One of the goals of the government-sponsored R&D program is to develop next generation algorithms to discriminate various types of battlefield ordnance in near real-time. Applications that could utilize this capab...
详细信息
ISBN:
(纸本)0780358465
One of the goals of the government-sponsored R&D program is to develop next generation algorithms to discriminate various types of battlefield ordnance in near real-time. Applications that could utilize this capability include early indication and warning of threats, support of battle damage assessment (BDA), level of conflict (LOC) assessment, and intelligence preparation of the battlefield (IPB). As part of this effort, we have previously investigated and reported on the performance of several classification algorithms applied to electro-optical data collected by a ground-based sensor[13]. That study included evaluation of our baseline algorithm OSCAR: Ordnance Statistical Classification And recognition. This paper discusses enhancements made to the algorithm over the last year and evaluates algorithm performance as applied to data obtained from remote assets where remote assets may be ground-, air-,or space-based. This remotely collected data has a larger noise component and higher intra-class variances than the ground-collected data, adding new challenges to the discrimination problem. Enhancements that we have made to the algorithm this year include 1) feature-based processing, 2) rejection of feature vectors from unknown classes, 3) addition of a confidence level in each classification result, 4) handling of multispectral data, and 5) handling of multiple input file formats. Enhancements we have made to the algorithm development workbench include analysis tools for displaying the feature space, the rotated feature space (via Principal Components Analysis (PCA)), and class boundaries/probability contours. These tools help the developer to understand the algorithm performance in insightful ways and help analyze class separability for various features, reveal why specific sample vectors get misclassified, highlight the normality of the data, identify data outliers. etc. Algorithm performance is evaluated for both broad and fine classes. A broad class is defined a
Clustering has been widely used in areas as patternrecognition, data analysis and imageprocessing. Recently, clustering algorithms have been recognized as one of a powerful tool for data mining. However, the well-kn...
详细信息
Clustering has been widely used in areas as patternrecognition, data analysis and imageprocessing. Recently, clustering algorithms have been recognized as one of a powerful tool for data mining. However, the well-known clustering algorithms offer no solution to the case of large mixed incomplete data sets. The authors comment the possibilities of application of the methods, techniques and philosophy of the logical combinatorial approach for clustering in these kinds of data sets. They present the new clustering algorithm DGLC for discovering /spl beta//sub 0/-density connected components from large mixed incomplete data sets. This algorithm combines the ideas of logical combinatorial patternrecognition with the density based notion of cluster. Finally, an example is showed in order to illustrate the work of the algorithm.
A methodology based on mathematical morphology to classify forest cover types in remotesensingimages is presented. The information automatically extracted at higher scales (aerial photographs) by morphological segme...
详细信息
ISBN:
(纸本)0769507506
A methodology based on mathematical morphology to classify forest cover types in remotesensingimages is presented. The information automatically extracted at higher scales (aerial photographs) by morphological segmentation approaches is afterwards used to classify different forest cover types at lower scales (satellite images). In this methodology the spectral process is guided by the spatial process, once the previous segmentation of the different textural elements is then used in the classification procedure, where the geometrical modelling of the shape of the training sets of points is also performed. Tests were done in a region of centre Portugal using aerial photographs and Landsat TM images for olive, cork oak, pine and eucalyptus trees.
An optimal line detector for the one-dimensional case is derived from Canny's criteria (1986). The detector is extended to the two-dimensional case by operating separately in the x and y directions. An efficient i...
详细信息
ISBN:
(纸本)0769507506
An optimal line detector for the one-dimensional case is derived from Canny's criteria (1986). The detector is extended to the two-dimensional case by operating separately in the x and y directions. An efficient implementation using an infinite impulse response (IIR) filter is provided. This implementation has an additional advantage that increasing the filter scale affects neither temporal nor spatial complexity. Our detector is faster than the Gaussian used by Steger (1998); e.g., when the scale is 3 our detector is 33 times faster. Experimental results using real images demonstrate the validity of the algorithm.
In the field of patternrecognition, the combination of an ensemble of neural networks has been proposed as an approach to the development of high performance image classification systems. However, previous work clear...
详细信息
In the field of patternrecognition, the combination of an ensemble of neural networks has been proposed as an approach to the development of high performance image classification systems. However, previous work clearly showed that such image classification systems are effective only if the neural networks forming them make different errors. Therefore, the fundamental need for methods aimed to design ensembles of "error-independent" networks is currently acknowledged. In this paper, an approach to the automatic design of effective neural network ensembles is proposed. Given an initial large set of neural networks, our approach is aimed to select the subset formed by the most error-independent nets. Reported results on the classification of multisensor remote-sensingimages show that this approach allows one to design effective neural network ensembles.
The dominant architecture for mobile robot perception uses sensors on-board the robot, providing a first-person perspective on the environment. We demonstrate a novel mobile robot architecture that uses an environment...
详细信息
The dominant architecture for mobile robot perception uses sensors on-board the robot, providing a first-person perspective on the environment. We demonstrate a novel mobile robot architecture that uses an environment-based sensor network, which provides third-person perception. The idea is that a mobile robot working in the area tunes in to broadcasts from the video camera network (or in this case from an environment-based computer processing the video frames) to receive sensor data. This distributed sensing configuration offers several advantages over on-board sensing.
The field of wavelets has opened up new opportunities for the compression of satellite sensory imagery. The paper examines the influence of wavelet compression on the automatic classification of urban environments. Ai...
详细信息
ISBN:
(纸本)0769507506
The field of wavelets has opened up new opportunities for the compression of satellite sensory imagery. The paper examines the influence of wavelet compression on the automatic classification of urban environments. Airborne laser scanning data is introduced as an additional channel along-side the spectral channels of colour infrared imagery. This effectively integrates the local height and multi-spectral information sources. To incorporate context information, the feature base is expanded to include both spectral and non-spectral features. A maximum likelihood classification approach is then applied. It is demonstrated that the classification of urban scenes is considerably improved by fusing multi-spectral and geometric data sets. The fused imagery is then systematically compressed (channel by channel) at compression rates ranging from 5 to 100 using a wavelet-based algorithm. The compressed imagery is then classified using the approach described here-above. Analysis of the results obtained indicates that a compression rate of up to 20 can conveniently be employed without adversely affecting the segmentation results.
The problem of selecting an appropriate wavelet filter is always present in the wavelet based compression. Different mother wavelets are characterized by their regularity, which describes the smoothness of the wavelet...
详细信息
ISBN:
(纸本)0769507506
The problem of selecting an appropriate wavelet filter is always present in the wavelet based compression. Different mother wavelets are characterized by their regularity, which describes the smoothness of the wavelet. Digital signals should be characterized similarly to enable the selection of a good wavelet filter. In this paper certain features and cooccurrence matrix are used in characterizing the spectra. Bayesian classification is used to classify the spectra into the classes defined by the best wavelet filter obtained from the compression of the training spectra. A training set is obtained from three multispectral images. The results show, that our method gives the correct result in wavelet filter selection for multispectral image compression.
This study presents a theoretical investigation of the rank-based multiple classifier decision problem for closed-set pattern classification. The case with classifier raw outputs in the form of candidate class ranking...
详细信息
ISBN:
(纸本)0769507506
This study presents a theoretical investigation of the rank-based multiple classifier decision problem for closed-set pattern classification. The case with classifier raw outputs in the form of candidate class rankings is considered and formulated as a discrete optimization problem with the objective function being the total probability of correct decision. The problem has a global optimum solution but is of prohibitive dimensionality. We present a partitioning formalism under which this dimensionality can be reduced by incorporating our prior knowledge about the problem domain and the structure of the training data. The formalism can effectively explain a number of rank-based combination approaches successfully used in the literature, one of which is discussed.
In this work, we investigate on the design of filter banks that allow to substitute part of the spectrum of one signal with that of another signal. One application of this technique is fusion of data collected by sens...
详细信息
In this work, we investigate on the design of filter banks that allow to substitute part of the spectrum of one signal with that of another signal. One application of this technique is fusion of data collected by sensors having different resolutions, that is a a typical problem encountered in remotesensing, when data from low-resolution multi-spectral sensors and high-resolution panchromatic sensors are to be merged, either to alleviate visual identification tasks, or to expedite automatic detection and recognition. The approach that is proposed here is based on the use of cosine-modulated uniform filter banks. We assume that the ratio of the sampling periods of the input data is not integer and show how to design the filter banks so that spectra from different signals can be integrated with minimum distortion.
暂无评论