Change detection is an important task in remotesensingimageprocessing and analysis. However, due to position errors and wind interference, bi-temporal low-altitude remotesensingimages collected by SUAVs often suf...
详细信息
Change detection is an important task in remotesensingimageprocessing and analysis. However, due to position errors and wind interference, bi-temporal low-altitude remotesensingimages collected by SUAVs often suffer from different viewing angles. The existing methods need to use an independent registration network for registration before change detection, which greatly reduces the integrity and speed of the task. In this work, we propose an end-to-end network architecture RegCD-Net to address change detection problems in the bi-temporal SUAVs' low-altitude remotesensingimages. We utilize global and local correlations to generate an optical flow pyramid and realize image registration through layer-by-layer optical flow fields. Then we use a nested connection to combine the rich semantic information in deep layers of the network and the precise location information in the shallow layers and perform deep supervision through the combined attention module to finally achieve change detection in bi-temporal images. We apply this network to the task of change detection in the garbage-scattered areas of nature reserves and establish a related dataset. Experimental results show that our RegCD-Net outperforms several state-of-the-art CD methods with more precise change edge representation, relatively few parameters, fast speed, and better integration without additional registration networks.
remote photoplethysmography (rPPG) is a promising technology for capturing physiological signals from facial videos, with potential applications in medical health, affective computing, and biometric recognition. The d...
详细信息
The rapid development of remotesensing technology provides wealthy data for earth observation. Land-cover mapping indirectly achieves biodiversity estimation at a coarse scale. Therefore, accurate land-cover mapping ...
详细信息
The rapid development of remotesensing technology provides wealthy data for earth observation. Land-cover mapping indirectly achieves biodiversity estimation at a coarse scale. Therefore, accurate land-cover mapping is the precondition of biodiversity estimation. However, the environment of the wetlands is complex, and the vegetation is mixed and patchy, so the land-cover recognition based on remotesensing is full of challenges. This paper constructs a systematic framework for multisource remotesensingimageprocessing. Firstly, the hyperspectral image (HSI) and multispectral image (MSI) are fused by the CNN-based method to obtain the fused image with high spatial-spectral resolution. Secondly, considering the sequentiality of spatial distribution and spectral response, the spatial-spectral vision transformer (SSViT) is designed to extract sequential relationships from the fused images. After that, an external attention module is utilized for feature integration, and then the pixel-wise prediction is achieved for land-cover mapping. Finally, land-cover mapping and benthos data at the sites are analyzed consistently to reveal the distribution rule of benthos. Experiments on ZiYuan1-02D data of the Yellow River estuary wetland are conducted to demonstrate the effectiveness of the proposed framework compared with several related methods.
The proceedings contain 16 papers. The topics discussed include: fully electrically controlled light-field camera via electrowetting liquid lens and liquid-crystal microlens array;study on transmission and nanofocusin...
ISBN:
(纸本)9781510674912
The proceedings contain 16 papers. The topics discussed include: fully electrically controlled light-field camera via electrowetting liquid lens and liquid-crystal microlens array;study on transmission and nanofocusing characteristics of surface array micronano metasurface;tuning of near-field optical properties based on magneto-tip array super-surfaces;ice area and 3D ice shape measurement method based on polarized light imaging;study on the polarization response of aluminum gratings with graphene;toroidal composite liquid crystal microlens array co-driven by four independent signal voltages;semi-supervised polarimetric SAR images classification based on FixMatch;and overview of remotesensingimage fusion based on deep learning.
The processing and recognition of geoscience images have wide applications. Most of existing researches focus on understanding the high-quality geoscience images by assuming that all the images are clear. However, in ...
详细信息
The processing and recognition of geoscience images have wide applications. Most of existing researches focus on understanding the high-quality geoscience images by assuming that all the images are clear. However, in many real-world cases, the geoscience images might contain occlusions during the image acquisition. This problem actually implies the image inpainting problem in computer vision and multimedia. As far as we know, all the existing image inpainting algorithms learn to repair the occluded regions for a better visualization quality, they are excellent for natural images but not good enough for geoscience images, and they never consider the following gescience task when developing inpainting methods. This paper aims to repair the occluded regions for a better geoscience task performance and advanced visualization quality simultaneously, without changing the current deployed deep learning based geoscience models. Because of the complex context of geoscience images, we propose a coarse-to-fine encoder-decoder network with the help of designed coarse-to-fine adversarial context discriminators to reconstruct the occluded image regions. Due to the limited data of geoscience images, we propose a MaskMix based data augmentation method, which augments inpainting masks instead of augmenting original images, to exploit the limited geoscience image data. The experimental results on three public geoscience datasets for remotesensing scene recognition, cross-view geolocation and semantic segmentation tasks respectively show the effectiveness and accuracy of the proposed method. The code is available at: https://***/HMS97/Task-driven-Inpainting.
remotesensingimage scene classification (RSSC), which assigns semantic labels to remotesensingimages, is very important for remotesensingimage interpretation. Thanks to the rapid development of deep learning, RS...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
remotesensingimage scene classification (RSSC), which assigns semantic labels to remotesensingimages, is very important for remotesensingimage interpretation. Thanks to the rapid development of deep learning, RSSC achieves significant breakthroughs by the use of convolutional neural network (CNN). However, CNN relies on local receptive fields and is difficult to capture long-range and global scene information. Moreover, the information of salient objects, which contributes to discriminate the category of scenes (e.g., airplanes indicate the airport scene), should be also exploited. To address this issue, a deep learning method, named multi-level representation learning (MLRL), is proposed to collaboratively extract pixel-level, patch-level, and object-level features, which respectively contain local, global, and object-oriented information. Specifically, pixel-level features are obtained by pixel-wise convolution operations within a CNN. Patch-level features are achieved by a patch-wise self-attention network. Object-level features are acquired by applying a CNN to a cropped sub-image, which conveys important information of salient objects. To this end, a three-branch network structure to respectively extract above features, is built. Finally, a decision fusion method is adopted to integrate multi-level features, and gives rise to refined classification results. Experiments conducted on widely-used datasets demonstrate the effectiveness of the proposed method.
remotesensingimages carry a wealth of information that is not easily accessible to end-users as it requires strong technical skills and knowledge. Visual Question Answering (VQA), a task that aims at answering an op...
详细信息
ISBN:
(数字)9781665490627
ISBN:
(纸本)9781665490627
remotesensingimages carry a wealth of information that is not easily accessible to end-users as it requires strong technical skills and knowledge. Visual Question Answering (VQA), a task that aims at answering an open-ended question in natural language from an image, can provide an easier access to this information. Considering the geographical information contained in remotesensingimages, questions often embed an important spatial aspect, for instance regarding the relative position of two objects. Our objective is to better model the spatial relations in the construction of a ground-truth database of image/question/answer triplets and to assess the capacity a VQA model has to answer these questions. In this article, we propose to use histograms of forces to model the directional spatial relations between geo-localized objects. This allows a finer modeling of ambiguous relationships between objects and to provide different levels of assessment of a relation (e.g. object A is slightly/strictly to the west of object B). Using this new dataset, we evaluate the performances of a classical VQA model and propose a curriculum learning strategy to better take into account the varying difficulty of questions embedding spatial relations. With this approach, we show an improvement in the performances of our model, highlighting the interest of embedding spatial relations in VQA for remotesensing applications.
remotesensing technology plays an important role in many tasks such as natural disaster detection, weather and climate monitoring and military defense. Currently, remotesensingimageprocessing predominantly relies ...
详细信息
作者：
Wang, YiqinJinzhong Univ
Sch Informat Technol & Engn 199 Wenhua St Jinzhong 030619 Shanxi Peoples R China
The current image semantic segmentation methods cannot meet the requirements of high precision and high speed for remotesensingimage analysis. The ENet network model builds a semantic segmentation network, which has...
详细信息
The current image semantic segmentation methods cannot meet the requirements of high precision and high speed for remotesensingimage analysis. The ENet network model builds a semantic segmentation network, which has the characteristics of few network parameters and fast operation speed. The attention mechanism module is integrated with the ENet network model, which can deeply mine image features in remotesensing datasets and ensure the accuracy of semantic segmentation. The author combines the ENet network with the attention mechanism to construct a new semantic segmentation network model. The model first constructed a remotesensingimage semantic segmentation network model based on the ENet network, and simplified the model to further improve the speed of image segmentation and recognition. Then, the attention mechanism module is fused with the ENet network model, which can conduct deep and orderly mining of the image features of the remotesensingimage data set. It can meet the accuracy requirements of remotesensingimage semantic analysis. Simulations are performed based on three general datasets, and the experimental results show high accuracy and high speed.
Hyperspectral image classification has become an important issue in remotesensing due to the significant amount of spectral information in HSI. The costly and time-consuming annotation task of HSIs makes the number o...
详细信息
Hyperspectral image classification has become an important issue in remotesensing due to the significant amount of spectral information in HSI. The costly and time-consuming annotation task of HSIs makes the number of labeled samples is limited. To address the above problem, we propose an enhanced semi-supervised support vector machine algorithm for spectral-spatial HSI classification. To fully capture the spectral and spatial information of HSI, we use local binary pattern to obtain spatial feature. The captured spatial features are concatenated with the spectral features to yield the hybrid spectral-spatial features. Self-training mechanism is then adopted to gradually select confident unlabeled samples with their pseudo-labels and add them to the labeled set. To further improve the classification performance of the semi-supervised support vector machine, we choose a cuckoo search algorithm based on the chaotic catfish effect to find its optimal combination of parameters. The experimental results on two publicly available HSI datasets show that the proposed model achieves excellent classification accuracy for each category in hyperspectral images, and also has superior overall accuracy compared with other comparative algorithms. Adequate experiments and analysis illustrate the promising potential and prospect of our proposed model for HSI classification.
暂无评论