Search Results (2)

Urban inland rivers are closely related to urban development, but high-density urbanisation has reduced the natural function of streams and the riverbanks are hardened into two parts, embankment walls and berms, which give rise to a variety of riparian landscapes. However, the difference in the height of riparian walkways affects the degree of their greening and landscape effects. In this paper, we studied single- and double-decker urban greenways, constructed quantitative indicators of spatial elements based on deep learning algorithms using an image semantic segmentation (ISS) model that simulates human visual perception, used random forests and multivariate linear regression models to study the impact of the height difference of the linear riverfront greenway on visual perception, clarified the impact of the visual landscape differences caused by different types of space on landscape aesthetic preferences (LP) and confirmed the impact of the specific extent to which landscape components influence preferences. The results of the study showed that there were significant differences in landscape perception scores between the single and double layers. (1) The influence of WED (negative correlation) and NI (positive correlation) is large in the single-layer greenway. The colour, material and structure of the guardrail can be beautified and diversified and the quality of the greenery can be taken into account to maintain the visibility of the greenery in order to improve the score of the single-layer greenway. (2) The significant influence of BVI in the double-layered greenway is positive. Water-friendly or water-viewing spaces can be added appropriately to improve the landscape score of double-layered greenways. This study is applicable to the regional landscape feature identification of single- and double-decker greenways on large-scale urban hard barge bank images, which realises the whole-region feature identification of a large-scale human perspective and is an effective expansion of analysis techniques for sustainable landscape planning and the design of riparian greenways. Full article

Image semantic segmentation (ISS) is used to segment an image into regions with differently labeled semantic category. Most of the existing ISS methods are based on fully supervised learning, which requires pixel-level labeling for training the model. As a result, it is often very time-consuming and labor-intensive, yet still subject to manual errors and subjective inconsistency. To tackle such difficulties, a weakly supervised ISS approach is proposed, in which the challenging problem of label inference from image-level to pixel-level will be particularly addressed, using image patches and conditional random fields (CRF). An improved simple linear iterative cluster (SLIC) algorithm is employed to extract superpixels. for image segmentation. Specifically, it generates various numbers of superpixels according to different images, which can be used to guide the process of image patch extraction based on the image-level labeled information. Based on the extracted image patches, the CRF model is constructed for inferring semantic class labels, which uses the potential energy function to map from the image-level to pixel-level image labels. Finally, patch based CRF (PBCRF) model is used to accomplish the weakly supervised ISS. Experiments conducted on two publicly available benchmark datasets, MSRC and PASCAL VOC 2012, have demonstrated that our proposed algorithm can yield very promising results compared to quite a few state-of-the-art ISS methods, including some deep learning-based models. Full article