A CNN-Based Layer-Adaptive GCPs Extraction Method for TIR Remote Sensing Images

Abstract

Ground Control Points (GCPs) are of great significance for applications involving the registration and fusion of heterologous remote sensing images (RSIs). However, utilizing low-level information rather than deep features, traditional methods based on intensity and local image features turn out to be unsuitable for heterologous RSIs because of the large nonlinear radiation difference (NRD), inconsistent resolutions, and geometric distortions. Additionally, the limitations of current heterologous datasets and existing deep-learning-based methods make it difficult to obtain enough precision GCPs from different kinds of heterologous RSIs, especially for thermal infrared (TIR) images that present low spatial resolution and poor contrast. In this paper, to address the problems above, we propose a convolutional neural network-based (CNN-based) layer-adaptive GCPs extraction method for TIR RSIs. Particularly, the constructed feature extraction network is comprised of basic and layer-adaptive modules. The former is used to achieve the coarse extraction, and the latter is designed to obtain high-accuracy GCPs by adaptively updating the layers in the module to capture the fine communal homogenous features of the heterologous RSIs until the GCP precision meets the preset threshold. Experimental results evaluated on TIR images of SDGSAT-1 TIS and near infrared (NIR), short wave infrared (SWIR), and panchromatic (PAN) images of Landsat-8 OLI show that the matching root-mean-square error (RMSE) of TIS images with SWIR and NIR images could reach 0.8 pixels and an even much higher accuracy of 0.1 pixels could be reached between TIS and PAN images, which performs better than those of the traditional methods, such as SIFT, RIFT, and the CNN-based method like D2-Net.

Static and Dynamic Hand Gestures: A Review of Techniques of Virtual Reality Manipulation

“…Visual hardware first captures an image with a camera, which is usually infrared to obtain more accurate information [ 44 ]. At this stage, the images usually present background or ambient noise, which provides information to the images and complicates gesture detection with pre-established patterns [ 45 , 46 ]. Therefore, noise-eliminating filters are applied.…”

The Contribution of the Pore Size of Titanium DC (Direct Current) Sputtered Condensation Polymer Materials to Electromagnetic Interruption and Thermal Properties

“…Regarding infrared rays, research on infrared image fusion, object-oriented attention, infrared and visible light image fusion, and remote sensing images is also being actively conducted [29][30][31][32]. The aim of infrared and visible image fusion techniques is to extract and integrate features from images captured using various sensors using specific algorithms to create complementary images that contain both rich detailed features of visible images and target information of infrared images.…”

“…Zhao et al propose a CNN-based layer adaptive GCP extraction method for TIR RSI. Specifically, the built feature extraction network consists of a primary module and a layer-adaptive module [32].…”

Multi-Resolution Feature Extraction and Fusion for Traditional Village Landscape Analysis in Remote Sensing Imagery

“…In recent years, deep learning techniques have demonstrated considerable success in remote sensing image analysis [13][14][15][16][17][18], with convolutional neural networks (CNNs) in particular exhibiting remarkable feature extraction and image classification capabilities. Nonetheless, deep learning methods exhibit certain limitations when analyzing traditional village landscapes in remote sensing images [19][20][21][22].…”