Izvorni znanstveni članak

https://doi.org/10.17559/TV-20240717001853

Research on Landscape Space Design Optimization of Green Buildings based on Virtual Generation Algorithm

Zhi Ji ; School of Intelligent Architecture and Environmental Engineering, Chengdu Textile College, Chengdu Sichuan, China
Huajie Yang ; Faculty of Innovation and Design, City University of Macau, Macau SAR, 999078 China *
Jinhong Xian ; School of Intelligent Architecture and Environmental Engineering, Chengdu Textile College, Chengdu Sichuan, China
Yutong Xie ; Chengdu Textile College, Chengdu Sichuan, China
Yaou Zhang ; College of Landscape Architecture, Sichuan Agricultural University, Chengdu Sichuan, China; Faculty of Innovation and Design, City University of Macau, Macau SAR, 999078 China
Xiaobo Ma ; Department of Civil & Architectural Engineering & Mechanics, University of Arizona, 1209 E 2nd St, Tucson, Pima County 85719, USA

* Dopisni autor.

Sažetak

Traditional architectural landscape design often has problems such as poor visual effect, low satisfaction of residents, and unreasonable design in green environmental protection. Therefore, this paper proposes a green landscape space environment optimization design scheme based on virtual generation algorithm. Firstly, according to the virtual generation algorithm, the virtual generation of the building landscape size is optimized, and the optimization targets include the cost of the building and public facilities such as green belt. Secondly, a new algorithm for generating panoramic landscape images in virtual reality based on feature extraction is proposed. The determinant of Hessian matrix of each pixel in the landscape image is calculated to obtain SURF feature point values. The similarity between landscape images is measured by the Euclidean distance between feature points to achieve feature matching. Through orthographic projection, all landscape images to be synthesized are projected onto the cylindrical surface, and the overlapping parts of adjacent landscape images are fused. After obtaining the projected landscape images, the three-dimensional landscape images generated by orthographic projection are seamlessly spliced through feature matching, and the three-dimensional landscape images generated by orthographic projection are split from a certain position and projected on a plane to obtain a visual consistency of the panoramic landscape image. Finally, the experimental results show that there is almost no gap between the overlapping regions of the panoramic landscape images generated by the proposed algorithm. The brightness difference is not large, the "ghost" phenomenon can be eliminated, and the memory and time consumption are less.

Ključne riječi

feature matching; green building; landscape space; panoramic landscape; virtual generation algorithm; virtual reality

Hrčak ID:

325845

URI

https://hrcak.srce.hr/325845

Datum izdavanja:

31.12.2024.

Posjeta: 9 *