The Spatial Form of Digital Nonlinear Landscape Architecture Design Based on Computer Big Data

Before the emergence of landscape ecology, the status of landscape architecture belonged to architecture. Landscape design serves the construction and provides a place for leisure and entertainment in the crowded city. With the emergence of landscape ecology, the discipline of landscape architecture has gradually moved towards a self-independent development path. Landscape architecture in this period went hand in hand with architecture and urban planning, and each developed in its professional field [1]. This separation of disciplines destroys the diversity of places and humans to a certain extent. Due to the fragmentation of disciplines in the Chinese context, part of the landscape design is still attached to architecture, and the landscape is fragmented and homogeneous. Design pursues single issues such as form, function, style, and ecology within the discipline, and the limitations brought about by the separation of disciplines are becoming increasingly apparent.

Nonlinear science is regarded as “another revolution” in the history of science in the 20th century. It has almost spread to all fields of natural sciences, humanities, and social sciences and is changing people's traditional views of the real world. The scientific community believes that the research of nonlinear science has great scientific significance, and nonlinearity is the source of complexity and variety in nature [2]. The development of nonlinear science has also caused profound changes in philosophy and methodology. Our contribution to its exploration is creating a new subject field. Still, the more far-reaching significance is that it has brought about a significant change in the outlook on nature, the outlook on science, the methodology, and even thinking. Nonlinear interaction has a more intrinsic and essential meaning than linear interaction. Linear interaction is only a highly simplified and approximate treatment of nonlinear interaction. Information technology closely related to nonlinear science is developing rapidly, making today's society a transformation from a post-industry to an information society. Information technology is based on computer, communication and control technology. In recent years, information technology has developed database technology, artificial intelligence, expert system, remote sensing technology, geographic information system, global positioning system, computer-aided decision-making system, automatic control technology, monitoring technology, multimedia technology, computer network technology and other applications. And these technologies have widely influenced all aspects of people's lives.

With the continuous advancement of ecological landscape construction and improved living environment, people's quality standards for landscape construction are constantly improving. So we need to carry out multi-dimensional nonlinear landscape design. Combining the functional positioning of the landscape to optimise the structure of multi-dimensional nonlinear landscape design can improve its urban design. The multi-dimensional nonlinear landscape design method is of great significance in improving the rationality of architectural design and improving the living environment of human settlements [3]. The planning and design of landscape construction based on its development orientationand government policy-oriented factors can help us improve the spatial expression ability. The article uses computer vision with structured image analysis and parameter simulation methods to optimise multi-dimensional design. This paper proposes a multi-dimensional nonlinear landscape design method based on parameterised models to improve the quantitative analysis ability.

We combined the RGB feature decomposition method to perform parameterised segmentation of multi-dimensional nonlinear landscape models [9]. The thesis extracts the fuzzy degree matching feature quantity of the multi-dimensional nonlinear landscape image. The estimated value of the parameters of the salient feature points in the inner and outer areas of the target edge of the landscape image is: (10) NLM[g(i)]=∑j∈Ωω(i,j)g(j) NLM[g(i)] = \sum\limits_{j \in \Omega } \omega (i,j)g(j) Based on the idea of metric learning, the three-dimensional feature reconstruction parameters of multi-dimensional nonlinear landscape images are uniformly distributed [10]. That is, the pixel sequence satisfies n∈N(0,σn2) n \in N(0,\sigma _n^2) . The paper uses a parameterised model for structural analysis to find the second moment of multi-dimensional nonlinear landscape design: (11) μpq=∑m=1M∑n=1N(x−x.)p(y−y.)qf(x,y) {\mu _{pq}} = \sum\limits_{m = 1}^M \sum\limits_{n = 1}^N {(x - \mathop x\limits^. )^p}{(y - \mathop y\limits^. )^q}f(x,y) We use first-order moments m₀₁ and m₀₂ to denote the edge blur characteristics of multi-dimensional nonlinear landscape images, respectively [11]. We use the block fusion technology to obtain the edge template area of the landscape design that satisfies the normal distribution in the n_c × n_r sub-blocks to obtain the fine-grained feature point extraction results of the landscape image: (12) I1=n20n02−n112n004 {I_1} = {{{n_{20}}{n_{02}} - n_{11}^2} \over {n_{00}^4}} (13) Emij=∑k=0255emkij E_m^{ij} = \sum\limits_{k = 0}^{255} e_{mk}^{ij} (14) emkij={−pklogpkpk≠00,pk=0 e_{mk}^{ij} = \left\{ {\matrix{ { - {p_k}\log {p_k}} & {{p_k} \ne 0} \cr {0,} & {{p_k} = 0} \cr } } \right. p_k is the vectorised dimension of the spatial data of the landscape image; m = 1, 2,···, N. We extract parameterised models of multi-dimensional nonlinear landscape views [12]. The vector quantisation feature set of the multi-dimensional nonlinear landscape image obtained based on the parameterised model method to achieve the multi-dimensional nonlinear landscape design is: (15) pij(A)={ωijωi,i≠jeij∈A0i≠jeij∈A1−∑j:eij∈Aωijωi,i=j {p_{ij}}(A) = \left\{ {\matrix{ {{{{\omega _{ij}}} \over {{\omega _i}}},} & {i \ne j{e_{ij}} \in A} \cr 0 & {i \ne j{e_{ij}} \in A} \cr {1 - {{\sum\limits_{j:{e_{ij}} \in A} {\omega _{ij}}} \over {{\omega _i}}},} & {i = j} \cr } } \right. We use the block template matching method to perform the parameterized segmentation of the multi-dimensional: (16) x(k)=[x1(k),x2(k),⋯,xm(k)],i=1,2,⋯,m x(k) = [{x_1}(k),{x_2}(k), \cdots ,{x_m}(k)],i = 1,2, \cdots ,m k is the sampling node of multi-dimensional nonlinear landscape data; x_i(k) is the data stream. The realisation process of the whole model is shown in Figure 1.

Fig. 1

The experiment adopts a Matlab simulation design. The sample pixel scale of landscape images is 100~ 600. The pixel distribution of the landscape image is 300 × 300. The modulus of metric learning is 100. The number of iterations is 1200. The descriptive statistical analysis results of the correlation parameters are shown in Table 1.

According to the descriptive statistical analysis results in Table 1, multi-dimensional nonlinear landscape image sampling is performed. We use the virtual scene space visual planning method to simulate multi-dimensional nonlinear landscape design parameters to achieve landscape optimisation design [13]. The design effect results are shown in Figures 2–4. Analysing Figures 2–4, we know that our multi-dimensional nonlinear landscape design method has a better effect.

Fig. 2

Fig. 3

Fig. 4

The test regression analysis value and test value, and other parameter results are shown in Table 2. Analysing Table 2 shows that we adopt the method of this paper to carry out multi-dimensional nonlinear landscape design with better visual feature expression ability. The parameter distribution structured fitting accuracy is high.

This paper proposes a multi-dimensional nonlinear landscape design method based on a parameterised model to extract a multi-dimensional nonlinear landscape view model. We use parameteried model methods to achieve multi-dimensional lnonlinear landscape design. The article uses the computer vision image analysis method to carry out multi-dimensional nonlinear landscape design and constructs a parametric analysis model of multi-dimensional nonlinear landscape design. We combine fuzzy feature extraction methods to optimise the collection and feature recognition of multi-dimensional nonlinear landscape images. The analysis shows that we have a solid ability to express visual characteristics of multi-dimensional nonlinear landscape design using this method. Moreover, the landscape design effect obtained by this method is better.