it according to the corresponding training data set to obtain a corresponding style design model.

[0080] Step S6: using the style selection module 60 to select the corresponding design style preference as the optimal design style preference according to the planning and design parameters of the area to be planned.

[0081] Step S7, using the initial image generation module 70 to input the image of the area to be planned into the style design model corresponding to the optimal design style preference to obtain an initial design image.

[0082] Step S8, using the optimization module 80 to perform multi-objective optimization on the initial design drawing to obtain an auxiliary design drawing.

[0083] Functions and Effects of the Embodiments

[0084] According to the urban design generation auxiliary method and device involved in this embodiment, first, the design style preference of the works of existing designers is identified through the data extraction module, feature calculation module and classification module; secondly, the corresponding style design model is trained according to each design style preference through the training set construction module and the model construction module; then, the appropriate design style preference is selected according to the design requirements of the planned area through the style selection module and the initial map generation module, and the initial design map containing the design style preference is generated; finally, the initial design map is optimized and adjusted through the optimization module, so as to obtain an auxiliary design map that better meets the design requirements. In short, this method can generate a regional planning design map with a specified designer's style preference, providing a reference for regional planning and design.

[0085] Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An urban design generation auxiliary method for generating an auxiliary design drawing with a specified designer style corresponding to a planned area, characterized in that: The following steps are involved: Step S1, collecting the regional planning image of the existing design scheme of the designated designer and the corresponding original image of the unplanned region to generate a plurality of original image-planning image pairs; Step S2, for each original image-planning image pair, calculating a plurality of urban design morphological parameters of the regional planning image, and classifying the regional original image to obtain a classification result; Step S3, clustering the urban design form parameters and the classification results corresponding to each of the original image-planning image pairs to obtain a plurality of clusters as design style preferences; Step S4, for each of the design style preferences, using all the corresponding original image-planned image pairs as corresponding training data sets; Step S5, for each of the design style preferences, construct a corresponding initial model and perform training according to the corresponding training data set to obtain a corresponding style design model; Step S6, selecting the corresponding design style preference as the optimal design style preference according to the planning and design parameters of the area to be planned; Step S7, inputting the image of the area to be planned into the style design model corresponding to the optimal design style preference to obtain an initial design drawing; Step S8, performing multi-objective optimization on the initial design drawing to obtain the auxiliary design drawing.

2. The urban design generation auxiliary method according to claim 1, characterized in that: in, The urban design morphological parameters include building density, volume ratio, average height, building height variance, building configuration variance, green space ratio, water surface ratio and hard square ratio.

3. The urban design generation auxiliary method according to claim 1, characterized in that: in, The classification results include unconstructed and constructed types. When the construction ratio in the original image of the region is less than a preset value, the original image of the region is unconstructed. When the construction ratio in the original image of the region is greater than a preset value, the original image of the region is a constructed type.

4. The urban design generation auxiliary method according to claim 1, Features: Wherein, in the step S5, for each of the design style preferences, generating a corresponding style design model includes the following steps: Step T1, construct the generator and discriminator as a generative adversarial network; Step T2, training the generative adversarial network according to the training data set to obtain a trained generator; Step T3, inputting the original images of each region in the training data set into the trained generator respectively, and obtaining the corresponding prediction planning images respectively; Step T4, for each of the predicted planning images, calculating the similarity between the predicted planning image and the corresponding regional planning image to obtain a similarity value; Step T5, determine whether each of the similarity values ​​is greater than or equal to a preset similarity value threshold. If so, use the trained generator as the style design model. If not, remove the original image of the area where the similarity value is less than the similarity value threshold from the training data set and execute step T2.

5. The urban design generation auxiliary method according to claim 1, characterized in that: in, In the step S6, the planning and design parameters are compared with the urban design form parameters corresponding to each of the design style preferences and the classification result for similarity, and the design style preference with the highest similarity is selected as the optimal design style preference.

6. The urban design generation auxiliary method according to claim 1, characterized in that: in, In step S8, the multi-objective optimization is to adjust the building combination in the initial design drawing until the parameters of the initial design drawing meet the planning and design parameters, thereby obtaining the auxiliary design drawing.

7. An urban design generation auxiliary device, used to generate an auxiliary design drawing with a specified designer style corresponding to the area to be planned, characterized in that: include: A data extraction module, used to collect the regional planning images of the existing design solutions of the designated designer and the corresponding original images of the unplanned regions, and generate a plurality of original image-planning image pairs; A feature calculation module, used for calculating a plurality of urban design morphological parameters of the regional planning image for each original image-planning image pair, and classifying the regional original image to obtain a classification result; A classification module, used for clustering the urban design form parameters and the classification results corresponding to each of the original image-planning image pairs to obtain a plurality of classes as design style preferences; A training set construction module, for taking all the corresponding original image-planned image pairs as corresponding training data sets for each of the design style preferences; A model building module, used to build a corresponding initial model for each of the design style preferences and train it according to the corresponding training data set to obtain a corresponding style design model; A style selection module, used for selecting the corresponding design style preference as the optimal design style preference according to the planning and design parameters of the area to be planned; An initial image generation module, used for inputting the image of the area to be planned into the style design model corresponding to the optimal design style preference to obtain an initial design image; The optimization module is used to perform multi-objective optimization on the initial design drawing to obtain the auxiliary design drawing.

Images