Mobile QR Code QR CODE

2024

Acceptance Ratio

21%


  1. (College of Art, Shangqiu University, Shangqiu 476000, China)



Garden landscape, Space environment, Image vision, Sparse Bayesian reconstruction, Multilevel feature decomposition

1. Introduction

With urbanization construction advancing and developing, various industries are facing the impact of information technology reform and innovation. More and more landscape designers are also trying to combine computer technology to design the spatial environment of landscape architecture. At present, most landscape designers in China have insufficient understanding and application of modern computer information technology when designing spatial environments. This leads to a lack of clear visual characteristics when conceptualizing design concepts and subjects, resulting in the inability to design a feasible construction plan [1,2]. Therefore, following the development of the times, utilizing computer vision information processing technology to improve data processing capabilities has become the main way for personnel from all walks of life to enhance their core competitiveness. With the progress of human society, the images to be processed have become increasingly complex and diverse. Various sectors of society have also raised higher requirements for the performance of visual information processing technology [3,4]. Image processing technology mainly covers areas such as object detection, image recognition, image reconstruction, and image segmentation. The 2D and 3D optical image reconstruction features of image reconstruction can obtain depth information in complex scenes. This feature makes image reconstruction technology often used in space exploration, environmental design, aerial photography and other fields [5]. Sparse Bayesian algorithm, as a unique algorithm model in reconstruction technology, can fully mine and utilize the prior information of images. Therefore, in order to improve the design effect of garden landscape spatial environment, a sparse Bayesian algorithm is studied to construct an optimized image visual reconstruction model based on garden landscape spatial environment design. And combine the gray pixel feature separation method and multi-level feature decomposition method in the model for image visual feature reconstruction. In this context, combining image visual reconstruction (IVR) methods to construct an image visual reconstruction model for the design of garden landscape spatial environment, to improve the design effect of garden landscape spatial environment.

2. Literature Background

With the rapid development of computer vision information processing technology, computer vision processing technology has the ability to extract features and process information from a large amount of image information data. This made experts realize computer image processing technology's application value in landscape space design. Many experts and scholars have conducted relevant research on the design of garden landscape spatial environment. Researchers such as Gaminiesfahani believe that the design of garden landscapes has a promoting effect on the rehabilitation of patients. They proposed using database search to analyze the relationship between landscape design and the impact on patient health and happiness. They conducted extensive evaluation and analysis of relevant data and papers. The results indicate that landscape design with architectural environmental features such as exposure to nature, reduced crowding, reduced noise, and natural light can increase patients' chances of recovery [6]. The research team of Uwajeh proposes innovative applications of medical technology combined with nature to optimize the clinical outcomes of Alzheimer's disease patients. And they studied the positive effects of therapeutic gardens in the medical environment on Alzheimer's disease through the Institute of Scientific Information Research. The research results indicate that natural environment and virtual reality technology have positive effects on the treatment of Alzheimer's disease [7]. Scholars such as Faraji believe that botanical gardens are important to scientific exploration and public landscape aesthetics. They proposed combining genetics to study the role of botanical gardens in biodiversity, climate change, and ecological assessment. The research results indicate that botanical gardens have scientific potential and application value in plant science research and agriculture [8]. Lotfi et al. proposed using vertical garden design as a tool for restoring urban space in order to increase the green space of the city. In the study, 20 experimental personnel were used to simulate the vertical garden in a specific environment, and a recovery scale was used to test and score the experimental personnel. The test results showed that the experimental personnel in a vertical garden environment had better recovery effects in heart rate and blood pressure [9]. Researchers such as Lamoreaux proposed an architectural philosophy of preventing crime through environmental design to design a warm, comfortable, safe, and reliable campus environment. The design covers technologies such as access control, natural surveillance and territorial reinforcement, aiming to use environmental design to prevent or reduce anti-social behavior. The results indicate that this design can improve students' safety perception and has certain reference value in the field of architectural philosophy [10]

IVR plays an important role in the design of garden landscape spatial environment. To improve the resolution of image extraction in different spatial environments, Zhang's research team proposed an unequal single image analysis method based on saliency driven gradient. This method utilizes the characteristics of single image super-resolution and multiple receiving domains to extract high-level and low-level features of the image. The results confirm that this method has achieved significant results in visual performance of image extraction [11]. To improve the denoising ability of image extraction, Liu and other scholars proposed an image denoising prior framework based on the denoising regularization algorithm. This method utilizes a convolutional neural network model to remove artifacts from images for training, and incorporates sampled image information into the measurement dataset for prior learning. The results confirm that this method has practical value in the field of image denoising [12]. To improve the scanning efficiency of images, Dar's research team proposes to construct an adversarial network model based on a deep learning framework. This model uses the shared high-frequency Apriori algorithm to retain the high spatial frequency details of the source contrast, and uses the undersampling contrast low-frequency Apriori algorithm to prevent the loss of feature information. The results confirm that this method improves scanning efficiency and has higher performance compared to the reconstruction synthesis method [13]. To improve the spatial resolution of images, researchers such as Greffier proposed a new deep learning image reconstruction algorithm. This algorithm utilizes the filtered backprojection method to measure changes in noise amplitude and noise texture, and uses AIDR3D to reconstruct the original data. The test results confirm that this method has a lower noise amplitude and can improve the spatial resolution of the image [14]. To improve single view image reconstruction accuracy, scholars such as Fang proposed a new single material 3D reconstruction method. This method adopts a self supervised approach to achieve end-to-end training, and utilizes rendering models to capture and generate new view images from dynamic scenes. The experimental results confirm the applicability of this method in the field of image reconstruction [15].

In summary, many researchers and scholars have recognized the importance of designing the spatial environment of garden landscapes for the psychological and physical health of urban residents. They have also proposed many designs for urban spatial environment. However, the design of garden landscape spatial environment combined with IVR model is still quite rare. Therefore, IVR for landscape spatial environment design is studied in order to achieve better results in landscape spatial environment design.

3. IVR Model for Landscape Space Environment Design

3.1. Model Reconstruction Based on Sparse Bayesian Algorithm

In the design of garden landscape spatial environment, images often have complex structures and diverse features, so traditional reconstruction methods cannot handle these complex situations well. Sparse Bayesian algorithm is a machine learning technique used for statistical inference, which has strong adaptability and high generalization ability, and is therefore widely used in image reconstruction tasks. Sparse representation theory is a multi-scale extended theoretical system that utilizes sparse constraints and control signals to achieve signal decomposition. It has advantages such as high efficiency and robustness in signal and data processing. Therefore, this theory has been widely applied in fields such as feature extraction, noise suppression, and sparse reconstruction [16]. The sparse Bayesian reconstruction algorithm in the IVR model can achieve the optimal sparse representation of image feature information. And based on this, a lower dimensional mathematical model can be established to solve practical application problems, thereby achieving full mining and utilization of image information [17]. The structure of the multi-scale fusion module is shown in Fig. 1.

Fig. 1. Structure diagram of multi-scale fusion module.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig1.png

According to Fig. 1, in sparse representation theory, all signals and data with sparse representation space have compressibility. That is, both data and signals have a certain degree of sparsity within a certain transformation domain [18]. Therefore, compress the decoder layers with scales of d0, d2, and d4 to their initial sizes. Then, the compressed feature maps to their initial size are used as inputs to the multi-scale fusion module through deconvolution layers. Finally, edge information from different depths was fused separately. Based on this theory, if the unknown signal is set to $x$ and the observation matrix is $\gamma $, then Eq. (1) is the mathematical expression of the signal observation equation.

(1)
$ y=\gamma x $

In Eq. (1), the value range of unknown signal $x$ is $x\in H^{N} $. The range of values for the observation matrix $\gamma $ is $\gamma \in H^{M\times N} $ where $N\ge M$. The range of values for the measurement value $y$ of the observation matrix is $y\in H^{M} $. $M$ and $N$ represent the columns and rows of the observation matrix, respectively. The signal observation equation can restore complete data information from incomplete data. However, in general, most raw data has insufficient sparsity. Therefore, it is not possible to directly use Eq. (1) for reconstruction processing. Therefore, before reconstruction, it is necessary to perform sparse decomposition on the original data through a transformation strategy. Eq. (2) is a mathematical expression for sparse decomposition.

(2)
$ x=\sum _{i=1}^{N}\varphi _{i} E. $

In Eq. (2), $\varphi _{i} $ means the sparse transformation matrix, i. e. the transformation domain. $E$ is sparsity, where the range of values for transformation domain and sparsity is $H^{N} $. Using the equivalent dictionary matrix $D=\varphi \gamma $, Eq. (3) is a mathematical expression for sparse reconstruction.

(3)
$ \left\{\begin{aligned} & y=\gamma x=\gamma \varphi E=DE, \\ & \arg \min \left\| E\right\| _{0} etDE=y. \end{aligned}\right. $

In Eq. (3), $D$ means the equivalent dictionary matrix. $\left\| \cdot \right\| _{0} $ is the norm of vector $l_{0} $, which represents the number of non zero elements in the vector. Taking into account factors such as noise and error during actual measurement, the expression for sparse reconstruction can also be shown by Eq. (4).

(4)
$ \arg \min \left\| E\right\| _{1} et\left\| y-DE\right\| _{2} \le q . $

In Eq. (4), $q$ represents the noise in actual measurement, and $\left\| E\right\| _{1} =\sum {}_{i} \left|E_{i} \right| $ is the sum of absolute values of signal elements. Considering practical applications, there is almost no situation without noise. Therefore, in the presence of noise, the basic linear model expression for sparse reconstruction is shown in Eq. (5).

(5)
$ y=D\sum _{i=1}^{N}E_{i} +q . $

In Eq. (5), $E_{i} $ means that the vector atom to be solved, i.e. the reconstructed signal source result. $y\in H^{M\times 1} $ is known observation information, while $q\in H^{M\times 1} $ means unknown observation noise, which is the noise vector in the actual environment. $D\in H^{M\times N} $ is an underdetermined equivalent dictionary matrix. Bayesian algorithm's core idea is to parameterize the prior information of unknown vectors through Gaussian distribution. Furthermore, Bayesian theory was used to update and alternate the implicit parameters of the posterior information of unknown vectors. Thus, the final posterior distribution estimation was obtained [19]. The observation image was set as $y$. If the mean $q$ of the ambient noise is 0 and obeys the Gaussian distribution, then Eq. (6) is the Gaussian likelihood function of the observed image $y$.

(6)
$ p(y\mid E;\varepsilon ^{2})=p(y\mid E;\theta )\sim N(\mu _{E} ,\theta )\nonumber\\ =\left(\frac{\theta }{2\pi } \right)^{\frac{M}{2} } \exp \left(-\frac{\theta }{2} (y-DE)^{T+1} \right). $

In Eq. (6), $p$ represents the observation point, $\theta $ represents the parameter of the Gaussian distribution, if the environmental noise is $q\sim N\left(0,\varepsilon ^{2} \right)$ and $\theta =\varepsilon ^{-2} $, and according to Bayesian theory, $E$ in the model follows a Gaussian distribution, then Eq. (7) is the prior distribution function.

(7)
$ p(E;\omega )=\prod _{n=1}^{N}N(E_{i} \mid 0,\omega )\nonumber\\ =(2\pi ) ^{-\frac{N}{2} } |\Omega |^{\frac{1}{2} } \exp \left(-\frac{1}{2} E^{T+1} \Omega ^{-1} \right) . $

In Eq. (7), $\Omega $ represents the implicit parameter controlling prior variance, with a value range of $\Omega \buildrel\Delta\over= diag(\omega _{1}$, $\omega _{2}$, ..., $\omega _{n})$. When $\omega _{i} $ is 0, the corresponding $E_{i} $ is also 0. Therefore, Fig. 2 shows a sparse Bayesian reconstruction model.

According to Fig. 2, sparse Bayesian reconstruction model's central idea is to parameterize unknown vectors' prior information by using Gaussian distribution. Therefore, this model can also be seen as a multi task linear regression model of sparse Bayesian [20]. Therefore, to cope with more complex design of garden landscape spatial environment, while considering more complex and diverse practical application scenarios, a multi-layer prior reconstruction model is constructed based on sparse Bayesian reconstruction model. Fig. 3 shows the IVR multi-layer prior model. This model essentially utilizes deeper hidden parameters to further express the reconstructed model's conjugate distribution. The accuracy parameter $\omega $ of IVR is controlled by the shape parameter $a$ and the scale parameter $b$, respectively. The accuracy parameter $\theta $ of noise model is controlled by shape parameter $g$ and scale parameter $f$, respectively, which control the noise.

Fig. 2. Sparse bayesian reconstruction model diagram.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig2.png

Fig. 3. Multilayer prior model graph for image visual reconstruction.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig3.png

3.2. Optimized IVR Model Based on Landscape Spatial Environment Design

To construct an IVR model for garden landscape spatial environment design, sparsity reconstruction can be performed on the image features of garden landscape spatial environment design. And feature matching method can be used to detect the features of garden landscape spatial environment design images. Atanassov extension method can be used to match the feature points of garden landscape spatial environment design. Fig. 4 shows the IVR model for garden landscape spatial environment design.

Fig. 4. Visual reconstruction model of landscape space environment design image in landscape architecture.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig4.png

According to Fig. 4, seed point $P$ means that landscape spatial environment design image's pixel set is $\left(i,j\right)$. And $P$ is also spatial environment's pixel center. The reconstruction model is composed of sharpened templates and blocks, where the grayscale value of the pixel center is represented by $G_{swk} $. The subbands number is set to $K$. Eq. (8) is the expression for the reconstructed model's gradient feature components.

(8)
$ \left\{\begin{aligned} & P_{ak} =\left(\frac{\sum _{j=1}^{b}G_{swk} \left(1,j\right) }{b} ,~\frac{\sum _{j=1}^{b}G_{swk} \left(2,j\right) }{b} ,~...,~\right.\\ & \left.\frac{\sum _{j=1}^{b}G_{swk} \left(i,j\right) }{b} ,~...,~\frac{\sum _{j=1}^{b}G_{swk} \left(a,j\right) }{b} \right), \\ & P_{bk} =\left(\frac{\sum _{i=1}^{a}G_{swk} \left(i,1\right) }{a} ,~\frac{\sum _{i=1}^{a}G_{swk} \left(i,2\right) }{a} ,~...,\right.\\ &\left.\frac{\sum _{i=1}^{a}G_{swk} \left(i,j\right) }{a} ,~...,~\frac{\sum _{i=1}^{a}G_{swk} \left(i,b\right) }{a} \right). \end{aligned}\right. $

In Eq. (8), $a$ is a motion blur feature vector. $b$ represents the number of columns in the vector quantization matrix. This image reconstruction model combines fuzzy pixel region feature fusion reconstruction method. From this, the pixel set of spatial environmental artistic features distribution in garden landscapes was obtained. This has improved garden landscapes' spatial environment design ability, as well as the three-dimensional perception ability and image information reconstruction ability of spatial environment design images. When constructing IVR models for landscape spatial environment design, visual feature space distributed detection methods, image gray pixel feature separation methods, and visual image multi-level feature decomposition methods are commonly used [21]. An image visual feature reconstruction model for garden landscape spatial environment design was constructed, where Eq. (9) is the calculation of the visual feature distribution of the image.

(9)
$ Z\left(\vec{x}\right)=\sum {}_{j=1}^{p} Z_{j} \left(\vec{x}\right) . $

In Eq. (9), $Z\left(\vec{x}\right)$ represents the visual feature distribution of the image, $p$ represents the pixel center of the seed point, $j$ represents the points in the pixel set, and $Z_{j} \left(\vec{x}\right)$ represents the visual feature distribution of point $j$ in the pixel set. The adaptive fusion method is used to reconstruct the edge visual model of the garden landscape design, thereby obtaining the fuzzy closeness function of the spatial environment image in Eq. (10).

(10)
$ fitness\left(\vec{x}\right)=f\left(\vec{x}\right)+\left(Ct\right)^{\alpha } \sum {}_{j=1}^{p} Z_{j}^{\beta } \left(\vec{x}\right) . $

In Eq. (10), $fitness\left(\vec{x}\right)$ represents the spatial environment image blur closeness function, while $\alpha $ and $\beta $ represent the similarity and difference degrees of blur closeness, respectively. From Eq. (10), it can be set that the coordinate of garden landscape spatial environment design point $P_{N} $ is $\left(X_{PN} ,Y_{PN} \right)$. The coordinate of spatial environment design edge point $L$ is $\left(x_{k} ,y_{k} \right)$. Eq. (11) represents the relationship between the coordinates of the design point $P_{N} $ and the edge point $L$.

(11)
$ \left\{\begin{aligned} & X_{PN} >x_{k} ,&&i_{L} =i_{L} -1, \\ & X_{PN} =x_{k} ,&&i_{L} =i_{L} +0, \\ & X_{PN} <x_{k} ,&&i_{L} =i_{L} +1. \end{aligned}\right. $

The relationship between spatial design points in Eq. (11) can determine the grayscale pixel level $f$ of landscape spatial environment design. Eq. (12) is the visual feature reconstruction model for the spatial environment of garden landscapes.

(12)
$ W\left(a,b\right)=\frac{K\cdot e^{2i\pi k\ln a} }{\sqrt{a} } \nonumber\\ \quad \times \Bigg\{2\pi j\left(b-b_{a} \right)[Ei(2\pi jf_{\max } (b-b_{a} ))]\nonumber\\ \quad +\Bigg[\frac{ae^{\frac{2\pi jf_{\min } }{a} \left(b-b_{a} \right)} }{f_{\min } } -\frac{e^{\frac{2\pi jf_{\max } }{a} \left(b-b_{a} \right)} }{f_{\max } } \Bigg]\nonumber\\ \quad -Ei\bigg(\frac{2\pi jf_{\min } }{a} (b-b_{a} )\bigg)\Bigg\}. $

In Eq. (12), $b_{a} =\left(\frac{1}{af_{\max } } -\frac{T}{2} \right)\left(1-a\right)$, where $Ei\left(\cdot \right)$ is visual information feature recombination output in garden landscape spatial environment design. The recombination result obtained by combining model recognition can be used for garden landscape spatial environment design. To improve environmental design capabilities, visual optimization of spatial environment in garden landscape design can be carried out on this model. Fig. 5 shows the optimized IVR model for garden landscape spatial environment design.

Fig. 5. Optimized image visual reconstruction model diagram.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig5.png

According to Fig. 5, after extracting the edge point $L$ of the spatial model, the local variance of the visual image of the spatial environment was set to $\zeta _{1}^{2} $. The optimal image coefficient for landscape spatial environment design is $\zeta _{\eta }^{2} $. The value range of parameter $\beta $ is $\max \left[\frac{\zeta _{1}^{2} -\zeta _{\eta }^{2} }{\zeta _{1}^{2} } ,~0\right]$. In IVR, the gradient descent method is used for block wise visual reconstruction during visual region segmentation, so that the designed image sparsity feature values meet $C\in Z$. The optimal visual reconstruction threshold for designed image $f_{n} \left(x,y\right)$ at frame $n$ can be obtained. At this point, garden landscape spatial environment design's image matching coefficient is shown in Eq. (13).

(13)
$ g_{i}^{*} =\left\{\begin{aligned} & Qz_{j} ,&&z\le j\le x-y, \\ & g_{i} ,&&\text{other}. \end{aligned}\right. $

In Eq. (13), $Q$ means a standardized constant. Distributed detection of the spatial environment design of gardens was carried out by using block based template matching method. At the same time, contour point matching method was used to extract edge features in gardens spatial environment design. The maximum visual grayscale value of the obtained garden landscape spatial environment design image is shown in Eq. (14).

(14)
$ N_{pb} =\frac{\mu _{pb} }{\left(\mu _{00} \right)^{\tau } } . $

In Eq. (14), the super-resolution reconstruction method and sparse representation method are used to visually reconstruct the image of the garden landscape spatial environment design. And the interactive genetic algorithm is combined to achieve landscape information fusion perception in the garden landscape spatial environment design. The visual information reconstruction model of the environmental design image obtained at this time is shown in Eq. (15).

(15)
$ g\left(x,y\right)=p\left(x,y\right)+f\left(x,y\right) . $

In Eq. (15), $p\left(x,y\right)$ refers to IVR model's grayscale image, $g\left(x,y\right)$ is reconstructed vision image, and $f\left(x,y\right)$ represents garden landscape space's initial environmental image.

4. Simulation and Verification of IVR Model for Landscape Space Environment Design

4.1. Optimization of Sparse Bayesian Algorithm for Reconstructing Model Performance Validation

To verify IVR model performance for landscape spatial environment design, experimental simulations were conducted to compare the optimized sparse Bayesian algorithm (OSBA) reconstruction model performance. The shape and scale parameters of $\omega $ in sparse Bayesian reconstruction model are set to $a=b=10^{-6} $. The shape and scale parameters of $\theta $ in noise model are set to $g=f=10^{-6} $. The maximum cyclic experiments number shall not exceed 50, and SBA model shall be compared with other four classic algorithms for experimental analysis. The classic algorithms are SBL-EM algorithm, IFSBL algorithm, Lasso FSBL algorithm, and GGAMP-SBL algorithm. The average running time, average reconstruction error value, average signal-to-noise ratio (PSNR), and average structural similarity value (SSIM) of reconstructed images using different algorithms were compared and verified. Table 1 shows the average data values of reconstructed images using different algorithms.

Table 1. Average data values of reconstructed images using different algorithms.

Algorithm/Average value

Error

Time/s

PSNR/dB

SSIM

SBL-EM

0.104

979.4

23.154

0.514

IFSBL

0.164

345.1

19.124

0.549

Lasso-FSBL

0.198

322.4

18.647

0.557

GGAMP-SBL

0.087

1465.9

25.674

0.679

OSBA

0.080

264.5

25.941

0.715

From Table 1, for average error, OSBA algorithm's minimum error value is 0.080, which is 59.6% less than Lasso FSBL algorithm's maximum error value of 0.198. From average running time, OSBA algorithm's shortest running time is 264.5 s, which is 81.96% less than GGAMP-SBL algorithm's longest running time 1465.9 s. PSNR and SSIM are reconstruction quality indicators, and maximum average PSNR and SSIM in OSBA are 25.941 dB and 0.715, respectively. In terms of average signal-to-noise ratio (PSNR), OSBA has improved by 12.04%, 34.25%, 39.12%, and 1.04% compared to SBL-EM, IFSBL, Lasso FSBL, and GGAMP-SBL, respectively. In terms of average structural similarity value SSIM, OSBA has increased by 39.11%, 30.24%, 2.37%, and 5.3% compared to the SBL-EM, IFSBL, Lasso FSBL, and GGAMP-SBL, respectively. Overall, optimizing SBA has smaller reconstruction errors, shorter runtime, and higher reconstruction quality, resulting in better performance. To further validate algorithm performance, the time and power consumption of various algorithms in image reconstruction were compared and analyzed. Fig. 6 shows the time and power consumption of image reconstruction using different algorithms.

Fig. 6. Time and power consumption diagrams for image reconstruction using different algorithms.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig6.png

From Fig. 6, OSBA algorithm's image reconstruction time and power loss are significantly lower than the other four algorithms. OSBA algorithm takes 62 seconds. Compared with the longest time-consuming GGAMP-SBL, its timeliness has been improved by nearly 8 times. In terms of power loss, the lowest power consumption of the OSBA algorithm is 18w, which is only 10.11% of GGAMP-SBL algorithm's maximum power consumption. In terms of efficiency in image reconstruction, optimizing SBA has higher performance.

4.2. Comparative Analysis of Sparse Bayesian Visual Image Reconstruction Methods

To verify the optimized SBAIVR model effectiveness, it is necessary to test sparse Bayesian visual image reconstruction method. The sparse Bayesian visual image reconstruction method is Method 1, the convolutional neural based image reconstruction method is represented by Method 2, and the non local variation based image reconstruction method is Method 3. Fig. 7 shows the peak signal-to-noise and reconstruction time ratios of different reconstruction methods.

Fig. 7. The peak signal-to-noise and reconstruction time of different reconstruction methods.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig7.png

From Fig. 7(a), three reconstruction methods' signal-to-noise ratio gradually increases when sampling rate increasing. Method 1 can achieve good image reconstruction results in a short period of time when the sampling rate is below 0.2. The reconstruction effect of Method 2 and Method 3 also improves with the increase of sampling rate. However, based on Fig. 7(b), Method 2 and Method 3 take longer to run. Overall, method 1 has a good image reconstruction effect and the shortest running time. This is because Method 1 combines denoising models to denoise the graphics and eliminate the interference of noise. Therefore, it improves the peak signal-to-noise ratio while shortening the reconstruction time. To further validate the overall effectiveness of SBAIVR model, Fig. 8 compares and analyzes the image resolution of three reconstruction methods.

Fig. 8. Image resolution results of three reconstruction methods.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig8.png

From Fig. 8, the lowest resolution of Method 1 is 408, which is significantly higher than other two reconstruction methods. This is because SBAIVR model preserves image edge details during image reconstruction. And a larger scale was used to distinguish between noise in the image and the original signal, thereby improving the denoising effect and resolution of the image. In order to analyze the accuracy of the three methods in more detail, Table 2 compares the mean, maximum, and standard deviation of errors for different reconstruction methods.

Table 2. Mean, maximum, and standard deviation of errors for different reconstruction methods.

Reconstruction method

MEAN

Maximum value

Standard deviation

Method 1

0.0505

0.123

0.0261

Method 2

0.0716

0.221

0.0434

Method 3

0.0863

0.264

0.0397

From Table 2, Method 1's lowest error is 0.0505 pixels. Its maximum value and standard deviation are 0.123 and 0.0261 pixels, respectively, achieving ideal accuracy. Method 1's error values were reduced by 0.0211 and 0.0358 pixels compared to Method 2 and 3, respectively. Method 1's maximum values were reduced by 0.098 and 0.141 pixels, and its standard deviation was reduced by 0.0173 and 0.0136 pixels, respectively. This indicates that Method 1 has high stability and robustness. Due to experiment randomness, to verify accurate experimental data, 50 experiments' average value are used as comparative data, which are conducted by using three reconstruction methods. Fig. 9 shows different structural similarities for different reconstruction methods.

Fig. 9. Comparison of different structural similarities of different reconstruction methods.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig9.png

From Fig. 9, the average value of Method 1 did not change significantly in 50 experiments, and the structural similarity was significantly better than the other two reconstruction methods. From Fig. 9, the average structural similarity of 50 experimental methods 1 is about 3.54, while the average values of methods 2 and 3 are about 2.29 and 1.88. The structural similarity of method 1 has increased by 54.6% and 88.3% compared to methods 2 and 3, respectively. This indicates that the optimized SBAIVR model has superior performance and higher accuracy in visual images.

4.3. Sparse Analysis of Reconstruction Coefficients in Sparse Bayesian Algorithm Image Visual Reconstruction Model

The sparsity of the reconstructed model was verified by randomly selecting coefficient vectors from 200 image sub blocks from the SBAIVR model. Fig. 10 shows the sparsity analysis of reconstruction coefficients.

Fig. 10. Reconstruction coefficient sparsity analysis graph.

../../Resources/ieie/IEIESPC.2025.14.4.483/fig10.png

From Fig. 10, SBAIVR model's minimum sparsity value is 0.90, and the sparsity of the reconstructed model is greater than 0.9. The sparsity approaching 1 indicates that the feature parameters of the model meet relevant requirements, and the sparse reconstruction effect is relatively stable.

5. Conclusion

To improve the design effect of garden landscape spatial environment, an optimized IVR model based on garden landscape spatial environment design was constructed by combining SBA's IVR method. And simulation verification was conducted on the reconstructed model. The results show that the optimized sparse Bayesian OSBA algorithm's minimum error value is 0.080, its minimum running time is 264.5 s, and its maximum average PSNR and SSIM values are 25.941 dB and 0.715, respectively. OSBA algorithm's time consumption is 62 seconds, which is nearly 8 times higher than GGAMP-SBL algorithm's timeliness. OSBA algorithm's minimum power consumption is 18 w, which is only 10.11% of GGAMP-SBL algorithm's power consumption. When sparse Bayesian visual image reconstruction method's sampling rate is below 0.2, good image reconstruction results can be obtained in a relatively short period. Sparse Bayesian visual image reconstruction method's minimum resolution is 408, its minimum error value is 0.0505 pixels, its maximum value and standard deviation are 0.123 and 0.0261 pixels, respectively. Sparse Bayesian visual image reconstruction method's average structural similarity in 50 experiments is about 3.54. This method improves by 54.6% and 88.3% respectively compared to convolutional neural based image reconstruction methods and non local variation based image reconstruction methods. Sparse Bayesian visual image reconstruction method's minimum sparsity value is 0.90. In summary, IVR method for garden landscape spatial environment design has a good design effect. However, the experimental results are not comprehensive enough due to the limited data used, and further improvement is needed.

REFERENCES

1 
A. Kuznetsova, H. Rom, N. Alldrin, J. Uijlings, I. Krasin, J. Pont-Tuset, and V. Ferrari, ``The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale,'' International Journal of Computer Vision, vol. 128, no. 7, pp. 1956-1981, 2020.DOI
2 
G. Gallego, T. Delbrück, G. Orchard, C. Bartolozzi, B. Taba, A. Censi, and D. Scaramuzza, ``Event-based vision: A survey,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 1, pp. 154-180, 2020.DOI
3 
L. Mennel, J. Symonowicz, S. Wachter, D.-K. Polyushkin, A.-J. Molina-Mendoza, and T. Mueller, ``Ultrafast machine vision with 2D material neural network image sensors,'' Nature, vol. 579, no. 7797, pp. 62-66, 2020.DOI
4 
H. Pei, B. Yang, J. Liu, and K.-C.-C. Chang, ``Active surveillance via group sparse Bayesian learning,'' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 3, pp. 1133-1148, 2020.DOI
5 
L. Zhao, W.-J. Gao, and W. Guo, ``Sparse Bayesian learning of delay-doppler channel for OTFS system,'' IEEE Communications Letters, vol. 24, no. 12, pp. 2766-2769, 2020.DOI
6 
H. Gaminiesfahani, M. Lozanovska, and R. Tucker, ``A scoping review of the impact on children of the built environment design characteristics of healing spaces,'' HERD: Health Environments Research and Design Journal, vol. 13, no. 4, pp. 98-114, 2020.DOI
7 
P.-C. Uwajeh, T.-O. Iyendo, and M. Polay, ``Therapeutic gardens as a design approach for optimising the healing environment of patients with Alzheimer's disease and other dementias: A narrative review,'' Explore, vol. 15, no. 5, pp. 352-362, 2019.DOI
8 
L. Faraji, and M. Karimi, ``Botanical Gardens as Valuable Resources in Plant Sciences,'' Biodiversity and Conservation, vol. 31, no. 12. pp. 2905-2926, 2022.DOI
9 
Y.-A. Lotfi, M. Refaat, M. El Attar, and A.-A.-L. Salam, ``Vertical gardens as a restorative tool in urban spaces of new cairo,'' Ain Shams Engineering Journal, vol. 11, no. 3, pp. 839-848, 2020.DOI
10 
D. Lamoreaux, and M.-L. Sulkowski, ``An alternative to fortified schools: Using crime prevention through environmental design (CPTED) to balance student safety and psychological well‐being,'' Psychology in the Schools, vol. 57, no. 1, pp. 152-165, 2020.DOI
11 
L. Zhang, D. Chen, J. Ma, and J. Zhang, ``Remote-sensing image superresolution based on visual saliency analysis and unequal reconstruction networks,'' IEEE Transactions on Geoscience and Remote Sensing, vol. 58, no. 6, pp. 4099-4115, 2020.DOI
12 
J. Liu, Y. Sun, C. Eldeniz, W. Gan, H. An, and U.-S. Kamilov, ``RARE: Image reconstruction using deep priors learned without groundtruth,'' IEEE Journal of Selected Topics in Signal Processing, vol. 14, no. 6, pp. 1088-1099, 2020.DOI
13 
S-U-H. Dar, M. Yurt, M. Shahdloo, M-E. Ildız, B. Tınaz, and T. ?ukur, ``Prior-guided Image Reconstruction for Accelerated Multi-contrast MRI Via Generative Adversarial Networks,'' IEEE Journal of Selected Topics in Signal Processing, vol. 14, no. 6, pp. 1072-1087, 2020.DOI
14 
J. Greffier, D. Dabli, J. Frandon, A. Hamard, A. Belaouni, P. Akessoul, and J.-P. Beregi, ``Comparison of two versions of a deep learning image reconstruction algorithm on CT image quality and dose reduction: A phantom study,'' Medical Physics, vol. 48, no. 10, pp. 5743-5755, 2021.DOI
15 
B. Fang and N. Xiao, ``Self-supervised reflectance-guided 3d shape reconstruction from single-view images,'' Applied Intelligence, vol. 53, no. 6, pp. 6966-6977, 2023.DOI
16 
P.-A. Ejegwa and J.-M. Agbetayo, ``Similarity-distance decision-making technique and its applications via intuitionistic fuzzy pairs,'' Journal of Computational and Cognitive Engineering, vol. 2, no. 1, pp. 68-74, 2023.DOI
17 
Y. Yang and X. Song, ``Research on face intelligent perception technology integrating deep learning under different illumination intensities,'' Journal of Computational and Cognitive Engineering, vol. 1, no. 1, pp. 32-36, 2022.DOI
18 
S. Liu, H. Wu, Y. Huang, Y. Yang, and J. Jia, ``Accelerated structure-aware sparse Bayesian learning for three-dimensional electrical impedance tomography,'' IEEE Transactions on Industrial Informatics, vol. 15, no. 9, pp. 5033-5041, 2019.DOI
19 
J. Glaubitz, A. Gelb, and G. Song, ``Generalized sparse Bayesian learning and application to image reconstruction,'' SIAM/ASA Journal on Uncertainty Quantification, vol. 11, no. 1, pp. 262-284, 2023.DOI
20 
Z. Zhang, Y. Li, C. Huang, Q. Guo, C. Yuen, and Y-L. Guan, ``DNN-aided block sparse Bayesian learning afor user activity detection and channel estimation in grant-free non-orthogonal random access,'' IEEE Transactions on Vehicular Technology, vol. 68, no. 12, pp. 12000-12012, 2019.DOI
21 
S. Nimrah and S. Saifullah, ``Context-free word importance scores for attacking nmural networks,'' Journal of Computational and Cognitive Engineering, vol. 1, no. 4, pp. 187-192, 2022.DOI

Author

Yuan Ren
../../Resources/ieie/IEIESPC.2025.14.4.483/au1.png

Yuan Ren born in August 1982, male, from Nanyang, Henan Province, Han ethnicity, a member of the Communist Party of China, holds a master's degree and is an associate professor. He graduated with a Bachelor's degree in Art and Design from the School of Fine Arts of Henan Normal University in 2006, and a Master's degree in Fine Arts from the School of Fine Arts of Henan Normal University in 2012. Research direction: Environmental Art Design.

Work experience: From July 2006 to July 2013, he worked as a teacher (teaching assistant) at Shangqiu University; From August 2013 to November 2018, he worked as a teacher (lecturer) at Shangqiu University; From December 2018 to present, he worked as a teacher (associate professor) at Shangqiu University.

Academic situation: Published more than ten scientific research and teaching papers, led and participated in more than ten social science planning projects of the Henan Provincial Department of Education, Henan Provincial Federation of Social Sciences, and Shangqiu City. Among them, one won the first prize and four won the second prize. Published 1 monograph and edited 5 textbooks. Participated in one invention patent and two utility model patents.