In recent years, researchers have paid increasing attention on hyperspectral image (HSI) classification using deep learning methods. To improve the accuracy and reduce training samples, we propose a double-branch dual-attention mechanism network (DBDA) for HSI in this paper. Two branches are designed DBDA to capture plenty of spectral spatial features contained HSI. Furthermore, channel block applied these two respectively, which enables refine optimize extracted feature maps. A series...

In recent years, researchers have paid increasing attention on hyperspectral image (HSI) classification using deep learning methods. To improve the accuracy and reduce training samples, we propose a double-branch dual-attention mechanism network (DBDA) for HSI in this paper. Two branches are designed DBDA to capture plenty of spectral spatial features contained HSI. Furthermore, channel block applied these two respectively, which enables refine optimize extracted feature maps. A series...

Accurate and efficient text detection in natural scenes is a fundamental yet challenging task computer vision, especially when dealing with arbitrarily-oriented texts. Most contemporary methods are designed to identify horizontal or approximately text, which cannot satisfy practical requirements for various real-world images such as image streams videos. To address this lacuna, we propose novel method called Rotational You Only Look Once (R-YOLO), robust real-time convolutional neural...

Three-dimensional (3D) line segments are helpful for scene reconstruction. Most of the existing 3D-line-segment reconstruction algorithms deal with two views or dozens small-size images; while in practice there usually hundreds thousands large-size images. In this paper, we propose an efficient segment method called ELSR <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> Available at https://skyearth.org/publication/project/ELSR. exploits...

Loop closure detection (LCD) is a crucial technology within the field of simultaneous localization and mapping (SLAM), aiming to mitigate cumulative errors maintain globally consistent map. Traditional fixed-Light Detection Ranging (LiDAR) scans have demonstrated reliable LCD performance when employing global descriptors. However, for rotary-LiDAR scans, these methods fall short in place recognition due significant view-angle changes. To address this challenge, we propose novel algorithm...

Loop closure detection (LCD) plays an essential role in the Simultaneous Localization and Mapping (SLAM) process, effectively reducing cumulative trajectory errors. However, conventional LCD methods often encounter challenges when dealing with variations illumination, changes viewpoint, environments weak textures. This is due to their reliance on low-level geometric or image features. To address these issues, we propose a robust method named SL-LCD, which integrates semantic information line...

Structural color filters based on all-dielectric materials are considered to be promising alternatives metal nanostructures due significant advantages, such as high-quality resonance effects and low losses of Ohmic effects. We demonstrate a polarization tunable filter metasurfaces, which is the arrays asymmetric monocrystalline silicon nanoblocks flexible substrate. By adjusting physical dimensions nanoblocks, can exhibit variety bright transmission colors. Furthermore, designed dielectric...

Digital reconstruction of ancient buildings is very challenging due to its architectural complexity and structural delicacy. Therefore, how apply new Earth Observation (EO) technology digitally reconstruct the complete real model without damage has become an urgent issue. This article proposes a multi-scan point cloud hierarchical registration method for 3D Chinese Great Wild Goose Pagoda based on Terrestrial Laser Scanner (TLS) data. First, elaborately designed reconstruction. Then, outline...

Accurate and efficient three-dimensional (3D) streetscape reconstruction is the fundamental ability for an exploration vehicle to navigate safely perform high-level tasks. Recently, remarkable progress has been made in with visual images light detection ranging (LiDAR), but they have difficulties either scaling reconstructing large-scale outdoors or processing. To address these issues, this paper proposed automatic method incremental dense of 3D streetscapes from coarse fine at near real...

Seed geometric parameters are important in yielding trait scorers, quantitative loci, and species recognition classification. A novel method for automatic measurement of three-dimensional seed phenotypes is proposed. First, a handheld (3D) laser scanner employed to obtain the point cloud data batches. Second, cloud-based phenotyping proposed single-seed 3D model extract 33 phenotypes. It connected by an pipeline, including segmentation, pose normalization, completion ellipse fitting method,...

We propose a practical approach to identify plasmonic absorption profile on back focal plane (BFP) of surface plasmon microscopy (SPM). Compared previous morphology approaches, the proposed one features: i) it is applicable in BFP images with non-concentricity; ii) algorithm Fourier Correlation Analysis (FCA) maximizes SPP while minimizes random coherent noises, which makes extremely suited for significant low-quality; (iii) takes much less time identification process and entire can be...

For positioning problems of the complex environment urban space, this paper proposed an image-based visual Iocalization approach to space by smartphone, realizing high-precision spatial city's $\circ $ In method, firstly, sequence buildings in prepositioned city is collected and SIFT used for rough matching. The 3D point cloud model building obtained through optimization matching precisely solving fundamental matrix projection matrix. By feature points, pixel coordinates object side library....

This article proposes a two-stage simultaneous localization and mapping (SLAM) method based on using the red green blue-depth (RGB-D) camera in dynamic environments, which can not only improve tracking robustness trajectory accuracy but also reconstruct clean dense static background model environments. In first stage, to accurately exclude interference of features region from tracking, object mask is extracted by Mask-RCNN optimized connected component analysis reference frame-based method....

In order to retrieve the positioning image efficiently and quickly from a large number of different images realize three-dimensional spatial positioning, in this article, based on photogrammetry computer vision theory, new method big data under bag words model guidance is proposed. The consists two parts: retrieving positioning. First, complete retrieval by feature extraction, K-means clustering, building other processes, thus improve efficiency matching. Second, achieve interior exterior...

Image matching is the main flow of a three-dimensional reconstruction. With development computer processing technology, seeking image to be matched from large date sets which acquired different formats, scales and locations has put forward new request for matching. To establish three dimensional reconstruction based on big data images, this paper effective method visual bag words model. The technologies include building model First, extracting SIFT feature points images in database,...

Given a set (or multiset) S of n numbers and target number t, the subset sum problem is to decide if there that sums up t. There are several methods for solving this problem, including exhaustive search, divide-and-conquer method, Bellman's dynamic programming method. However, none them could generate universal light code. In paper, we present new deterministic algorithm based on novel data arrangement, which such code return all solutions. If small enough, it efficient usual purpose. We...