Automated Simultaneous Multiple Feature Classification of MTI Data Neal R. Harvey, James Theiler, Lee Balick, Paul Pope, John J. Szymanski, Simon J. Perkins, Reid B. Porter, Steven P. Brumby, Jeffrey J. Bloch, Nancy A. David, Mark Galassi Space and Remote Sensing Sciences Group Los Alamos National Laboratory Los Alamos, New Mexico 87545 ABSTRACT Los Alamos National Laboratory has developed and demonstrated a highly capable system, GENIE, for the two- class problem of detecting a single feature against a background of non-feature. In addition to the two-class case, however, a commonly encountered remote sensing task is the segmentation of multispectral image data into a larger number of distinct feature classes or land cover types. To this end we have extended our existing system to allow the simultaneous classification of multiple features/classes from multispectral data. The technique builds on previous work and its core continues to utilize a hybrid evolutionary-algorithm-based system capable of searching for image processing pipelines optimized for specific image feature extraction tasks. We describe the improvements made to the GENIE software to allow multiple-feature classification and describe the application of this system to the automatic simultaneous classification of multiple features from MTI image data. We show the application of the multiple-feature classification technique to the problem of classifying lava flows on Mauna Loa volcano, Hawaii, using MTI image data and compare the classification results with standard supervised multiple-feature classification techniques. Keywords: supervised classification, multiple features, evolutionary computation, lava flow classification 1. INTRODUCTION Much interest has been shown in recent years in the development of feature extraction tools which can assist in the exploitation of the ever-increasing quantities of multi-spectral data that are becoming available. Creation and development of task-specific feature-detection algorithms is important, yet can be extremely expensive, often requiring a significant investment of time and effort by highly skilled personnel. At Los Alamos National Laboratory we have developed an automated system for the generation of feature extraction/classification tools, which we refer to as GENIE. Our particular interest is the pixel-by-pixel classification of multi-spectral remotely-sensed images, both to locate and identify and also to delineate particular features of interest. The large number of features in which we are interested, together with the variety of instruments which are available, make the hand-coding of suitable feature- detection algorithms impractical. We therefore employ a supervised learning approach that can generate image processing pipelines capable of distinguishing features of interest. Until recently our approach has been to only consider the two-class problem (distinguishing a single class against a background of “other” classes), however, many applications require the segmentation of an image into a larger number of distinct features or land-cover types. To this end we have extended GENIE’s capability to allow the simultaneous classification of multiple features/classes from multispectral data. The technique builds on previous work and its core continues to utilize a hybrid evolutionary- algorithm-based system capable of searching for image processing pipelines optimized for specific image feature extraction tasks. To demonstrate and evaluate the system we gave it the task of classifying lava flows on Mauna Loa Volcano, Hawaii, from multispectral data obtained from the Multispectral Thermal Imager (MTI) Satellite. In order to have some bench-mark against which to compare GENIE’s performance, we gave the same classification tasks to some standard, commonly-used supervised classification techniques. Work supported by the U.S. Department of Energy. Emails: {harve,jt,lbalick,papope,szymanski,s.perkins,rporter,brumby,jbloch,ndavid,mgalassi}@lanl.gov.