ABSTRACT Redirected Walking (RDW) effectively extends the navigable area of virtual environments but frequently induces cybersickness due to vestibular‐visual conflicts. To mitigate this, this study proposes a hierarchical visual guidance framework. We first introduce a Steering Visual Guidance (SVG) model, which employs a grid‐based pattern to provide a stable visual reference frame. While effective, our evaluation revealed that the static, full‐field nature of SVG can occlude the user's view and reduce visual clarity. To address this limitation, we optimized the model into Dynamic Steering Visual Guidance (DSVG). Grounded in the physiological principles of central and peripheral vision, DSVG dynamically renders stability cues exclusively within the foveal region, fading them in the periphery based on retinal sensitivity. A user study demonstrates that both SVG and DSVG significantly reduce subjective discomfort and physiological markers of sickness compared to a control condition without cues. Crucially, DSVG achieves sickness mitigation comparable to SVG while reducing visual occlusion by approximately 70%. These findings suggest that DSVG offers a robust, occlusion‐minimizing solution for deploying RDW in constrained physical spaces.

With the continuous advancement of medical imaging technology, sparse-view X-ray 3D reconstruction has found widespread applications in low-dose imaging and rapid scanning. However, traditional reconstruction methods have limitations in capturing details and expressing complex structures. To address this issue, we propose a sparse-view X-ray 3D reconstruction method based on a hierarchical hybrid Transformer. This research uses NeRF as the core framework, combining sparse-view X-ray images. First, a module combining focal attention and multi-scale mixing is employed to partition focal regions, extract multi-scale features, and perform channel attention-based fusion, effectively integrating key regions and hierarchical structural information in the image. This enables the collaborative representation of fine details and global semantics. To address complex anatomical structures such as bones in chest, jaw and foot, as well as soft tissues like aneurism and pancreas, a hierarchical hybrid Transformer module is introduced to model dependencies between local and global regions, enhancing the ability to capture fine details and represent overall structure. Experimental results show that the proposed method achieves significant performance improvements in sparse-view X-ray 3D reconstruction tasks across various medical structures.

Salient object detection in optical remote sensing images (RSI-SOD) is critical in remote sensing, yet it faces challenges such as dependency on intensive pixel-level annotations and limited research on low-cost, weakly supervised methods. These challenges are compounded by difficulties in handling complex backgrounds and varying salient object features with existing CNN-based methods. We introduce the Global-Local Semantic Interaction Network (GLSIN), a high-performance, cost-effective RSI-SOD approach based on scribble supervision. GLSIN employs an encoder-decoder framework, blending a Transformer and CNN to create a Dual Branch Encoder that effectively captures both global and local features of images. The Global-Local Affinity Block (GLAB) and Feature Shrinkage Decoder with the Global-Local Fusion Block (GLFB) are integrated to enhance feature interaction and precision in saliency map generation. Experimental results on two public datasets show that our method achievesFmaxβ,Emaxξ,Sα, andMscores of 86.6%, 96.5%, 91.8%, and 0.7% on the EORSSD dataset, and 90.1%, 97.2%, 91.7%, and 1.1% on the ORSSD dataset, respectively. The performance surpasses existing weakly-supervised or unsupervised SOD methods and even some fully-supervised models.

In real-world scenarios, due to factors like sensor noise, point cloud data often exhibits low overlap, posing challenges for traditional registration methods. To address this issue, we propose a low-overlap point cloud registration with Transformer. This algorithm employs a dynamic positional encoding strategy that adaptively computes position encodings for each point based on its distribution. This enables better capturing of richer spatial relationships between point clouds and facilitates adaptation to diverse point cloud distributions across various scenes. Furthermore, we combine the mechanisms of self-attention and graph convolutions. The self-attention mechanism captures global dependencies among points, while the graph convolutions capture local neighborhood information between points. Lastly, in the context of cross-attention, adaptive weights are introduced during the attention calculation process. This involves multiplying attention scores by adaptive weights, enhancing the model's ability to focus on crucial registration areas. In scenarios with low overlap, this algorithm significantly enhances the success rate of successful registrations. It achieves notable improvements and attains a new state-of-the-art performance in the 3DLoMatch benchmark test, reaching a registration recall rate of 71.7%.

Point cloud registration is a critical issue in 3D reconstruction and computer vision, particularly challenging in cases of low overlap and different datasets, where algorithm generalization and robustness are pressing challenges. In this paper, we propose a point cloud registration algorithm called Neighborhood Multi-compound Transformer (NMCT). To capture local information, we introduce Neighborhood Position Encoding for the first time. By employing a nearest neighbor approach to select spatial points, this encoding enhances the algorithm’s ability to extract relevant local feature information and local coordinate information from dispersed points within the point cloud. Furthermore, NMCT utilizes the Multi-compound Transformer as the interaction module for point cloud information. In this module, the Spatial Transformer phase engages in local-global fusion learning based on Neighborhood Position Encoding, facilitating the extraction of internal features within the point cloud. The Temporal Transformer phase, based on Neighborhood Position Encoding, performs local position-local feature interaction, achieving local and global interaction between two point cloud. The combination of these two phases enables NMCT to better address the complexity and diversity of point cloud data. The algorithm is extensively tested on different datasets (3DMatch, ModelNet, KITTI, MVP-RG), demonstrating outstanding generalization and robustness.

Salient object detection in optical remote sensing images (RSI-SOD) has recently become a key area of research, driven by the unique challenges posed by the variability in remote sensing imagery. Traditional approaches, largely based on Convolutional Neural Networks (CNNs), are limited in handling the diverse scenarios of remote sensing due to their static network construction and reliance on local feature extraction. To tackle these limitations, we present the Adaptive Semantic Network (ASNet), a novel framework specifically designed for RSI-SOD. ASNet innovatively integrates Transformer and CNN technologies in a Dual Branch Encoder, which captures both global dependencies and local fine-grained image details. The network also features an Adaptive Semantic Matching Module (ASMM) for dynamically harmonizing filter responses to global and local contexts, an Adaptive Feature Enhancement Module (AFEM) that effectively enhances salient region features while restoring image resolution, and a Multi-scale Fine-grained Inference Module (MFIM) which refines high-level semantic features by integrating detailed low-level information, leading to the generation of precise, high-quality saliency maps. These components work in concert to adaptively respond to the complex nature of remote sensing images. Extensive experimental evaluations confirm that ASNet substantially outperforms existing models in the RSI-SOD task.

Point cloud registration plays a key role in the fields of computer vision, particularly in scenarios with low overlap, large scenes, different datasets, where difficulties, such as difficulty matching, scale changes and geometric deformations, local feature loss are commonly encountered. In this article, we propose a point cloud registration algorithm named multicompound adaptive transformer, which introduces adaptive position encoding, dynamically adjusting the local coordinates and feature information of scattered points within the point cloud through an adaptive threshold enhancement mechanism. Simultaneously, the multicompound transformer is introduced. In the spatial transformer stage, it accomplishes the local position-local feature interaction of individual point clouds through adaptive position encoding. Then, in the temporal transformer stage, it achieves local–local interaction and local–global information interaction between two point clouds through a dual-branch multiscale transformer. Through experiments on different datasets, we validate the algorithm's superior generalization performance in scenarios with low overlap, large scenes, and different datasets.

Three-dimensional (3D) modeling is currently a creative task that requires modelers with strong professional skills and background knowledge, especially in the field of 3D modeling of Chinese ancient architecture (CAA). At present, most of the studies on 3D CAA modeling are based on hard-coded constructive rules, which need completed, complex and formalized descriptions. We present a generative system bridging the gap between the Chinese text and 3D models that allows users to generate 3D models by natural language. First, a Bayesian network is learned from existing CAA data to provide relationships of different structural components. Second, by parsing the Chinese text inputted by the user, key components of the CAA will be determined; and other matched structural components will be calculated by inferencing the trained Bayesian network. Third, the synthesis of all components is achieved by a proposed placement optimizing algorithm. Finally, we evaluate the effectiveness of the trained Bayesian network and demonstrate the application to generate 3D CAA model rapidly from the Chinese text.

In this paper we propose an improve algorithm for ancient porcelain classification, which contain three steps. First, image preprocessing. A color quantization method in HSI color space is performed to generate gray image. Second, feature extraction. An approach of color-texture features extraction is proposed based on Gabor filter, which is only depended on frequency. In order to reduce number of features, principal component analysis is adopted. Last, porcelain shards classifications. Nearest Neighbor Method is adopted to classify shards. An ancient shards classification prototype system is developed and help archeologist easily to do research on porcelain and restore the broken porcelain. The system has been practical to the recovery of Yao Zhou's porcelains, which are famous in ancient China.