Accurate estimation of gross domestic product (GDP) at small geographies is great significance to evaluate the distribution and dynamics socio-economic development. Nighttime light (NTL) data becoming increasingly important in estimation. However, previous research has found that using NTL alone insufficient accurately measure GDP geographies, contribution for time-series unreliable. This article proposed a deep learning method Contiguous United States (CONUS) (2012-2015) county level. The...

Due to increasingly complex factors of image degradation, inferring high-frequency details remote sensing imagery is more difficult compared ordinary digital photos. This paper proposes an adaptive multi-scale feature fusion network (AMFFN) for super-resolution. Firstly, the features are extracted from original low-resolution image. Then several extraction (AMFE) modules, squeeze-and-excited and gating mechanisms adopted fusion. Finally, sub-pixel convolution method used reconstruct...

Since the normal matrix is often ill-conditioned when solving Rational Function Model (RFM) for satellite remote sensing imagery, combining with orthogonal decomposition, Levenberg–Marquardt algorithm and Compute Unified Device Architecture high-performance computing technique, a fast robust method directly based on error equation coefficient proposed. The analysed by different parameters of control point grid compared common methods, namely L-curve ridge estimation Iteration Correcting...

Aim to the difficulty of acquisition conjugate points from multi-source remote sensing imagery, a novel matching method based on SIFT and CRA similarity measure is proposed. Firstly, operator adopted extract feature coarse match performed, approximate transformation relationship rotation angle between matched images are estimated by above results. Secondly, if there exists large images, compensation image window should be carried out, used search corresponding points. Finally, mismatched in...

Abstract As a generic sensor model, the rational function model ( RFM ) has been widely used in geometric processing of optical images, but its application to synthetic aperture radar SAR datasets not developed sufficiently. In this paper thorough study ‐based bundle adjustment for simultaneous positioning multi‐sensor spaceborne imagery is made. Different control point layouts are designed analyse accuracy, and block adjustments based on rigorous compared with . Experiments show that can...

High-resolution optical remote sensing image registration is still a challenging task due to non-linearity in the intensity differences and geometric distortion. In this paper, an efficient method utilizing hyper-graph matching algorithm proposed, which can simultaneously use high-order structure information radiometric information, obtain thousands of feature point pairs for accurate registration. The mainly consists following steps: firstly, initial by Uniform Robust Scale-Invariant...

This paper proposes a robust matching method for the multi-sensor imagery. Firstly, SIFT feature and relaxation are integrated in highest pyramid to derive approximate relationship between reference slave image. Then, normalized Mutual Information multi-grid multi-level RANSAC algorithm adopted find correct conjugate points. Iteratively perform above steps until original image level, facet- based transformation model is used carry out registration. Experiments have been made, results show...

To enable multidimensional agricultural monitoring, this paper presents a lightweight transformer model for winter wheat yield estimation, which is based on the cell structure of original network, replacing decoding layer with linear layer, and trained using multiple sources data such as crop growth variables environmental from 2002 to 2020. Our approach compared other methods, including long short term memory (A-LSTM) gated recursive unit (GRU) models, introduced by self-focused mechanism....

In this paper, we present a matching method for DSM generation from multiple images based on feature points, which introduce the coarse-to-fine strategy, geometrically constrained and relaxation technology, is guided by information in object make full use of both image space. A match appearing any pair has chance to survive, very dense disparity maps are obtained. Experiments have been performed height accuracy derived about 3 pixels.

For the aerial images characterized by low contrast or containing repetitive patterns homogeneous textures, robust tie point matching is still a challenging task. In this letter, an effective hypergraph-based method proposed. Firstly, feature points are divided into several clusters in overlapping area of images. Secondly, two-stage pipeline, that is, candidate and high-order graph matching, performed. The relationship utilized to establish similarity tensor without any information loss....

Aim to the difficulty of automatic and robust registration optical imagery with point cloud data, this paper propose a new method based on SIFT Mutual Information (MI). The features are firstly extracted matched, whose result is used derive coarse geometric relationship between data. Secondly, MI-based similarity measure conjugate points. And then RANSAC algorithm adopted eliminate erroneous matching Repeating procedure MI mismatching points deletion until finest pyramid image level. Using...