Optical Remote Sensing Imagery Research Articles

Multi-feature sea–land segmentation based on pixel-wise learning for optical remote-sensing imagery

ABSTRACTRobust sea–land segmentation in optical remote-sensing images is challenging because of the complex sea–land environment and scene diversity. Here, we propose a novel multi-feature sea–land segmentation method via pixel-wise learning for optical remote-sensing images. Multiple features such as greyscale, local statistical information, edge, texture, and structure are first extracted from each pixel in training images and then used to learn a multi-feature sea–land classifier, which transforms the segmentation issue into pixel-wise binary classification problem. In our approach, a new multi-feature sea–land segmentation algorithm is put forward based on the approximation of Newton method. Experiments on Google-Earth, Venezuelan Remote Sensing Satellite-1 (VRSS-1) and Gaofen-1 images demonstrate that the proposed approach yields more robust and accurate sea–land segmentation results.

Landslide Identification and Information Extraction Based on Optical and Multispectral UAV Remote Sensing Imagery

Landslide is one of the most serious natural disasters which caused enormous economic losses and casualties in the world. Fast and accurate identification of newly occurred landslide and extraction of relevant information are the premise and foundation for landslide disaster assessment and relief. As the places where landslides occur are often inaccessible for field observation because of the temporary failure in transportation and communication. Therefore, UAV remote sensing can be adopted to collect landslide information efficiently and quickly with the advantages of low cost, flexible launch and landing, safety, under-cloud-flying, and hyperspatial image resolution. Newly occurred landslides are usually accompanied with those phenomena such as vegetation burying and bedrock or bare soil exposure, which can be easily detected in optical or multispectral UAV images. By taking one typical landslide occurred in Wenchuan Earthquake stricken area in 2010 as an example, this paper demonstrates the process of integration of multispectral camera with UAV platform, NDVI generation with multispectral UAV images, three-dimensional terrain and orthophoto generation with optical UAV images, and identification and extraction of landslide information such as its location, impacted area, and earthwork volume.

Statistical characterization of the sea ice extent during different winter scenarios in the Gulf of Riga (Baltic Sea) using optical remote-sensing imagery

ABSTRACTThis study focuses on the statistical characterization of ice conditions (extent, sea ice occurrence probability (SIOP), and length of ice season) in the Gulf of Riga, Baltic Sea, using remote-sensing data. The optical remote-sensing data with 250 m resolution acquired by a Moderate Resolution Imaging Spectroradiometer (MODIS) during 2002–2011 were used for statistical characterization of sea ice. A method based on bimodal histogram analysis of remote-sensing reflectance data was developed to discriminate ice from water. In general, ice extent information obtained from MODIS data agrees with the official ice chart data (synthetic aperture radar (SAR) and in situ measurements) and multi-sensor product containing data from microwave and infrared instruments (R2 >0.83). However, in case of severe winters and extremely mild winters there are differences in the dates when maximum ice extent is registered. MODIS data can be used for detailed analysis of ice extent in specific basins of Baltic Sea. Depending on the year, the ice season length in the Gulf of Riga ranged from 68 to 146 days, and the maximum ice extent varied greatly from 329 to 15,350 km2. SIOP and number of ice days increased significantly in areas where the depth is less than 15 m. Based on negative-degree days and ice cover characteristics (SIOP and ice season length), three winter scenarios were defined: severe (2003, 2006, 2010, and 2011), medium (2004 and 2005), and mild (2007, 2008, and 2009).

A Survey of Building Extraction Methods from Optical High Resolution Remote Sensing Imagery

A Survey of Building Extraction Methods from Optical High Resolution Remote Sensing Imagery

A Comparative Study of Sampling Analysis in the Scene Classification of Optical High-Spatial Resolution Remote Sensing Imagery

Scene classification, which consists of assigning images with semantic labels by exploiting the local spatial arrangements and structural patterns inside tiled regions, is a key problem in the automatic interpretation of optical high-spatial resolution remote sensing imagery. Many state-of-the-art methods, e.g., the bag-of-visual-words model and its variants, the topic models and unsupervised feature learning-based approaches, share similar procedures: patch sampling, feature learning and classification. Patch sampling is the first and a key procedure, and it has a considerable influence on the results. In the literature, many different sampling strategies have been used, e.g., random sampling and saliency-based sampling. However, the sampling strategy that is most suitable for the scene classification of optical high-spatial resolution remote sensing images remains unclear. In this paper, we comparatively study the effects of different sampling strategies under the scenario of scene classification of optical high-spatial resolution remote sensing images. We divide the existing sampling methods into two types: random sampling and saliency-based sampling. Here, we consider the commonly-used grid sampling to be a specific type of random sampling method, and the saliency-based sampling consists of keypoint-based sampling and salient region-based sampling. To compare their performances, we rely on a standard bag-of-visual-words model to learn the global features for testing because of its simplicity, robustness and efficiency. In addition, we conduct experiments using a Fisher kernel framework to validate our conclusions. The experimental results obtained on two commonly-used datasets using different feature learning methods show that random sampling can provide comparable and even better performance than all of the saliency-based strategies.

Foreword to the Special Issue on Information Extraction From High-Spatial-Resolution Optical Remotely Sensed Imagery

The articles in this special section focus on information extraction from high-spatial-resolution optical remotely sensed imagery technology and applications.

Object Detection Based on Sparse Representation and Hough Voting for Optical Remote Sensing Imagery

We present a novel method for detecting instances of an object class or specific object in high-spatial-resolution optical remote sensing images. The proposed method integrates sparse representations for local-feature detection into generalized-Hough-transform object detection. Object parts are detected via class-specific sparse image representations of patches using learned target and background dictionaries, and their co-occurrence is spatially integrated by Hough voting, which enables object detection. We aim to efficiently detect target objects using a small set of positive training samples by matching essential object parts with a target dictionary while the residuals are explained by a background dictionary. Experimental results show that the proposed method achieves state-of-the-art performance for several examples including object-class detection and specific-object identification.

Remote-sensing monitoring for spatio-temporal dynamics of sand dredging activities at Poyang Lake in China

As the largest freshwater lake in China, Poyang Lake is suffering from declining water quality related to the excessive dredging of sand. Field supervision is difficult due to the size of the lake (>3000 km2, wet season) and limited human resources. In this study, an approach is proposed to monitor sand-dredging activities using medium-resolution optical remote-sensing imagery, including 45 Landsat TM/ETM+ images from 2002 to 2012 and 140 HJ1A/B CCD images from 2009 to 2012. The procedure for detecting dredging vessels involves three steps. (1) The entire image is segmented into different homogeneous partitions to overcome water body heterogeneity, and ships in each partition with different levels of water clarity are detected using three types of contrast box architecture. (2) Dredging vessels are then identified based on a spatial overlay analysis of ships and dredging plumes, which are extracted from remote-sensing imagery. (3) False alarms (FAs) of dredging vessels are screened according to the distribution of the sandy lake bed. The results showed significant spatio-temporal variation in dredging activities; sand-dredging activities were concentrated at the northern part of Poyang Lake prior to 2008 and have expanded southwards since 2009. The northern part of the lake experienced persistent dredging operations throughout the year, whereas dredging was observed only during the wet seasons in the southern portion of the lake. A high intensity of illegal dredging was discovered based on two lines of evidence: dredging vessels were detected during the sand-dredging ban, and the estimated quantities of sand dredging were much higher than those planned by the authority. The sediment balance in Poyang Lake has continued to be disrupted, and the lake has become a sediment-exporting system. This study provides an effective solution for monitoring sand-dredging dynamics as well as useful information for managing sand dredging in fresh water environments and assessing its potential impacts on aquatic ecosystems.

An Integrated Method for Urban Main-Road Centerline Extraction From Optical Remotely Sensed Imagery

Road information has a fundamental role in modern society. Road extraction from optical satellite images is an economic and efficient way to obtain and update a transportation database. This paper presents an integrated method to extract urban main-road centerlines from satellite optical images. The proposed method has four main steps. First, general adaptive neighborhood is introduced to implement spectral-spatial classification to segment the images into two categories: road and nonroad groups. Second, road groups and homogeneous property, measured by local Geary's C, are fused to improve road-group accuracy. Third, road shape features are used to extract reliable road segments. Finally, local linear kernel smoothing regression is performed to extract smooth road centerlines. Road networks are then generated using tensor voting. The proposed method is tested and subsequently validated using a large set of multispectral high-resolution images. A comparison with several existing methods shows that the proposed method is more suitable for urban main-road centerline extraction.

Shoreline movement monitoring based on SAR images in Shanghai, China

In this article, a divide-and-combine approach and multiscale normalized cut techniques are combined to extract water–land boundaries from synthetic aperture radar (SAR) images. A validation study shows that waterline extraction from SAR using the proposed algorithm is consistent with that derived from Landsat optical remote-sensing imagery. Analysis of six European Remote Sensing-1/2 (ERS-1/2) satellite and Envisat SAR images acquired at spring tides in Lingang New City, a new land-reclamation city in the suburb of Shanghai, China, between 1993 and 2005 shows the shoreline movement of the study region. We find that the shoreline had minimum seaward movement in the region due to natural forces between 1993 and 1996. Between 1996 and 1999, the shoreline moved more in the eastern and slower in the western part due to artificial impelling siltation and deposition of silt in the Yangtze River. Later, between 1999 and 2005, the shoreline significantly changed following commencement of a tidal flat reclamation project.

The Inylchek Glacier in Kyrgyzstan, Central Asia: Insight on Surface Kinematics from Optical Remote Sensing Imagery

Mountain chains of Central Asia host a large number of glaciated areas that provide critical water supplies to the semi-arid populated foothills and lowlands of this region. Spatio-temporal variations of glacier flows are a key indicator of the impact of climate change on water resources as the glaciers react sensitively to climate. Satellite remote sensing using optical imagery is an efficient method for studying ice-velocity fields on mountain glaciers. In this study, temporal and spatial changes in surface velocity associated with the Inylchek glacier in Kyrgyzstan are investigated. We present a detailed map for the kinematics of the Inylchek glacier obtained by cross-correlation analysis of Landsat images, acquired between 2000 and 2011, and a set of ASTER images covering the time period between 2001 and 2007. Our results indicate a high-velocity region in the elevated part of the glacier, moving up to a rate of about 0.5 m/day. Time series analysis of optical data reveals some annual variations in the mean surface velocity of the Inylchek during 2000–2011. In particular, our findings suggest an opposite trend between periods of the northward glacial flow in Proletarskyi and Zvezdochka glacier, and the rate of westward motion observed for the main stream of the Inylchek.

Road Detection from Optical Remote Sensing Imagery Using Circular Projection Matching and Tracking Strategy

A road detection method based on Circular Projection (CP) matching and tracking strategy is proposed in this paper. The CP matching is a template matching-based algorithm with rotation invariant and low computation characteristics. This advantage is beneficial for twisty roads detection. Take the good use of the advantage, we improve the CP matching to make it more applicable for road detection, including: optimal template radius selection, approximation control parameter, and anti-interference correction for suppressing brightness and random noise affects. The overall road detection is using a dual-directional tracking strategy. In road tracking, we use a fan-shaped scope to limit the direction of template matching to prevent backtracking. In tracking procedure, we propose an adaptive template matching algorithm, overcoming changes between the selected template and targets. Finally, SPOT-5 and GeoEye-1 remote sensing images are used for experiment. The experimental result shows that the proposed method is rotation invariant, flexible and fast, which is applicable to the twisty road detection. The experiment results also show that road detection error is about 2 ~ 5 m, and detection accuracy is higher than 80 %.

Segmentation for High-Resolution Optical Remote Sensing Imagery Using Improved Quadtree and Region Adjacency Graph Technique

An approach based on the improved quadtree structure and region adjacency graph for the segmentation of a high-resolution remote sensing image is proposed in this paper. In order to obtain the initial segmentation results of the image, the image is first iteratively split into quarter sections and the quadtree structure is constructed. In this process, an improved fast calculation method for standard deviation of image is proposed, which significantly increases the speed of quadtree segmentation with standard deviation criterion. A spatial indexing structure was built using improved Morton encoding based on this structure, which provides the merging process with data structure for neighborhood queries. Then, in order to obtain the final segmentation result, we constructed a feature vector using both spectral and texture factors, and proposed an algorithm for region merging based on the region adjacency graph technique. Finally, to validate the method, experiments were performed on GeoEye-1 and IKONOS color images, and the segmentation results were compared with two typical algorithms: multi-resolution segmentation and Mean-Shift segmentation. The experimental results showed that: (1) Compared with multi-resolution and Mean-Shift segmentation, our method increased efficiency by 3–5 times and 10 times, respectively; (2) Compared with the typical algorithms, the new method significantly improved the accuracy of segmentation.

Filling the ‘white ribbon’ – a multisource seamless digital elevation model for Lizard Island, northern Great Barrier Reef

Hydrographers have traditionally referred to the nearshore area as the ‘white ribbon’ area due to the challenges associated with the collection of elevation data (elevation hereafter refers to both topography and bathymetry) in this highly dynamic transitional zone between terrestrial and marine environments. Accordingly, available information in this zone is typically characterized by a range of data sets from disparate sources. In this article, we propose a framework to fill the white ribbon area of a coral reef system by integrating multiple elevation data sets acquired by a suite of remote-sensing technologies into a seamless digital elevation model (DEM). A range of data sets are integrated, including field-collected global positioning system (GPS) elevation points, topographic and bathymetric light detecting and ranging (lidar), single and multibeam echosoundings, nautical charts, and bathymetry derived from optical remote-sensing imagery. The proposed framework ranks data reliability internally, thereby avoiding the requirements to quantify absolute error and results in a high-resolution, seamless product. Nested within this approach is an effective spatially explicit technique for improving the accuracy of bathymetry estimates derived empirically from optical satellite imagery through modelling the spatial structure of residuals. The approach was applied to data collected on and around Lizard Island in northern Australia. Collectively, the framework holds promise for filling the white ribbon zone in coastal areas characterized by similar data availability scenarios.

Radiometric and photometric determinations of simulated shallow-water environment

Optical remote sensing is increasingly becoming a preferred economic alternative to the traditional in situ observations and physical sampling for mapping and monitoring habitats. Submersed habitats, such as eelgrass and corals, are especially challenging for field work. Even for remote-sensing work, a priori knowledge of environmental factors is required for highly accurate analysis. Background illumination and water clarity are two key factors that affect the optical remote-sensing imagery, which may vary widely with season, time of year, geographic location, or water depth. This article presents efforts to simulate natural oceanic conditions in a laboratory setting. Solar radiation predicted at different latitudes under varying water clarity conditions and depth were replicated using a 2.5 m deep wave tank at the University of New Hampshire. The goals of the study were: (1) to simulate illumination and water clarity conditions that approximate coastal and oceanic waters, and (2) to quantify the impact of the simulated illumination and water clarity conditions at different depths on the apparent colours that can be observed from an aerial platform. The empirical radiometric measurements included irradiance, radiance, and remote-sensing reflectance from an underwater array of light sources. The results of the study show good correlation (r 2 = 0.89–0.93) between the natural daylight spectrum at the water surface and the irradiance measurements between 350 nm and 590 nm, at 3.5 m from the light array. The colours of the clear and murky water types were photometrically calculated from the radiometric measurements and validated using underwater video imagery. Using this methodology, illumination and water clarity can be replicated under controlled laboratory conditions and used to assist in studying the physical, chemical, and biological processes in habitats, at varied geographic locations and differing environments.

A General Approach for Suppressing Vegetation in Optical Remotely Sensed Imagery

Vegetation coverage is always an obstacle in the lithological information extraction from optical remotely sensed imagery.Aimed at enhancing lithological and soil information in vegetation coverage area,this paper used a three-step(i.e.,masking,enforced variance,histogram equalization) general vegetation suppression approach.Four experiments were conducted by using different datasets in different locations,which included a Landsat ETM+ image in Hangzhou of Zhejiang,a Landsat ETM+ image in Funing of Yunnan,a Hyperion image in Rongtang of Jiangxi and a MODIS image in Guangdong.The results indicate the effectiveness of the proposed approach in suppressing vegetation in optical remote sensing images with different spatial and spectral resolutions.