Resolution Remote Sensing Images Research Articles

A HIGH-PRECISION EXPLICIT FOREST CARBON STOCK MODEL BASED ON REMOTE SENSING

Abstract. The current status as well as the potential absorption capability of the forest carbon sinks in terrestrial ecosystems are in urgent need of further studies. The ground observation-based methods are labor-intensive, and the resulting statistics from samples are difficult to evaluate, while inversion methods based on remote sensing data lack theoretical explanation and universality. This paper proposes a pixel-level, multi-scale, high-precision Explicit Forest carbon stock Model (EFM) that is universal and adaptive. First, four key variables were used in the construction of the EFM: remote sensing image resolution, forest canopy density, terrain slope, and forest height; Second, simulated forest scene were generated based on the growth characteristics of individual trees, and EFM parameters were solved and analyzed by these simulated pixels; Third, the EFM was tested at various scales and forest saturation levels to verify its accuracy, robustness, and applicability, the simulation and real-life experiments show that the correlation coefficient is greater than 0.98 and the relative error is about 20%. The EFM solves the problem that the existing methods are lack in theoretical interpretation and universal applicability, thus can be used to map forest carbon stocks at high resolution and large scale and even monitor forest carbon dynamics at global scale.

Contemporary and historical detection of small lakes using super resolution Landsat imagery: promise and peril

ABSTRACT Landsat is the longest-running environmental satellite program and has been used for surface water mapping since its launch in 1972. However, its sustained 30 m resolution since 1982 prohibits the detection of small water bodies, which are globally far more prevalent than large. Remote sensing image resolution is increasingly being enhanced through single image super resolution (SR), a machine learning task typically performed by neural networks. Here, we show that a 10× SR model (Enhanced Super Resolution Generative Adversarial Network, or ESRGAN) trained entirely with Planet SmallSat imagery (3 m resolution) improves the detection of small and sub-pixel lakes in Landsat imagery (30 m) and produces images (3 m resolution) with preserved radiometric properties. We test the utility of these Landsat SR images for small lake detection by applying a simple water classification to SR and original Landsat imagery and comparing their lake counts, sizes, and locations with independent, high-resolution water maps made from coincident airborne camera imagery. SR images appear realistic and have fewer missed detections (type II error) compared to low resolution (LR), but exhibit errors in lake location and shape, and yield increasing false detections (type I error) with decreasing lake size. Even so, lakes between ~500 and ~10,000 m2 in area are better detected with SR than with native-resolution Landsat 8 imagery. SR transformation achieves an F-1 score for water detection of 0.75 compared to 0.73 from native resolution Landsat. We conclude that SR enhancement improves the detection of small lakes sized several Landsat pixels or less, with a minimum mapping unit (MMU) of ~ 2/3 of a Landsat pixel – a significant improvement from previous studies. We also apply the SR model to a historical Landsat 5 image and find similar performance gains, using an independent 1985 air photo map of 242 small Alaskan lakes. This demonstration of retroactively generated 3 m imagery dating to 1985 has exciting applications beyond water detection and paves the way for further SR land cover classification and small object detection from the historical Landsat archive. However, we caution that the approach presented is suitable for landscape-scale inventories of lake counts and lake size distributions, but not for specific geolocational positions of individual lakes. Much work remains to be done surrounding technical and ethical guidelines for the creation, use, and dissemination of SR satellite imagery.

Semantic understanding of high spatial resolution remote sensing images using directional geospatial relationships

ABSTRACT Semantic understanding of high spatial resolution remote sensing (RS) images can be divided into object detection, object labelling, identification of geospatial relationships, and semantic description generation. Geographical relations represent the spatial distribution dependencies between geospatial entities such as points, lines, and polygons, and the topologies among them. Geospatial relations play a very important role in describing the relations between geographic objects. These relations can be broadly classified as topological, directional, and proximity relations. These relations describe the adjacency and association relations between geospatial objects. An approach to identify an appropriate directional geospatial relationship between geo-objects present in high spatial resolution RS images is proposed in this paper. Geospatial objects in the form of the closed boundary are taken as input and relationship triplets are generated. Two approaches have been used in the identification of directional relationships and the results of both approaches are compared. The first approach is based on the centroid of the objects and the second considers whole objects while calculating the direction. These relations are then further represented using a knowledge graph, where nodes represent objects and edges represent their relationship. Knowledge graph plays a very important role in overall scene understanding. It shows the association of all objects with each other. These relationships are then represented in the form of descriptions by using template-based sentence generation. Results show that these directional relationships are accurately identified between each pair of objects using both approaches, but relations generated by considering whole objects are closer to human cognition. Semantic understanding of remote sensing images is of great significance in different applications such as urban surveys, urban planning, and management, military intelligence, etc.

Single-Image Super Resolution of Remote Sensing Images with Real-World Degradation Modeling

Limited resolution is one of the most important factors hindering the application of remote sensing images (RSIs). Single-image super resolution (SISR) is a technique to improve the spatial resolution of digital images and has attracted the attention of many researchers. In recent years, with the advancement of deep learning (DL) frameworks, many DL-based SISR models have been proposed and achieved state-of-the-art performance; however, most SISR models for RSIs use the bicubic downsampler to construct low-resolution (LR) and high-resolution (HR) training pairs. Considering that the quality of the actual RSIs depends on a variety of factors, such as illumination, atmosphere, imaging sensor responses, and signal processing, training on “ideal” datasets results in a dramatic drop in model performance on real RSIs. To address this issue, we propose to build a more realistic training dataset by modeling the degradation with blur kernels and imaging noises. We also design a novel residual balanced attention network (RBAN) as a generator to estimate super-resolution results from the LR inputs. To encourage RBAN to generate more realistic textures, we apply a UNet-shape discriminator for adversarial training. Both referenced evaluations on synthetic data and non-referenced evaluations on actual images were carried out. Experimental results validate the effectiveness of the proposed framework, and our model exhibits state-of-the-art performance in quantitative evaluation and visual quality. We believe that the proposed framework can facilitate super-resolution techniques from research to practical applications in RSIs processing.

Luojia-HSSR: A high spatial-spectral resolution remote sensing dataset for land-cover classification with a new 3D-HRNet

ABSTRACT High Spatial and Spectral Resolution (HSSR) remote-sensing images can provide rich spectral bands and detailed ground information, but there is a relative lack of research on this new type of remote-sensing data. Although there are already some HSSR datasets for deep learning model training and testing, the data volume of these datasets is small, resulting in low classification accuracy and weak generalization ability of the trained models. In this paper, an HSSR dataset Luojia-HSSR is constructed based on aerial hyperspectral imagery of southern Shenyang City of Liaoning Province in China. To our knowledge, it is the largest HSSR dataset to date, with 6438 pairs of 256 × 256 sized samples (including 3480 pairs in the training set, 2209 pairs in the test set, and 749 pairs in the validation set), covering area of 161 km2 with spatial resolution 0.75 m, 249 Visible and Near-Infrared (VNIR) spectral bands, and corresponding to 23 classes of field-validated ground coverage. It is an ideal experimental data for spatial-spectral feature extraction. Furthermore, a new deep learning model 3D-HRNet for interpreting HSSR images is proposed. The conv-neck in HRNet is modified to better mine the spatial information of the images. Then, a 3D convolution module with attention mechanism is designed to capture the global-local fine spectral information simultaneously. Subsequently, the 3D convolution is inserted into the HRNet to optimize the performance. The experiments show that the 3D-HRNet model has good interpreting ability for the Luojia-HSSR dataset with the Frequency Weighted Intersection over Union（FWIoU） reaching 80.54%, indicating that the Luojia-HSSR dataset constructed in this paper and the proposed 3D-HRnet model have good applicable prospects for processing HSSR remote sensing images.

An End-to-End Atrous Spatial Pyramid Pooling and Skip-Connections Generative Adversarial Segmentation Network for Building Extraction from High-Resolution Aerial Images

Automatic building extraction based on high-resolution aerial imagery is an important challenge with a wide range of practical applications. One of the mainstream methods for extracting buildings from high-resolution images is deep learning because of its excellent deep feature extraction capability. However, existing models suffer from the problems of hollow interiors of some buildings and blurred boundaries. Furthermore, the increase in remote sensing image resolution has also led to rough segmentation results. To address these issues, we propose a generative adversarial segmentation network (ASGASN) for pixel-level extraction of buildings. The segmentation network of this framework adopts an asymmetric encoder–decoder structure. It captures and aggregates multiscale contextual information using the ASPP module and improves the classification and localization accuracy of the network using the global convolutional block. The discriminator network is an adversarial network that correctly discriminates the output of the generator and ground truth maps and computes multiscale L1 loss by fusing multiscale feature mappings. The segmentation network and the discriminator network are trained alternately on the WHU building dataset and the China typical cities building dataset. Experimental results show that the proposed ASGASN can accurately identify different types of buildings and achieve pixel-level high accuracy extraction of buildings. Additionally, compared to available deep learning models, ASGASN also achieved the highest accuracy performance (89.4% and 83.6% IoU on these two datasets, respectively).

SR4RS: A Tool for Super Resolution of Remote Sensing Images

The SR4RS software includes tools to apply super-resolution methods on remote sensing images. It employs TensorFlow on the deep learning side, and relies on GDAL and the Orfeo ToolBox to deal with geospatial data. The software is written in simple python, and provides user-oriented applications to train and apply state of the art models on images.

Sonar Image Target Detection Based on Deep Learning

Due to the nonlinearity of the imaging of sonar equipment and the complexity of the underwater sound field environment, the gray level of the target area of the acquired underwater sonar image is relatively small. These characteristics are the target of the subsequent sonar image. Work such as detection and location tracking has brought great challenges. It has brought great challenges to solving the work of positioning and tracking, which makes the research of sonar image target detection based on deep learning very important. This article aims at studying the use of sonar to detect image targets based on deep learning technology. This article proposes a variety of sound image denoising methods based on multiresolution tools. The purpose of this article is to divide the natural image into blocks at an appropriate rate according to the change of the sampling matrix and measure the underwater natural image. The sound image defines an information model. These methods have greatly changed the image and period of using remote and temporary information. The translation results of these methods are all valid. The sharpening separation method based on filtered image and bidirectional detection should be published through a solution algorithm and different frames, and the expected algorithm can be reused and extracted as an action to improve the similarity of the image and should be saved and separated in detail. The result is correct. This article studies the application of deep learning methods in sonar image target detection and designs corresponding algorithms for improvement and functional realization in view of the current deficiencies and needs in this field. The experimental results show that the improved scheme and applied algorithm proposed in this article can achieve good results, the verification sample set includes 184 remote-sensing aircraft targets, and the resolution of remote-sensing images is unified to 1644 × 971 size. The accuracy of the target detection algorithm has been significantly improved, reaching 74.6%, and the detection speed has also been greatly improved. Compared with the RNN algorithm, the speed has been increased by 7 times. The detection results confirmed that the improved algorithm has higher positioning accuracy and faster detection speed.

SRSe-Net: Super-Resolution-Based Semantic Segmentation Network for Green Tide Extraction

Due to the phenomenon of mixed pixels in low-resolution remote sensing images, the green tide spectral features with low Enteromorpha coverage are not obvious. Super-resolution technology based on deep learning can supplement more detailed information for subsequent semantic segmentation tasks. In this paper, a novel green tide extraction method for MODIS images based on super-resolution and a deep semantic segmentation network was proposed. Inspired by the idea of transfer learning, a super-resolution model (i.e., WDSR) is first pre-trained with high spatial resolution GF1-WFV images, and then the representations learned in the GF1-WFV image domain are transferred to the MODIS image domain. The improvement of remote sensing image resolution enables us to better distinguish the green tide patches from the surrounding seawater. As a result, a deep semantic segmentation network (SRSe-Net) suitable for large-scale green tide information extraction is proposed. The SRSe-Net introduced the dense connection mechanism on the basis of U-Net and replaces the convolution operations with dense blocks, which effectively obtained the detailed green tide boundary information by strengthening the propagation and reusing features. In addition, the SRSe-Net reducs the pooling layer and adds a bridge module in the final stage of the encoder. The experimental results show that a SRSe-Net can obtain more accurate segmentation results with fewer network parameters.

Deep Covariance Alignment for Domain Adaptive Remote Sensing Image Segmentation

Unsupervised domain adaptive (UDA) image segmentation has recently gained increasing attention, aiming to improve the generalization capability for transferring knowledge from the source domain to the target domain. However, in high spatial resolution remote sensing image (RSI), the same category from different domains (e.g., urban and rural) can appear to be totally different with extremely inconsistent distributions, which heavily limits the UDA accuracy. To address this problem, in this article, we propose a novel deep covariance alignment (DCA) model for UDA RSI segmentation. The DCA can explicitly align category features to learn shared domain-invariant discriminative feature representations, which enhance the ability of model generalization. Specifically, a category feature pooling (CFP) module is first used to extract category features by combining coarse outputs and deep features. Then, we leverage a novel covariance regularization (CR) to enforce the intracategory features to be closer and the intercategory features to be further separate. Compared with the existing category alignment methods, our CR aims to regularize the correlation between different dimensions of the features, and thus performs more robustly when dealing with divergent category features of imbalanced and inconsistent distributions. Finally, we propose a stagewise procedure to train the DCA to alleviate error accumulation. Experiments on both rural-to-urban and urban-to-rural scenarios of the LoveDA dataset demonstrate the superiority of our proposed DCA over other state-of-the-art UDA segmentation methods. Code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/Luffy03/DCA</uri> .

A Residual Attention and Local Context-Aware Network for Road Extraction from High-Resolution Remote Sensing Imagery

Extracting road information from high-resolution remote sensing images (HRI) can provide crucial geographic information for many applications. With the improvement of remote sensing image resolution, the image data contain more abundant feature information. However, this phenomenon also enhances the spatial heterogeneity between different types of roads, making it difficult to accurately discern the road and non-road regions using only spectral characteristics. To remedy the above issues, a novel residual attention and local context-aware network (RALC-Net) is proposed for extracting a complete and continuous road network from HRI. RALC-Net utilizes a dual-encoder structure to improve the feature extraction capability of the network, whose two different branches take different feature information as input data. Specifically, we construct the residual attention module using the residual connection that can integrate spatial context information and the attention mechanism, highlighting local semantics to extract local feature information of roads. The residual attention module combines the characteristics of both the residual connection and the attention mechanism to retain complete road edge information, highlight essential semantics, and enhance the generalization capability of the network model. In addition, the multi-scale dilated convolution module is used to extract multi-scale spatial receptive fields to improve the model’s performance further. We perform experiments to verify the performance of each component of RALC-Net through the ablation study. By combining low-level features with high-level semantics, we extract road information and make comparisons with other state-of-the-art models. The experimental results show that the proposed RALC-Net has excellent feature representation ability and robust generalizability, and can extract complete road information from a complex environment.

Single-frame super-resolution for high resolution optical remote-sensing data products

ABSTRACT The resolution of remote-sensing images is directly related to the application value of data products. Due to differences in imaging characteristics between digital cameras and remote-sensing cameras, the existing network models cannot get the best super resolution (SR) results of satellite images. In response to the requirements of remote-sensing image production, we propose a single image super-resolution (SISR) reconstruction method for specific type of remote-sensing satellite. First, we measure and model the imaging degradation phenomenon of remote-sensing satellite, including the image blur and noise model. Then, high-quality aerial images are down-sampled and degraded to construct paired training image datasets. We chose the most popular Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) as the basic structure and optimized the number of Residual-in-Residual Dense Block (RRDB) modules to further improve the processing efficiency. Finally, we perform a series of quantitative measurements of the SR image results, including image interpretation capability, reconstruction accuracy, ground resolution distance, and data processing efficiency, using higher resolution remote-sensing images as the benchmark. The experimental results demonstrate that our proposed method has higher interpretation capability and reconstruction accuracy for the SR processing of specific type remote-sensing satellite. Our proposed method is evaluated within a real satellite image product, that demonstrated it has the capability of pipeline super-resolution processing of remote sensing data products.

Estimating the Influence of Oyster Reef Chains on Freshwater Detention at the Estuary Scale Using Landsat-8 Imagery

Oyster reef chains grow in response to local hydrodynamics and can redirect flows, particularly when reef chains grow perpendicular to freshwater flow paths. Singularly, oyster reef chains can act as porous dams that may facilitate nearshore accumulation of fresh or low-salinity water, in turn creating intermediate salinities that support oyster growth and estuarine conditions. However, oyster-driven freshwater detention has only been confirmed by limited, point-scale observational data, and simplified models. Oyster reef-driven freshwater detention in real ecosystems at the estuary scale remains largely unexplored. In this study, we analyzed the visible bands in 30-m resolution remote sensing (RS) images recorded by the Operational Land Imager aboard Landsat-8 to characterize the freshwater detention effect of oyster reef chains across a set of hydrologic conditions. Our results support prior findings indicating that 30-m resolution RS images recorded by the Operational Land Imager aboard Landsat-8 are useful for analyzing coastal dynamics after atmospheric correction, despite having been originally designed for terrestrial studies. Statistical models of water-leaving reflectance revealed that freshwater detention by oyster reefs was evident across the estuary, with the greatest effect occurring in the region closest to shore. Additionally, statistical modeling results and spatial patterns apparent in the satellite images suggested that reef-driven freshwater detention occurred under high riverine discharge conditions, but was less evident when flow was low. Beyond offering insight on the potential role of oyster reefs as mediators of estuarine hydrology, this study presents a transferable methodological framework for exploring estuarine biophysical feedbacks in blackwater river estuaries using satellite remote sensing.

Road Feature Extraction from LANDSAT-8 and ResourceSat-2 Images

This paper presents a methodology of road feature extraction from the different resolutions of Remote Sensing images of Landsat-8 Operational Lander Image (OLI) and ResourceSat-2 of Linear Imaging Self Sensor-3 (LISS-3) and LISS-4 sensors with the spatial resolutions of 15 m, 24 m, and 5 m. In the methodology of road extraction, an index is proposed based on the spectral profile of Roads, also involving Morphological transform (Top-Hat or Bot-Hat) and Markov Random Fields (MRF). In the proposed index, Short Wave Infrared (SWIR) band has a significant role in the detection of roads from sensors, and it is named Normalized Difference Road Index (NDRI). To enhancement of features from the index, Bot-Hat transforms used. To segment the road features from this image, MRF used. The methodology is performed on the OLI, LISS-3 and LISS-4 images, and presented with results.

Learning a robust CNN-based rotation insensitive model for ship detection in VHR remote sensing images

ABSTRACT Deep convolutional neural networks (CNN) have been widely applied in various fields, especially in the field of object detection. Deep CNN-based models showed great advantages over many traditional methods, even so, there are still many specific problems in the application of certain scenarios. In very high resolution (VHR) remote-sensing image datasets, the uncertainty of the object direction angle causes big trouble to the learning of the detector. Although the pooling operation can slightly alleviate the deviation caused by small angle, the feature learning of the objects with larger angle rotation still relies mainly on the sufficiency of sample data or effective data augmentation, which means the insufficiency of the training instances may cause serious performance degradation of the detector. In this paper, we propose a multi-angle box-based rotation insensitive object detection structure (MRI-CNN), which is an extended exploration for typical region-based CNN methods. On the one hand, we defined a set of directionally rotated bounding boxes before learning, and restricted the classification scene in a small angular range by rotated RoI (Region of Interest) pooling. On the other hand, we proposed a more effective screening method of bounding boxes, enabling the detector to adapt to diverse ground truth annotation methods and learn more accurate object localization. We trained our detector with different datasets containing different amount of training data, and the test results showed that the method proposed in this paper performs better than some mainstream detection methods when limited training data are provided in VHR remote-sensing datasets.

A Remote-Sensing Image Pan-Sharpening Method Based on Multi-Scale Channel Attention Residual Network

Pan-sharpening is a significant task that aims to generate high spectral- and spatial- resolution remote-sensing image by fusing multi-spectral (MS) and panchromatic (PAN) image. The conventional approaches are insufficient to protect the fidelity both in spectral and spatial domains. Inspired by the robust capability and outstanding performance of convolutional neural networks (CNN) in natural image super-resolution tasks, CNN-based pan-sharpening methods are worthy of further exploration. In this paper, a novel pan-sharpening method is proposed by introducing a multi-scale channel attention residual network (MSCARN), which can represent features accurately and reconstruct a pan-sharpened image comprehensively. In MSCARN, the multi-scale feature extraction blocks comprehensively extract the coarse structures and high-frequency details. Moreover, the multi-residual architecture guarantees the consistency of feature learning procedure and accelerates convergence. Specifically, we introduce a channel attention mechanism to recalibrate the channel-wise features by considering interdependencies among channels adaptively. The extensive experiments are implemented on two real-datasets from GaoFen series satellites. And the results show that the proposed method performs better than the existing methods both in full-reference and no-reference metrics, meanwhile, the visual inspection displays in accordance with the quantitative metrics. Besides, in comparison with pan-sharpening by convolutional neural networks (PNN), the proposed method achieves faster convergence rate and lower loss.

A Deep Multi-Attention Driven Approach for Multi-Label Remote Sensing Image Classification

Deep learning (DL) based methods have been found popular in the framework of remote sensing (RS) image scene classification. Most of the existing DL based methods assume that training images are annotated by single-labels, however RS images typically contain multiple classes and thus can simultaneously be associated with multi-labels. Despite the success of existing methods in describing the information content of very high resolution aerial images with RGB bands, any direct adaptation for high-dimensional high-spatial resolution RS images falls short of accurate modeling the spectral and spatial information content. To address this problem, this paper presents a novel approach in the framework of the multi-label classification of high dimensional RS images. The proposed approach is based on three main steps. The first step describes the complex spatial and spectral content of image local areas by a novel KBranch CNN that includes spatial resolution specific CNN branches. The second step initially characterizes the importance scores of different local areas of each image and then defines a global descriptor for each image based on these scores. This is achieved by a novel multi-attention strategy that utilizes the bidirectional long short-term memory networks. The final step achieves the classification of RS image scenes with multilabels. Experiments carried out on BigEarthNet (which is a large-scale Sentinel-2 benchmark archive) show the effectiveness of the proposed approach in terms of multi-label classification accuracy compared to the state-of-the-art approaches. The code of the proposed approach is publicly available at https://gitlab.tubit.tuberlin.de/rsim/MAML-RSIC.

Research on Method of Farmland Obstacle Boundary Extraction in UAV Remote Sensing Images.

Aimed at the problem of obstacle detection in farmland, the research proposed to adopt the method of farmland information acquisition based on unmanned aerial vehicle landmark image, and improved the method of extracting obstacle boundary based on standard correlation coefficient template matching and assessed the influence of different image resolutions on the precision of obstacle extraction. Analyzing the RGB image of farmland acquired by unmanned aerial vehicle remote sensing technology, this research got the following results. Firstly, we applied a method automatically registering coordinates, and the average deviations on the X and Y direction were 4.6 cm and 12.0 cm respectively, while the average deviations manually by ArcGIS were 4.6 cm and 5.7 cm. Secondly, with an improvement on the step of the traditional correlation coefficient template matching, we reduced the time of template matching from 12.2 s to 4.6 s. The average deviation between edge length of obstacles calculated by corner points extracted by the algorithm and that by actual measurement was 4.0 cm. Lastly, by compressing the original image on a different ratio, when the pixel reached 735 × 2174 (the image resolution reached 6 cm), the obstacle boundary was extracted based on correlation coefficient template matching, the average deviations of boundary points I of six obstacles on the X and Y were respectively 0.87 and 0.95 cm, and the whole process of detection took about 3.1 s. To sum up, it can be concluded that the algorithm of automatically registered coordinates and of automatically extracted obstacle boundary, which were designed in this research, can be applied to the establishment of a basic information collection system for navigation in future study. The best image pixel of obstacle boundary detection proposed after integrating the detection precision and detection time can be the theoretical basis for deciding the unmanned aerial vehicle remote sensing image resolution.

SLIC segmentation method for full‐polarised remote‐sensing image

The rapid development of remote-sensing data acquisition technology means that the resolution of remote-sensing image has been continuously improved, resulting in the large scale of remote-sensing image data and the increase of redundant information, which restricts the image processing and analysis. However, super-pixel segmentation method mainly aimed at the general image segmentation algorithm, rarely for remote-sensing image. Therefore, here, the polarisation decomposition of full-polarised remote-sensing images is used to synthesise pseudo-colour images. Then, based on the region, the image segmentation algorithm based on the colour feature of SLIC super-pixel segmentation algorithm is used to make it have important application value in remote-sensing image target extraction. Here, the image of Danjiangkou reservoir obtained by high-resolution space-borne SAR is selected as the research object. Finally, the polarisation decomposition and segmentation of full-polarised remote-sensing images are realised, and the advantages of SLIC super-pixel algorithm are proved.

Learning rotation-invariant binary codes for efficient object detection from remote sensing images

Object detection is one of the most important research topics for remote sensing image (RSI) analysis. Because of the rapid development of satellite imaging technology, the resolution of RSI has increased dramatically, which brings a great challenge to the efficiency of RSI-based object detection. We present an object detection model, which greatly improves the detection speed by using learning-based hashing techniques. Specifically, first, a selective search method is used to generate many high quality object proposals that may contain objects. Then for each proposal, we learn rotation-invariant binary codes, which can handle the objects with arbitrary orientations in RSI, to quickly eliminate most nonobject proposals in Hamming space. And finally, the object detection task can be achieved by classifying the left (very limited amount of) proposals with more discriminating classification model. Experimental evaluations on a public very high-resolution remote sensing dataset show the superiority of the proposed method. Specifically, the rotation-invariant learning metric has been combined with three popular hashing methods, and performance improvements have been obtained for all the cases. Furthermore, comparisons with deep learning-based detectors show that hash-based methods fail to achieve state-of-the-art detection performance, but when the graphics processing unit is inaccessible, it achieves a good compromise between performance and runtime.