Ralph Bernstein

Digital image processing of earth observation sensor data

This paper describes digital image processing techniques that were developed to precisely correct Landsat multispectral Earth observation data and gives illustrations of the results achieved, e.g., geometric corrections with an error of less than one picture element, a relative error of one-fourth picture element, and no radiometric error effect. Techniques for enhancing the sensor data, digitally mosaicking multiple scenes, and extracting information are also illustrated.

Gaussian Maximum Likelihood and Contextual Classification Algorithms for Multicrop Classification

In this paper we review some of the ways in which context has been handled in the remote-sensing literature, and we introduce additional possibilities. The problem of computing exhaustive and normalized class-membership probabilities from the likelihoods provided by the Gaussian maximum likelihood classifier (to be used as initial probability estimates to start relaxation) is discussed. An efficient implementation of probabilistic relaxation is proposed, suiting the needs of actual remote-sensing applications. A modified fuzzy-relaxation algorithm using generalized operations between fuzzy sets is presented. Combined use of the two relaxation algorithms is proposed to exploit context in multispectral classification of remotely sensed data. Results on both one artificially created image and one MSS data set are reported.

Combined Surface Display and Reformatting for the Three-Dimensional Analysis of Tomographic Data

Radiologic examinations increasingly are based on sequences of cross-sectional images. In current clinical applications, the three-dimensional (3D) relationships contained in these examinations must be inferred by the observer through analysis of multiple two-dimensional (2D) images. In this article, methods for the direct display of 3D gray-level data are investigated. In the chosen approach, the 3D presentation of bone and skin surface serves to orient the viewer, while planar reformation and/or transparent projections can be applied for the assessment of soft-tissue structures in regions of interest. The resulting images represent the original image data in a way that is more suitable for observation of 3D relationships than the conventional cross-sectional viewing mode. This may facilitate the diagnostic process and enhance the interpretability of the images. Routine clinical application of this technique requires special computer hardware. Research applications, however, can be performed within tolerable times (10-30 sec/view) with computers found in radiologic research environments.

Analysis And Processing of LANDSAT-4 Sensor Data Using Advanced Image Processing Techniques And Technologies

Techniques have been developed or improved to calibrate, repair, geometrically correct, and extract information from Landsat-4 satellite data. Statistical techniques to correct data radiometry have been evaluated and have minimized striping and banding. It is shown that unless these statistical techniques are used, striping will result even with perfect calibration parameters. Algorithms have been developed to replace data from failed detectors and to reduce coherent noise. The Landsat-4 data have been geometrically corrected to conform to a 1:100 000 map reference to an accuracy of about 41 m. The data were then recorded onto film, and image products produced that can serve as low-cost accurate map products. To decrease the dimensionality of the Landsat-4 data, principal component transformation of the data to four significant new bands was performed, and the results compared with latest available land use maps. The transformation is useful for land use analysis and in delineating vegetation anomalies which appear to reflect areas underlain by altered serpentinite. A range of image processing systems have been used to process the satellite data, including general purpose, special purpose, and personal computers. These systems are described, along with their processing performance. Index Terms-Digital Image Processing, Thematic Mapper, Multispectral Scanner, Calibration, Geometric Correction, Mapping, Digital Terrain, Enhancement, Noise Removal, Personal Computer, Entropy, Principal Components, Banding, Striping, Information Extraction, Geology, Land Use.

Effect of terrain orientation and solar position on satellite-level luminance observations

Abstract Changes in the luminance of a terrain along the local nadir direction are studied as a function of terrain slope, and the solar azimuth and elevation angles. The digital terrain data for the Santa Clara valley-Mt. Hamilton region of California, available from the U.S. National Cartographic Information Center, are used for this purpose. The elevation above mean sea-level of this selected region varies between 24 and 1,282 m. Luminance computations are performed by assuming the Lommel-Seeliger law of reflection, and the terrain reflectance to be a constant and independent of its geographic location. The atmospheric effects and contributions due to multiple reflections are neglected. It is shown that the ratio of the actual terrain luminance to that for the corresponding flat surface for 64 equal-size (∼ 4 km 2 ) finite subregions, changes by a very significant amount over the selected region, and also with changes in the solar position. Selected Landsat MSS band 7 data of the same region are presented for validation purposes.

Clustering Algorithms for MRI

Magnetic Resonance Imaging (MRI) plays a relevant role in the design of systems for computer assisted diagnosis. MR-images are multi-dimensional in nature; physicians have to combine several perceptual information images to perform the tissue classification needed for diagnosis. Automatic clustering methods help to discriminate relevant features and to perform a preliminary segmentation of the image; it can guide the final manual classification of body-tissues. Three clustering techniques and their integration in a MRI-system are described. Their performance and accuracy was evaluated on synthetic and real image-data. A comparison of our approach with the tissue-classification done by a radiologist was performed.

Gaussian Maximum Likelihood and Contextual Classification Algorithms for Multicrop Classification Experiments Using Thematic Mapper and Multispectral Scanner Sensor Data

In this paper we present the results of a study of performance of a previously proposed classification technique on real remotesensor imagery. Testing has been achieved in the framework of an analysis of variance experiment designed to compare thematic mapper (TM) versus multispectral scanner (MSS) image data under the view-point of classification accuracy. The improvements of TM relative to MSS consist in (Fl) three additional spectral bands, (F2) increased radiometric resolution, and (F3) increased spatial resolution. The impacts of factors FI-F3, with or without context (factor F4), were evaluated by a four-factor analysis of the variance experiment, by repeated classification runs on 1) a TM data set, and 2) suitably degraded versions of the same set. Figures of increase/decrease in classification accuracy due to any combinations of the four factors have been computed, along with the corresponding levels of significance. Simultaneously acquired TM and MSS data sets have been used, together with photographic data acquired in coincidence with the satellite overpass (as control data for classification accuracy computation). The relaxation algorithms proposed in a previous paper have been used to assess the impact of the contextual factor.

Modeling and image processing for visualization of volcanic mapping

In countries such as Italy, Japan, and Mexico, where active volcanoes are located in highly populated areas, the problem of risk reduction is very important. Actual knowledge about volcanic behavior does not allow deterministic event prediction or the forecasting of eruptions. However, areas exposed to eruptions can be analyzed if eruption characteristics can be inferred or assumed. Models to simulate volcanic eruptions and identify hazardous areas have been developed by collaboration between the IBM Italy Pisa Scientific Center and the Earth Science Department of Pisa University (supported by the Italian National Group of Volcanology of the Italian National Research Council). The input to the models is the set of assumed eruption characteristics: the typology of the phenomenon (ash fall, pyroclastic flow, etc.), vent position, total eruptible mass, wind profile, etc. The output of the models shows volcanic product distribution at ground level. These models are reviewed and their use in hazard estimation (compared with the more traditional techniques currently in use) is outlined. Effective use of these models, by public administrators and planners in preparing plans for the evacuation of hazardous zones, requires the clear and effective display of model results. Techniques to display and visualize such data have been developed by the authors. In particular, a computer program has been implemented on the IBM 7350 Image Processing System to display model outputs, representing both volume (in two dimensions) and distribution of ejected material, and to superimpose the displays upon satellite images that show 3D oblique views of terrain. This form of presentation, realized for various sets of initial conditions and eruption times, represents a very effective visual tool for volcanic hazard zoning and evacuation planning.

Information Extraction From Multi-Modality Medical Imaging

Two information. extraction methods were applied to selected diagnostic medical images obtained from different imaging. modalities. Sample images consisted of magnetic resonance tomograms, transmission computed tomograms, and emission computed tomograms of patients with focal and generalized brain abnormalities. Color composite analysis applied to image pairs afforded a limited but potentially useful approach to enable rapid, objective. depiction of relative and absolute contribu-tions from dualmodality parameters. Cluster analysis applied to image pairs or triplets was computationally more demanding, but this approach afforded a method for objective classification of tissues that could be generalized. Technical problems arise from patient factors and machine factors. Patient factors include motion and positioning. Machine factors include resolution, orientation, format, registration and scaling.

A Shipboard Oceanographic Data Processing and Control System

In June of 1962 a digital computer (IBM 1710) was placed aboard the Research Vessel Chain of the Woods Hole Oceanographic Institution. This system has made it possible to automatically sample, compute, and record the ships heading and speed, latitude and longitude, water depth, gravity in terms of total acceleration, free-air and Bouguer anomalies, and the magnetic field of the earth. The system was expanded in November, 1963, to provide online plotting of bathymetric, gravity anomaly, and magnetic field profiles; computer control of gravity meter spring tension; processing of ocean surface temperature and sound velocity measurements; reduction of Loran C and VLF radio navigation data to latitude and longitude; display of ships position and numerical data at remote stations aboard the ship; and malfunction detection and alarm message generation. Three magnetic-disk storage units are used for data and program storage and provide the ability to merge real-time on-line computations with background off-line computations (time-sharing). Experiments are being made using three input/output typewriters at remote locations on the ship. The advantages of a shipboard system are described.