Zehua Chen

Identifying spatial domains is crucial in spatial transcriptomics, yet effectively integrating gene expression, spatial location, and histology remains challenging. We present STESH, a Spatial Transcriptomics clustering method that combines Expression, Spatial information and Histology. STESH extracts histological features using a convolutional neural network and generates expression, histology, spatial, and collaborative convolution modules for a multi-view graph convolutional network with a decoder and attention mechanism. We evaluated STESH on multiple tissue types and technology platforms. STESH consistently outperformed ten state-of-the-art methods, achieving superior clustering accuracy with the highest scores in adjusted Rand index, normalized mutual information, and Fowlkes-Mallows index.

UAV path planning in complex 3D terrain faces the dual challenges of computational efficiency and reliable obstacle avoidance. To address these issues, this paper proposes a Keypoint–Sparse Cache (KSC) strategy and a hierarchical KSC-PPO (Proximal Policy Optimization) framework for mountainous environments with both static terrain and dynamic obstacles. The KSC strategy reduces search complexity through orthogonal slice-based sparse keypoint extraction and path caching reuse, thereby improving the efficiency of global path planning. On this basis, PPO-based local obstacle avoidance is activated only when safety thresholds are exceeded, while the remaining path is replanned globally after threat clearance, which confines avoidance computation to a local scope while preserving global path quality. Experiments in static mountainous environments show that KSC requires substantially less computation time than RRT* and Informed RRT* while maintaining competitive path efficiency, and it also outperforms four bio-inspired optimization algorithms across terrains of increasing complexity. Hybrid navigation validation experiments further show that KSC-PPO achieves high mission success, low collision rates, and low avoidance overhead in dynamic mountainous environments. Experiments demonstrate that KSC-PPO decomposes exponential global search space into controllable linear subproblems, significantly enhancing efficiency while ensuring path quality, providing an effective solution for UAV navigation in complex terrain.

Road extraction from remote sensing images (RSIs) has long been a critical task due to its broad application prospects such as digital twins for smart cities and intelligent transportation systems. However, extracting high-precision roads from RSIs remains challenging for the following reasons: (1) Ambiguity caused by background objects with geometric similarities to roads; (2) Topological discontinuities resulting from background occlusions. To address these issues, we propose a Direction-Aware Road Extraction network (DARENet). First, to address the ambiguity caused by insufficient structural features of extracted roads, we introduce a Multiscale Feature Extraction (MSFE) module with different directions in the skip connections; Second, to alleviate the issue of road topological discontinuities, we propose a Multidirectional Global-Local Fusion (MDGLF) module, which captures the relationship between roads and their environment through global and local processing. Finally, evaluations on three public datasets demonstrate that DARENet outperforms existing models. Our source code is available at https://github.com/ZehuaChenLab/DARENet.

Change detection (CD) is a fundamental task in remote sensing imagery, playing a pivotal role in diverse applications such as urban planning, environmental monitoring, and natural disaster assessment. However, most existing methods focus on processing entire images, with limited exploration of cropping images into strip patches and performing operations within these patches. These methods suffer from a waste of computational cost and memory resources in many areas unchanged. In this article, we present a novel paradigm for CD task and introduce a lightweight yet powerful framework, termed SP4CD. Specifically, SP4CD employs a Siamese backbone as a shared feature extractor to capture discriminative and hierarchical representations from bi-temporal remote sensing imagery. Furthermore, an interactive fusion mechanism is devised to integrate the extracted representations across spatial and channel dimensions, thereby enhancing cross-temporal feature interaction. To this end, we design a Patch-wise Intertemporal Channel Integration (PICI) module. By introducing a patch-swapping strategy, PICI integrates dual-temporal feature channels in a nearly parameter-free manner, achieving adaptive channel enhancement while effectively alleviating misclassification errors. In addition, we explore an Intertemporal Synergistic Patch Scanning (ISPS) module. It refines spatial representations through difference-guided selective scan (SS2D) of patches, optimizing detail in changing regions to reduce false negatives. Moreover, two complementary scanning modes (parallel and series modes) are designed based on guidance sequences derived from distinct feature representations, enabling adaptive spatial exploration from multiple perspectives. Extensive experiments conducted on four challenging benchmark datasets verify that the proposed SP4CD achieves competitive or superior performance across multiple evaluation metrics, confirming its effectiveness and generalization capability. Our source code is available at: https://github.com/ZehuaChenLab/SP4CD.

Gantry cranes play a key role in modern industrial logistics. However, the traditional dynamic model based on the assumption of cable rigidity faces difficulty in accurately describing the complex swing characteristics of flexible cables, resulting in low load positioning accuracy and limited operation efficiency. To address this problem, this paper proposes a cable modeling method that considers the flexible deformation and nonlinear dynamic characteristics of the cable. Based on the theory of continuum mechanics, a flexible cable dynamic model that can accurately describe the flexible deformation and distributed mass characteristics of the cable is established. In order to solve the transportation time optimization and full-state constraint problems, a velocity trajectory optimization algorithm based on a discretization framework is proposed. Through inverse kinematics analysis and numerical integration technology, a reverse angle enumeration reasoning (RAER) method is proposed to suppress the swing of the load. Under the same constraints of distance, velocity, acceleration, cable swing angle, and residual swing angle, RAER requires a longer transportation time but achieves smaller peak swing and residual swing, making it the only algorithm that satisfies full-state constraints. Under the energy criterion, the proposed algorithm also requires the least amount of energy. Comprehensive comparisons through simulations and experiments show that the predicted swing angles of the flexible cable are highly consistent with the experimental results.

Road extraction from remote sensing imagery plays a pivotal role in a wide range of geospatial and urban applications. Nevertheless, this task remains inherently challenging due to the intricate morphological variations of roads and frequent occlusions or interference caused by complex background environments. To address these challenges, we propose a road extraction network based on Gated Global-Local Linear Attention (G²L²Net). Firstly, we introduce a linear deformable convolution and design a Linear Input-Dependent Deformable Convolution (LID²Conv) which adaptively modulates convolution offsets and weights in a content-aware manner. In addition, we design a Top-K based sparse gated weight (TGW). We use this gated mechanism as a shared weight to multiply with local and global information to achieve gated global-local linear attention (G²L²Attention). Local information is obtained by LID²Conv and we gain global information by introducing 2D Selective Scan (SS2D). These two pathways are integrated through the proposed G²L²Attention, enabling an efficient and consistent fusion of hierarchical spatial features. The extracted features are passed to the decoder. This approach improves road detail representation and provides accurate contextual information. Experiments conducted on three public road datasets demonstrate that G²L²Net outperforms existing methods in various evaluation metrics. Our source code is available at: https://github.com/ZehuaChenLab.