A "Reef-Up" approach to classifying coral habitats from IKONOS imagery

Monitoring coral reef benthos with satellites has typically followed a "sensor-down" approach, with the classification algorithm driven by statistics derived from the imagery. I adopt a "reef-up" method, drawing on statistics derived from hyperspectral optical field measurements of substrate reflectance to train image classification. In order to calibrate the satellite data with direct physical measurements of reflectivity, it is necessary to process both the imagery and in situ data to common units of albedo. Building upon a proof-of-concept study conducted by the author in the Red Sea, the link is made by correcting the remote sensing data for the effect of varying bathymetry using in situ measurement of water column optical properties and a digital elevation model constructed from a vessel-based acoustic survey, thereby yielding units of substrate reflectance. Extensive ground verification of the predictive benthic habitat map resulting from image classification showed that eight substrate classes were resolved with an overall accuracy of 69% down to a depth of 6 m, including live and dead coral framework. As compared to conventional from-image classification techniques, the reef-up method offers the potential for higher thematic accuracy while maintaining a greater degree of flexibility for repeat survey using platforms of higher spectral and spatial resolution, expected to come online in the near future. The fact that image acquisition and optical ground-truthing did not occur concurrently, is of particular relevance in confirming that in situ measurements can be made independent of image acquisition and retrospectively linked to appropriate substrate classes. Considering the wealth of hyperspectral data already acquired for shallow reef facies, the work highlights the potential of the reef-up approach for quantifying substrate distribution in coral environments using both air- and spaceborne platforms.     

Integrated spectral and spatial information mining in remote sensing imagery

Most existing remote sensing image retrieval systems allow only simple queries based on sensor, location, and date of image capture. This approach does not permit the efficient retrieval of useful hidden information from large image databases. This paper presents an integrated approach to retrieving spectral and spatial patterns from remotely sensed imagery using state-of-the-art data mining and advanced database technologies. Land cover information corresponding to spectral characteristics is identified by supervised classification based on support vector machines with automatic model selection, while textural features characterizing spatial information are extracted using Gabor wavelet coefficients. Within identified land cover categories, textural features are clustered to acquire search-efficient space in an object-oriented database with associated images in an image database. Interesting patterns are then retrieved using a query-by-example approach. The evaluation of the study results using coverage and novelty measures validates the effectiveness of the proposed remote sensing image information mining framework, which is potentially useful for applications such as agricultural and environmental monitoring.     

Supervised classification of K-distributed SAR images of naturaltargets and probability of error estimation

A radiometric and textural classification method for the single-channel synthetic aperture radar (SAR) image is proposed, which explicitly takes into account the probability density function (pdf) of the underlying cross section for K-distributed images. This method makes extensive use of adaptive preprocessing methods (e.g. Gamma-Gamma MAP speckle filtering), in order to ensure good classification accuracy as well as fair preservation of the spatial resolution of the final result. Error rates can be estimated during the training step, allowing one to select only relevant reflectivity classes and to save computation time in trials. The classification method is based on a maximum likelihood (ML) segmentation of the filtered image. Finally, a texture criterion is introduced to improve the accuracy of the classification result     

Forest canopy closure from classification and spectral unmixing ofscene components-multisensor evaluation of an open canopy

Three types of remote sensing data, color infrared aerial photography (CIR), compact airborne spectrographic imager (CASI) imagery, and airborne visible/infrared imaging spectrometer (AVIRIS) imagery, have been used to estimate forest canopy closure for an open-canopy forest environment. The high-spatial-resolution CIR and CASI images were classified to generate forest canopy closure estimates. These estimates were used to validate the forest canopy closure estimation accuracy obtained using the AVIRIS image. Reflectance spectra extracted from the spectral-mode CASI image were used to normalize the raw AVIRIS image to a reflectance image. Classification and spectral unmixing methods have been applied to the AVIRIS image. Results indicate that both the classification and the spectral unmixing methods can produce reasonably accurate estimates of forest canopy closure (within 3 percent agreement) when related statistics are extracted from the AVIRIS image and relatively pure reflectance spectra are extracted from the CASI image. However, it is more challenging to use the spectral unmixing technique to derive subpixel-scale components whose reflectance spectra cannot be directly extracted from the AVIRIS image     

Effect of spatial resolution on classification errors of pure andmixed pixels in remote sensing

It is observed in remote sensing that a finer spatial resolution does not necessarily improve the classification performance. These observations have been understood by using the conceptual explanation that "boundary effect" and "within-class variability" work against one another. Though easily understood, this conceptual explanation cannot be readily used for a quantitative investigation. The authors design a simulation scheme to evaluate systematically the impacts of various parameters on the classification accuracy. The authors employ a model for the class spectral covariance of pure pixels and a linear mixing model for the spectral responses of mixed pixels. Based on these models, the authors derive the statistical characteristics for mixed pixels and assess the corresponding classification errors. As the ratio of ground sampling distance to field size decreases, the classification error associated with pure pixels tends to increase, whereas the classification error associated with mixed pixels tends to decrease from the smaller area of mixed pixels. The simulation results show that the overall classification error first decreases with decreasing ratio of ground sampling distance to field width, reaches a minimum value, and then may increase with further decreasing ratio. The study on the classification error may help the development of classification schemes for high spatial resolution imagery     

Resolution Enhancement of Multilook Imagery for the Multispectral Thermal Imager

This paper studies the feasibility of enhancing the spatial resolution of multilook Multispectral Thermal Imager (MTI) imagery using an iterative resolution enhancement algorithm known as Projection Onto Convex Sets (POCS). A multiangle satellite image modeling tool is implemented, and simulated multilook MTI imagery is formed to test the resolution enhancement algorithm. Experiments are done to determine the optimal configuration and number of multiangle low-resolution images needed for a quantitative improvement in the spatial resolution of the high-resolution estimate. The issues of atmospheric path radiance and directional reflectance variations are explored to determine their effect on the resolution enhancement performance.     

Super-resolution target identification from remotely sensed imagesusing a Hopfield neural network

Fuzzy classification techniques have been developed recently to estimate the class composition of image pixels, but their output provides no indication of how these classes are distributed spatially within the instantaneous field of view represented by the pixel. As such, while the accuracy of land cover target identification has been improved using fuzzy classification, it remains for robust techniques that provide better spatial representation of land cover to be developed. Such techniques could provide more accurate land cover metrics for determining social or environmental policy, for example. The use of a Hopfield neural network to map the spatial distribution of classes more reliably using prior information of pixel composition determined from fuzzy classification was investigated. An approach was adopted that used the output from a fuzzy classification to constrain a Hopfield neural network formulated as an energy minimization tool. The network converges to a minimum of an energy function, defined as a goal and several constraints. Extracting the spatial distribution of target class components within each pixel was, therefore, formulated as a constraint satisfaction problem with an optimal solution determined by the minimum of the energy function. This energy minimum represents a “best guess” map of the spatial distribution of class components in each pixel. The technique was applied to both synthetic and simulated Landsat TM imagery, and the resultant maps provided an accurate and improved representation of the land covers studied, with root mean square errors (RMSEs) for Landsat imagery of the order of 0.09 pixels in the new fine resolution image recorded     

一种高分辨率遥感影像分类的特征指数

随着影像分辨率的提高,传统的光谱特征不能有效地描述复杂的高分辨率影像信息,从而影响高分辨率遥感影像的分类。为了弥补传统光谱方法的不足,提出了一种加权对象相关指数(WOCI),并将其应用到基于支持向量机(SVM)的影像分类中。该指数是通过考虑具有相似性光谱的对象来构建的,可全面地描述影像的上下文结构。结果表明与仅考虑光谱特征和像素空间特征进行分类的方法相比,基于WOCI特征的分类结果有更高的精确性,且分类精度提高了7.16%。     

Geometric accuracy of Ikonos: zoom in

Pioneering in the world, the high-resolution satellite Ikonos imagery has revolutionary changed the market - users now have the unique opportunity to get satellite images with spatial resolution comparable to middle-scale aerial photos. In our study, we tried to find an answer why such images, having submetre spatial resolution, still contain planimetric errors measured in dozens and even hundreds of metres. This paper is aimed at to not only show to potential users of high-resolution satellite images the main sources of image geometric distortions, but to uncover the errors, hardly to be corrected without possessing precise information about particular conditions on the spot at the moment of image acquisition, and evaluate sensibility of the corresponding image correction models for variations of such data.     

基于卷积神经网络的高分遥感影像单木树种分类

树种调查一直面临着成本高、效率低、精度不高等问题。利用遥感手段能大大提高树种类型调查的工作效率、节省成本;卷积神经网络(CNN)虽然已经在自然图像分类领域取得了许多突破,但是较少有人将CNN模型用于单木树种分类。基于上述考虑,搭建出CNN模型,并与高分遥感影像相结合,进行单木树种分类。在利用高分影像半自动化构建单木树种遥感影像样本集过程中,采用了影像冠层切片(CSI)圈定、人工标注、数据增强等方法;同时为了训练单木树种遥感影像样本集,对5个CNN模型进行针对性改写。通过对比分析发现:LeNet5_relu和AlexNet_mini都未取得最佳分类效果;GoogLeNet_mini56、ResNet_mini56和DenseNet_BC_mini56分别对不同的树种具有最佳分类效果;DenseNet_BC_mini56总体精度最高(94.14%),Kappa系数最高(0.90),是总体最佳分类模型。该研究证明了CNN在单木树种分类中的有效性,能为森林资源调查提供重要的解决方案。     

Estimation of the Number of Decomposition Levels for a Wavelet-Based Multiresolution Multisensor Image Fusion

The wavelet-based scheme for the fusion of multispectral (MS) and panchromatic (PAN) imagery has become quite popular due to its ability to preserve the spectral fidelity of the MS imagery while improving its spatial quality. This is important if the resultant imagery is used for automatic classification. Wavelet-based fusion results depend on the number of decomposition levels applied in the wavelet transform. Too few decomposition levels result in poor spatial quality fused images. On the other hand, too many levels reduce the spectral similarity between the original MS and the pan-sharpened images. If the shift-invariant wavelet transform is applied, each excessive decomposition level results in a large computational penalty. Thus, the choice of the number of decomposition levels is significant. In this paper, PAN and MS image pairs with different resolution ratios were fused using the shift-invariant wavelet transform, and the optimal decomposition levels were determined for each resolution ratio. In general, it can be said that the fusion of images with larger resolution ratios requires a higher number of decomposition levels. This paper provides the practitioner an understanding of the tradeoffs associated with the computational demand and the spatial and spectral quality of the wavelet-based fusion algorithm as a function of the number of decomposition levels     

Comparative study of semi-implicit schemes for nonlinear diffusion in hyperspectral imagery.

Nonlinear diffusion has been successfully employed over the past two decades to enhance images by reducing undesirable intensity variability within the objects in the image, while enhancing the contrast of the boundaries (edges) in scalar and, more recently, in vector-valued images, such as color, multispectral, and hyperspectral imagery. In this paper, we show that nonlinear diffusion can improve the classification accuracy of hyperspectral imagery by reducing the spatial and spectral variability of the image, while preserving the boundaries of the objects. We also show that semi-implicit schemes can speedup significantly the evolution of the nonlinear diffusion equation with respect to traditional explicit schemes.     

An Innovative Neural-Net Method to Detect Temporal Changes in High-Resolution Optical Satellite Imagery

The advent of new high spatial resolution optical satellite imagery has greatly increased our ability to monitor land cover changes from space. Satellite observations are carried out regularly and continuously, and provide a great deal of insight into the temporal changes of land cover use. High spatial resolution imagery better resolves the details of these changes and makes it possible to overcome the "mixed-pixel" problem that is inherent with more moderate resolution satellite sensors. At the same time, high-resolution imagery presents a new challenge over other satellite systems, in that a relatively large amount of data must be analyzed and corrected for registration and classification errors to identify the land cover changes. To obtain the accuracies that are required by many applications to large areas, very extensive manual work is commonly required to remove the classification errors that are introduced by most methods. To improve on this situation, we have developed a new method for land surface change detection that greatly reduces the human effort that is needed to remove the errors that occur with many classification methods that are applied to high-resolution imagery. This change detection algorithm is based on neural networks, and it is able to exploit in parallel both the multiband and the multitemporal data to discriminate between real changes and false alarms. In general, the classification errors are reduced by a factor of 2-3 using our new method over a simple postclassification comparison based on a neural-network classification of the same images.     

Use of Remotely Sensed Data for Assessing Forest Stand Conditions in the Eastern United States

The results of three interrelated research activities conducted by Goddard scientists in support of the AgRISTARS Renewable Resources Inventory (RRI) project are summarized. The central theme of the research conducted at Goddard was the development of techniques for the detection, classification, and measurement of forest disturbances using digital, remotely sensed data. Three study areas located in Pennsylvania, North Carolina, and Maine were investigated with respect to: a) the delineation and assessment of forest damage associated with two different forest insect defoliators, and b) an assessment of the improved capabilities to be expected from Landsat Thematic Mapper (TM) data relative to Multispectral Scanner (MSS) data for delineating forest stand characteristics. Key results include the development of a statewide MSS digital data base and associated image processing techniques for accurately delineating (approximately 90 precent correct classification accuracy) insect damaged and healthy forest. Comparison of analyses using MSS and TM Simulator (TMS) data indicated that for broad land cover classes which are spectrally homogeneous, the accuracy of the classification results are similar. However, TMS data provided superior results (20 percent overall accuracy increase relative to MSS results) when detailed (Level III) forest classes were mapped. These studies also illustrated the utility of having at least one band in the visible, near infrared, and middle infrared portion of the electromagnetic spectrum for assessing specific (Level III) forest cover types.     

基于流形学习的新高光谱图像降维算法

提出了一种新的基于图像块距离的邻域选择方法,并将其应用于流形学习中,得到一类新的高光谱图像非线性降维算法。该类算法利用高光谱图像物理特性,结合图像的光谱信息和空间信息,在最大限度减小图像信息冗余的基础之上,很好地保持了原始数据集的特性。与其它高光谱图像的降维算法相比,改进的流形学习算法不仅考虑到高光谱图像本身的空间关系,而且利用图像块距离更好地保持了数据点之间的局部特性,从而有效地去除原始数据集光谱维和空间维的冗余信息。实际高光谱数据的实验结果表明,所提出的算法在应用于高光谱图像分类时,与其它方法相比具有更高的分类精度。     

A Markov random field model for classification of multisourcesatellite imagery

A general model for multisource classification of remotely sensed data based on Markov random fields (MRF) is proposed. A specific model for fusion of optical images, synthetic aperture radar (SAR) images, and GIS (geographic information systems) ground cover data is presented in detail and tested. The MRF model exploits spatial class dependencies (spatial context) between neighboring pixels in an image, and temporal class dependencies between different images of the same scene. By including the temporal aspect of the data, the proposed model is suitable for detection of class changes between the acquisition dates of different images. The performance of the proposed model is investigated by fusing Landsat TM images, multitemporal ERS-1 SAR images, and GIS ground-cover maps for land-use classification, and on agricultural crop classification based on Landsat TM images, multipolarization SAR images, and GIS crop field border maps. The performance of the MRF model is compared to a simpler reference fusion model. On an average, the MRF model results in slightly higher (2%) classification accuracy when the same data is used as input to the two models. When GIS field border data is included in the MRF model, the classification accuracy of the MRF model improves by 8%. For change detection in agricultural areas, 75% of the actual class changes are detected by the MRF model, compared to 62% for the reference model. Based on the well-founded theoretical basis of Markov random field models for classification tasks and the encouraging experimental results in our small-scale study, the authors conclude that the proposed MRF model is useful for classification of multisource satellite imagery     

Classifying land development in high-resolution Satellite imagery using hybrid structural-multispectral features

It is well known that combining spatial and spectral information can improve land use classification from satellite imagery. Human activity on the ground, such as construction, induces changes in both the photometric structure of the image and in its spectral content owing to, primarily, changes in vegetation density and surface materials. This paper introduces a novel approach to combine spatial (more precisely, structural) information extracted from (1-m resolution) panchromatic Ikonos imagery with the multispectral response (4-m resolution) available from the same sensor. Of the prior work combining spatial and spectral information, none has extracted structural features as we do, and none has combined these information sources as early in the process. The classifier we describe here, discriminating urban and rural regions, is a front-end component of a fairly complete satellite image analysis system that identifies suburban residential areas and extracts their street networks and single-family houses. We extract structural information in the form of photometric straight lines and their spatial arrangement over relatively small neighborhoods. To capture the multispectral information, we turn to the well-known normalized difference vegetation index (NDVI) and an improved linearized version of our own development (details of the structural analysis and the theoretical development of the linearized NDVI appear elsewhere). This paper addresses the novel combination of these types of features (hybrids) by using the structural features, straight line support regions based on gradient orientation, as cue regions for multispectral analysis. We test the hybrid features in a range of parametric and nonparametric classifiers. We also implement and test a probabilistic relaxation algorithm followed by the maximum a priori decision rule. We report extensive results that indicate significant improvements in classification accuracy using the hybrid features.     

Multi-manifold-based skin classifier on feature space Voronoï regions for skin segmentation

Skin segmentation is a crucial and a challenging step in many face and gesture recognition techniques and it has various applications in human computer interaction, objectionable content filtering, image retrieval and many more. In this article, we propose a novel skin segmentation method, which uses multi-manifold-based skin classification of feature space skin candidate Voronoï regions to achieve accurate skin segmentation. The state-of-the-art skin segmentation techniques reported in this article focus on discrimination between textural feature vectors belonging to skin and non-skin classes. In contrast, the proposed method focuses on discrimination between textural feature vectors belonging to skin and skin-like (non-skin) classes, which lead to higher skin classification accuracy. Furthermore, we introduce a novel image segmentation technique based on spatial and feature space Dirichlet tessellation (also called a Voronoï diagram) to achieve feature space segmentation of skin candidate regions of an image. These feature space segments will then be classified using a multi-manifold-based skin classifier. The proposed skin segmentation method was evaluated on two benchmark skin segmentation data sets and its results were compared with four other state-of-the-art methods proposed for skin segmentation. The experimental results reported in this article confirm that the proposed method outperforms the existing skin segmentation approaches in terms of false alarm rates in the skin segmentation process. Also, the proposed method results in the lowest minimal detection error compared to the existing methods reported in this article.     

Efficient Texture Analysis of SAR Imagery

We address the problem of efficiency in texture analysis for synthetic aperture radar (SAR) imagery. Motivated by the statistical occupancy model, we introduce the notion of patch reoccurrences. Using the reoccurrences, we propose the use of approximate textural features in analysis of SAR images. We describe how the proposed approximate features can be extracted for two popular texture analysis methods—the gray-level cooccurrence matrix and Gabor wavelets. Results on image texture classification show that the proposed method can provide an improved efficiency in the analysis of SAR imagery, without introducing any significant degradation in the classification results.     

Enhanced SAR-Based Snow-Covered Area Estimation Method for Boreal Forest Zone

In this paper, an enhanced method for fractional snow-covered area (SCA) estimation for the boreal forest zone is presented. The new approach, based on utilizing weather station data alongside with spaceborne synthetic aperture radar (SAR) imagery, leads to a significantly improved estimation accuracy. While the Helsinki University of Technology (TKK) SAR-based SCA estimation method serves as a basic tool in the SCA estimation, the ground-based weather station observations are employed to still strengthen its performance at the nearly melt-off or totally melt-off conditions. The method is still improved by a new reference image selection process, leading to more accurate results and an easier adaptivity to new areas. The SCA estimation accuracy of the new enhanced method is compared with optical satellite-based SCA data. Evaluation of the method is carried out using Radarsat wide-swath data for the snow-melt seasons of 2004–2006. The results show a significant increase in accuracy when the enhanced SCA method is applied. Correlation between the radar-based and optical comparison data increases from 0.914 to 0.947 and root-mean-square error improves from 0.151 to 0.123 with the new method. Traditionally, the TKK method has provided SCA estimates for Finnish third-order subdrainage basins. In this paper, the method is adapted to produce SCA estimates also in 5 $times$ 5 km spatial resolution. The analyses for the 5 $times$ 5 km method indicate poorer estimation accuracy than the nominal drainage-basin-based method.