James C. Tilton

Advances in Spectral-Spatial Classification of Hyperspectral Images

Recent advances in spectral-spatial classification of hyperspectral images are presented in this paper. Several techniques are investigated for combining both spatial and spectral information. Spatial information is extracted at the object (set of pixels) level rather than at the conventional pixel level. Mathematical morphology is first used to derive the morphological profile of the image, which includes characteristics about the size, orientation, and contrast of the spatial structures present in the image. Then, the morphological neighborhood is defined and used to derive additional features for classification. Classification is performed with support vector machines (SVMs) using the available spectral information and the extracted spatial information. Spatial postprocessing is next investigated to build more homogeneous and spatially consistent thematic maps. To that end, three presegmentation techniques are applied to define regions that are used to regularize the preliminary pixel-wise thematic map. Finally, a multiple-classifier (MC) system is defined to produce relevant markers that are exploited to segment the hyperspectral image with the minimum spanning forest algorithm. Experimental results conducted on three real hyperspectral images with different spatial and spectral resolutions and corresponding to various contexts are presented. They highlight the importance of spectral-spatial strategies for the accurate classification of hyperspectral images and validate the proposed methods.

Multiple Spectral–Spatial Classification Approach for Hyperspectral Data

A new multiple-classifier approach for spectral-spatial classification of hyperspectral images is proposed. Several classifiers are used independently to classify an image. For every pixel, if all the classifiers have assigned this pixel to the same class, the pixel is kept as a marker, i.e., a seed of the spatial region with a corresponding class label. We propose to use spectral-spatial classifiers at the preliminary step of the marker-selection procedure, each of them combining the results of a pixelwise classification and a segmentation map. Different segmentation methods based on dissimilar principles lead to different classification results. Furthermore, a minimum spanning forest is built, where each tree is rooted on a classification-driven marker and forms a region in the spectral-spatial classification map. Experimental results are presented for two hyperspectral airborne images. The proposed method significantly improves classification accuracies when compared with previously proposed classification techniques.

Learning bayesian classifiers for scene classification with a visual grammar

A challenging problem in image content extraction and classification is building a system that automatically learns high-level semantic interpretations of images. We describe a Bayesian framework for a visual grammar that aims to reduce the gap between low-level features and high-level user semantics. Our approach includes modeling image pixels using automatic fusion of their spectral, textural, and other ancillary attributes; segmentation of image regions using an iterative split-and-merge algorithm; and representing scenes by decomposing them into prototype regions and modeling the interactions between these regions in terms of their spatial relationships. Naive Bayes classifiers are used in the learning of models for region segmentation and classification using positive and negative examples for user-defined semantic land cover labels. The system also automatically learns representative region groups that can distinguish different scenes and builds visual grammar models. Experiments using Landsat scenes show that the visual grammar enables creation of high-level classes that cannot be modeled by individual pixels or regions. Furthermore, learning of the classifiers requires only a few training examples.

Refining image segmentation by integration of edge and region data

A basic requirement for understanding the dynamics of the Earths major ecosystems is accurate quantitative information about the distribution and areal extent of the Earths vegetation formations. Some of this required information can be obtained through the analysis of remotely sensed data. Image segmentation is often one of the first steps of this analysis. This paper focuses on two particular types of segmentation: region-based and edge-based segmentations. Each approach is affected differently by various factors, and both types of segmentations may be improved by taking advantage of their complementary nature. Included among region-based segmentation approaches are region growing methods, which produce hierarchical segmentations of images from finer to coarser resolution. In this hierarchy, an ideal segmentation (ideal for a given application) does not always correspond to one single iteration, but map correspond to several different iterations. This, among other factors, makes it somewhat difficult to choose a stopping criterion for region growing methods. To find the ideal segmentation, the authors develop a stopping criterion for their Iterative Parallel Region Growing (IPRG) algorithm using additional information from edge features, and the Hausdorff distance metric. They integrate information from regions and edges at the symbol level, taking advantage of the hierarchical structure of the region segmentation results. Also, to demonstrate the feasibility of this approach in processing the massive amount of data that will be generated by future Earth remote sensing missions, such as the Earth Observing System (EOS), all the different steps of this algorithm have been implemented on a massively parallel processor. >

Best Merge Region-Growing Segmentation With Integrated Nonadjacent Region Object Aggregation

Best merge region growing normally produces segmentations with closed connected region objects. Recognizing that spectrally similar objects often appear in spatially separate locations, we present an approach for tightly integrating best merge region growing with nonadjacent region object aggregation, which we call hierarchical segmentation or HSeg. However, the original implementation of nonadjacent region object aggregation in HSeg required excessive computing time even for moderately sized images because of the required intercomparison of each region with all other regions. This problem was previously addressed by a recursive approximation of HSeg, called RHSeg. In this paper, we introduce a refined implementation of nonadjacent region object aggregation in HSeg that reduces the computational requirements of HSeg without resorting to the recursive approximation. In this refinement, HSegs region intercomparisons among nonadjacent regions are limited to regions of a dynamically determined minimum size. We show that this refined version of HSeg can process moderately sized images in about the same amount of time as RHSeg incorporating the original HSeg. Nonetheless, RHSeg is still required for processing very large images due to its lower computer memory requirements and amenability to parallel processing. We then note a limitation of RHSeg with the original HSeg for high spatial resolution images and show how incorporating the refined HSeg into RHSeg overcomes this limitation. The quality of the image segmentations produced by the refined HSeg is then compared with other available best merge segmentation approaches. Finally, we comment on the unique nature of the hierarchical segmentations produced by HSeg.

Image segmentation by region growing and spectral clustering with a natural convergence criterion

The image segmentation approach described is a new hybrid of region growing and spectral clustering. This approach produces a specified number of hierarchical segmentations at different levels of detail, based upon jumps in a dissimilarity criterion. A recursive implementation of this segmentation approach on a cluster of 66 Pentium Pro PCs is described, and the effectiveness of this segmentation approach on Landsat Multispectral Scanner data is discussed.

Contextual classification of multispectral image data

Abstract Compound decision theory is invoked to develop a method for classifying image data using spatial context. Methods for characterizing contextual information in an image are proposed and tested. Experimental results based on both simulated and real multispectral remote sensing data demonstrate the effectiveness of the contextual classifier. A number of practical problems associated with this approach are discussed and possible solutions are explored.

A Marker-Based Approach for the Automated Selection of a Single Segmentation From a Hierarchical Set of Image Segmentations

The Hierarchical SEGmentation (HSEG) algorithm, which combines region object finding with region object clustering, has given good performances for multi- and hyperspectral image analysis. This technique produces at its output a hierarchical set of image segmentations. The automated selection of a single segmentation level is often necessary. We propose and investigate the use of automatically selected markers for this purpose. In this paper, a novel Marker-based HSEG (M-HSEG) method for spectral-spatial classification of hyperspectral images is proposed. Two classification-based approaches for automatic marker selection are adapted and compared for this purpose. Then, a novel constrained marker-based HSEG algorithm is applied, resulting in a spectral-spatial classification map. Three different implementations of the M-HSEG method are proposed and their performances in terms of classification accuracies are compared. The experimental results, presented for three hyperspectral airborne images, demonstrate that the proposed approach yields accurate segmentation and classification maps, and thus is attractive for remote sensing image analysis.

Image Segmentation By Iterative Parallel Region Growing And Splitting

The spatially constrained clustering (SCC) iterative parallel region-growing technique is applied to image analysis. The SCC algorithm is implemented on the massively parallel processor at NASA Goddard. Most previous region-growing approaches have the drawback that the segmentation produced depends on the order in which portions of the image are processed. The ideal solution to this problem (merging only the single most similar pair of spatially adjacent regions in the image in each iteration) becomes impractical except for very small images, even on a massively parallel computer. The SCC algorithm overcomes these problems by performing, in parallel, the best merge within each of a set of local, possibly overlapping, subimages. A region-splitting stage is also incorporated into the algorithm, but experiments show that region splitting generally does not improve segmentation results. The SCC algorithm has been tested on various imagery data, and test results for a Landsat TM image are summarized.

Analysis of hierarchically related image segmentations

Describes an approach for producing high quality hierarchically related image segmentations and some first steps towards exploiting the information content of the segmentation hierarchy. Hierarchically related image segmentations are a set of image segmentations at different levels of detail in which the less detailed segmentations can be produced from specific merges of regions contained in the more detailed segmentations. After a general overview of other approaches to image segmentation, the Hierarchical Segmentation (HSEG) algorithm is presented, along with its recursive formulation (RHSEG). Finally, an approach is outlined for exploiting the information content from the segmentation hierarchy based on changes in region features from one hierarchical level to the next. Comparative results are presented with Landsat Thematic Mapper (TM) data.