What’s wrong with pixels? Some recent developments interfacing remote sensing and GIS

While remote sensing made enormous progress over the last years in terms of improved resolution, data availability and public awareness, a vast majority of applications rely on basic image processing concepts developed in the 70s: per-pixel classification of in a multi-dimensional feature space. It is argued that this methodology does not make use of any spatial concepts. Especially in high-resolution images it is very likely that neighbouring pixels belong to the same land cover class as the pixel under consideration. The authors argue for classification of homogeneous groups of pixels reflecting our objects of interest in reality and use algorithms to delineate objects based on contextual information in an image on the basis of texture or fractal dimension. ZUSAMMENFASSUNG Was ist mit den Pixeln los? Neue Entwicklungen zur Integration von Fernerkundung und GIS. Fernerkundung hat sich in den vergangenen Jahren bezüglich Bildauflösung, Datenverfügbarkeit und öffentlicher Präsenz enorm weiterentwickelt, trotzdem basieren nahezu alle Anwendungen auf den methodischen Grundlagen der Bildverarbeitung aus den 70er Jahren: individuelle Pixel werden im mehrdimensionalen Spektralraum klassifiziert, ohne irgendwelche räumlichen Konzepte zu berücksichtigen. Insbesondere bei hochauflösenden Bildern gehören benachbarte Pixel mit hoher Wahrscheinlichkeit zur selben Kategorie wie das aktuelle Pixel. Die Autoren argumentieren für Klassifikationsansätze homogener Gruppen von Pixeln, die realweltlichen Objekten entsprechen und aus kontextueller Bildinformation (Textur, fraktale Dimension) abgeleitet werden. Dr. Thomas Blaschke 1 Patterns do matter, or: the need for change We start our considerations of recent remote sensing practice from the user’s point of view and, more precisely, from a geographical or landscape ecology point of view: The world in its complexity and manifold relationships cannot easily be grasped in full depth. Creating models of the world or computer-based representations of its surface poses a series of problems. In landscape ecology, there is a growing awareness about continuity of phenomena and discontinuities of scales. Forman (1995) described this ambiguity through the metaphor of a person gradually descending with a spaceship or balloon. Human perception abruptly starts to discover patterns and mosaics. Many mosaics are quasi-stable or persistent for a while, separated by rapid changes that represent the “domains of scale”. Each domain exhibits certain spatial patterns, which in turn are produced by a certain causal mechanism or group of processes. Back to remote sensing: The ultimate goal is to mirror, elucidate, quantify and to describe surface patterns in order to contribute to an understanding of the underlying phenomena and processes. Since the start of the first Landsat satellite in 1972, we achieve this in more or less the same way: We measure some reflectance at the Earth’s surface. The smallest unit is called a ‘pixel’. In this paper, we do not question the pixel as an important and necessary entity. Instead, we argue for a somewhat different handling of our entities introducing the concepts of neighbourhood, distance and location. All these concepts are not new. In fact, entire disciplines like Geography are based on these conINTERFACING REMOTE SENSING AND GIS