A5003981917- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Image Processing and 3D Reconstruction
- Advanced Vision and Imaging
- Medical Image Segmentation Techniques
- 3D Surveying and Cultural Heritage
- Visual Attention and Saliency Detection
- Face Recognition and Perception
- Data Visualization and Analytics
- Remote Sensing and LiDAR Applications
- Medical Imaging and Analysis
- Robotics and Sensor-Based Localization
- Emotion and Mood Recognition
- Face and Expression Recognition
- Time Series Analysis and Forecasting
- Face recognition and analysis
- Video Analysis and Summarization
- Aortic aneurysm repair treatments
- Human Pose and Action Recognition
- Data Management and Algorithms
- Image and Video Quality Assessment
- Advanced X-ray and CT Imaging
- Satellite Image Processing and Photogrammetry
Swansea University
2015-2025
Cardiff University
2010-2014
Durham University
2006-2011
Zhejiang Normal University
2008-2011
City University of Hong Kong
2004
Three-dimensional surface registration transforms multiple three-dimensional data sets into the same coordinate system so as to align overlapping components of these sets. Recent surveys have covered different aspects either rigid or nonrigid registration, but seldom discuss them a whole. Our study serves two purposes: 1) To give comprehensive survey both types focusing on point clouds and meshes 2) provide better understanding from perspective fitting. Registration is closely related...
In this work, we present a study that traces the technical and cognitive processes in two visual analytics applications to common theoretic model of soft knowledge may be added into process for constructing decision-tree model. Both case studies involved development classification models based on "bag features" approach. compared approach using parallel coordinates with machine-learning information theory. found had some advantages over machine learning approach, especially when sparse...
Salient object detection identifies objects in an image that grab visual attention. Although contextual features are considered recent literature, they often fail real-world complex scenarios. We observe this is mainly due to two issues: First, most existing datasets consist of simple foregrounds and backgrounds hardly represent real-life Second, current methods only learn salient objects, which insufficient model high-level semantics for saliency reasoning scenes. To address these problems,...
3D representations of large-scale and urban scenes are crucial across various industries, including autonomous driving, planning, natural resource supervision many more. Large-scale industrial reconstructions inherently complex multifaceted. Many existing surveys primarily focus on academic progressions often neglect the intricate diverse needs industry. This survey aims to bridge this gap by providing a comprehensive analysis reconstruction methods, with requirements such as scalability...
Finding correspondences between two surfaces is a fundamental operation in various applications computer graphics and related fields. Candidate can be found by matching local signatures, but as they only consider geometry, many are globally inconsistent. We provide novel algorithm to prune set of candidate those most likely consistent. Our approach handle articulated surfaces, ones deformation which nonisometric, provided that the locally approximately isometric. uses an efficient diffusion...
3D models of humans are commonly used within computer graphics and vision, so the ability to distinguish between body shapes is an important shape retrieval problem. We extend our recent paper which provided a benchmark for testing non-rigid algorithms on human models. This far stricter challenge than previous benchmarks. have added 145 new use as separate training set, in order standardise data provide fairer comparison. also included experiments with FAUST dataset scans. All participants...
Psychology studies and behavioural observation show that humans shift their attention from one location to another when viewing an image of a complex scene. This is due the limited capacity human visual system in simultaneously processing multiple inputs. The sequential shifting on objects non-task oriented can be seen as form saliency ranking. Although there are methods proposed for predicting rank, they not able model this well, primarily based ranking values binary prediction. Following...
Non-rigid registration of deformed 3D shapes is a challenging and fundamental task in geometric processing, which aims to non-rigidly deform source shape into alignment with target shape. Current state-of-the-art methods assume deformations be near-isometric. This assumption does not reflect real-world conditions, for example large-scale deformation, where moderate anisotropic (e.g., stretches) are common. In this paper we propose two significant changes typical pipeline address such...
With the increasing popularity of 3D applications such as computer games, a lot geometry models are being created. To encourage sharing and reuse, techniques that support matching retrieval these emerging. However, only few them can handle deformable models, is, different poses, methods generally very slow. In this paper, we present novel method for efficient models. Our research idea stresses using both topological geometric features at same time. First, propose Topological Point Ring (TPR)...
Abstract Over the past decade, computer scientists and psychologists have made great efforts to collect analyze facial dynamics data that exhibit different expressions emotions. Such is commonly captured as videos are transformed into feature‐based time‐series prior any analysis. However, analytical tasks, such expression classification, been hindered by lack of understanding complex space associated algorithm space. Conventional graph‐based visualization also found inadequate support tasks....
Point cloud completion is the task of producing a complete 3D shape given an input partial point cloud. It has become vital process in computer graphics, vision and applications such as autonomous driving, robotics, augmented reality. These often rely on presence representation environment. Over past few years, many algorithms have been proposed substantial amount research carried out. However, there are not in-depth surveys that summarise progress way allows users to make informed choice...
Natural image statistics is an important area of research in cognitive sciences and computer vision. Visualization statistical results can help identify clusters anomalies as well analyze deviation, distribution, correlation. Furthermore, they provide visual abstractions symbolism for categorized data. In this paper, we begin our study visualization by considering representations power spectra, which are commonly used to visualize different categories images. We show that convey a limited...
Due to the popularity of computer games and animation, research on 3D articulated geometry model retrieval has attracted a lot attention in recent years. However, most existing works extract high-dimensional features represent models suffer from practical limitations. First, misalignment may produce unreliable euclidean distances affect accuracy. Second, curse dimensionality also degrades efficiency. In this paper, we propose an embedding framework improve practicability these methods. It is...
Abstract The human visual system has limited capacity in simultaneously processing multiple inputs. Consequently, humans rely on shifting their attention from one location to another. When viewing an image of complex scenes, psychology studies and behavioural observations show that prioritise sequentially shift among stimuli. In this paper, we propose predict the saliency rank objects by inferring shift. We first construct a new large-scale salient object ranking dataset, with defined order...
Matching and retrieval of motion sequences has become an important research area in recent years, due to the increasing availability popularity capture data. The main challenge matching two is diversity captured motions, including variable length, local shifting, global scaling. Most existing methods employ Dynamic Time Warping (DTW) or Uniform Scaling handle these problems. In this paper, we propose a novel content-based method for human We convert problem data into transportation problem....
Abstract Identifying multiple deformable parts on meshes and establishing dense correspondences between them are tasks of fundamental importance to computer graphics, with applications e.g. geometric edit propagation texture transfer. Much research has considered non‐rigid surfaces, but little work can both identify similar handle partial shape correspondences. This paper addresses two related problems, treating as a whole: (i) identifying mesh, by transformation given query part, (ii) point...
Fully automated 3D segmentation is not only challenging due to, for instance, ambiguities in appearance, but it also computationally demanding.We present a fullyautomatic, learning-based deformable modelling method segmenting the aortic root CT images using two-stage mesh deformation: non-iterative boundary with statistical shape model constraint, followed by an iterative refinement process.At both stages, we introduce B-spline regularisation technique to avoid entanglement during...