Color plays an important role in human visual perception, reflecting the spectrum of objects. However, existing infrared and visible image fusion methods rarely explore how to handle multi-spectral/channel data directly achieve high color fidelity. This paper addresses above issue by proposing a novel method with diffusion models, termed as Dif-Fusion, generate distribution multi-channel input data, which increases ability multi-source information aggregation fidelity colors. In specific,...

Hyperspectral Image (HSI) classification is an important issue in remote sensing field with extensive applications earth science. In recent years, a large number of deep learning-based HSI methods have been proposed. However, existing limited ability to handle high-dimensional, highly redundant, and complex data, making it challenging capture the spectral-spatial distributions data relationships between samples. To address this issue, we propose generative framework for diffusion models...

Stem characteristics of plants are great importance to both ecology study and forest management. Terrestrial laser scanning (TLS) may provide an effective way characterize the fine-scale structures vegetation. However, clumping plants, dense foliage thin structure could intensify shadowing effect pose a series problems in identifying stems, distinguishing neighboring merging disconnected stem parts point clouds. This paper presents new method automatically detect stems homogeneous using...

Accurate land cover classification information is a critical variable for many applications. This study presents method to classify using the fusion data of airborne discrete return LiDAR (Light Detection and Ranging) CASI (Compact Airborne Spectrographic Imager) hyperspectral data. Four LiDAR-derived images (DTM, DSM, nDSM, intensity) (48 bands) with 1 m spatial resolution were spatially resampled 2, 4, 8, 10, 20 30 resolutions nearest neighbor resampling method. These thereafter fused...

Recently, there have been significant advancements in Hyperspectral Image (HSI) classification methods employing Transformer architectures. However, these methods, while extracting spectral-spatial features, may introduce irrelevant spatial information that interferes with HSI classification. To address this issue, paper proposes a Spectral Query Spatial (SQSFormer) framework. The proposed framework utilizes the center pixel (i.e., to be classified) adaptively query relevant from neighboring...

Edges in mobile light detection and ranging (lidar) point clouds are important for many applications but usually overlooked. In this letter, we propose a fast edge extraction method lidar. First, an index based on geometric center is introduced then gradients unorganized 3-D defined. By analyzing the ratio between eigenvalues, candidates can be detected. Finally, linking algorithm named graph snapping proposed. The tested extensively experimental results demonstrate that proposed able to...

We present a novel approach to large-scale point cloud surface reconstruction by developing an efficient framework that converts irregular into signed distance field (SDF). Our backbone builds upon recent transformer-based architectures (i.e., PointTransformerV3), serializes the locality-preserving sequence of tokens. efficiently predict SDF value at aggregating nearby tokens, where fast approximate neighbors can be retrieved thanks serialization. serialize different levels/scales, and...

æ–°ä¸€ä»£æ˜Ÿè½½æ¿€å ‰é›·è¾¾å«æ˜ŸICESat-2å°†é‡‡ç”¨å¤šæ³¢æŸå¾®è„‰å†²å ‰å­è®¡æ•°æŠ€æœ¯,å¹¶è¿›è¡Œé«˜ç¨‹å‰–é¢å¼çš„å¯¹åœ°è§‚æµ‹ã€‚ç”±äºŽè¯¥ç‚¹äº‘æ•°æ®å ·æœ‰èƒŒæ™¯å™ªå£°å¤§ã€å¯†åº¦ä½Žå¹¶å‘ˆçº¿çŠ¶åˆ†å¸ƒç­‰ç‰¹ç‚¹,ä¼ ç»Ÿçš„ç‚¹äº‘æ»¤æ³¢ç®—æ³•å¹¶ä¸é€‚ç”¨,ç ”ç©¶æ–°çš„ç‚¹äº‘æ»¤æ³¢ç®—æ³•ååˆ†å¿ è¦ã€‚æœ¬æ–‡ä»¥ICESat-2的机载模拟器MABEL数据为例,é¦–å ˆä»‹ç»äº†å¾®è„‰å†²å ‰å­è®¡æ•°æ¿€å...

Color plays an important role in human visual perception, reflecting the spectrum of objects. However, existing infrared and visible image fusion methods rarely explore how to handle multi-spectral/channel data directly achieve high color fidelity. This paper addresses above issue by proposing a novel method with diffusion models, termed as Dif-Fusion, generate distribution multi-channel input data, which increases ability multi-source information aggregation fidelity colors. In specific,...

Extraction of individual façades from building point clouds collected by ground-based LiDAR is vital for urban mapping and modeling. It a challenging task due to the complexity façades, very few studies have been conducted. In this paper, we present new façade separation method that can divide connected into instances using coordinates only. The proposed consists two steps. first step extracting edges windows clouds. step, an improved window detection which detect complex windows, such as...

Metro subway systems with underground tunnels form the backbone of urban transportations and therefore, accurate monitoring maintenance such are extremely necessary for a hassle-free daily commutation billions people. Though 3-D models widely used deformation tunnels, existing model-based tunnel rely on coarse geometric hence fail to capture complete health information. We present two-stage algorithm create high-fidelity lining from Terrestrial Laser Scanning (TLS) point clouds. Tunnel...

This paper presents a novel framework to achieve 3D semantic labeling of objects (e.g., trees, buildings, and vehicles) from airborne laser-scanning point clouds. To this end, we propose which consists hierarchical clustering higher-order conditional random fields (CRF) labeling. In the clustering, raw clouds are over-segmented into set fine-grained clusters by integrating density classic K-means algorithm, followed proposed probability algorithm. Through process, not only obtain more...

Although mobile light detection and ranging (LiDAR) technology has excellent potential in mapping street scenes, there is little research constructing façade footprints from unorganized, uneven, incomplete LiDAR point clouds. In fact, footprint vectorization data still involves a lot of manual work, especially complex scenes with various types buildings. this paper, we present new effective framework for extracting 2-D The proposed consists three steps: 1) line segment extraction projected...

RSS (Received Signal Strength) values which are used in indoor position estimation based on fingerprinting affected by noise. The value received the fixed line-of-sight condition points obeys Gauss distribution and does not match distribution. WIFI signal transmission attenuation is also a nonlinear attenuation. This paper presents joint KPCA-ELM locating algorithm, use of KPCA (Kernel Principal Component Analysis) characteristics allow original data being replaced dimension reduction,...

Laser point cloud filtering is a fundamental step in various applications of light detection and ranging (LiDAR) data. The progressive triangulated irregular network (TIN) densification (PTD) algorithm classic method widely used due to its robustness effectiveness. However, the performance PTD depends on quality initial TIN-based digital terrain model (DTM). effect also limited by tuning number parameters cope with terrains. Therefore, an improved based multiscale cylindrical neighborhood...

Tree localization in point clouds of forest scenes is critical the inventory. Most existing methods proposed for TLS data are based on model fitting or point-wise features which time-consuming, sensitive to incompleteness and complex tree structures. Furthermore, these often require lots preprocessing such as ground filtering noise removal. The fast easy-to-use top-based that widely applied processing ALS not applicable localizing trees due canopy objective this study make clouds. To end, a...