Application of Pattern Recognition to Concept Discovery Export

@inbook{citeulike:4237774, abstract = {This paper explores the ways that pattern matching techniques can be applied to raster datasets to develop new concepts for geography. It is argued that it is becoming increasingly important for geographers to develop new ways of generalising datasets that will allow them to overcome the increasing data richness of the geocybersphere. The data set used for this study is a population density surface derived from the 1991 census of population by Bracken and Martin (1989). Using this data set the aim is to take the data-poor geographical theories of urban social structure of the first half of the century and make use of the data-rich environments of the 1990s to test the theories in a general and robust manner. To achieve this pattern matching techniques used in computer vision and other fields will be applied to raster data of population density and social and economic variables for Great Britain. Initially the raster data is segmented by the application of image analysis techniques to identify 129 urban areas in Britain. These urban areas are then compared to templates of the theoretical models of Burgess (1925) and Hoyt (1939) using methods developed in the fields of computer vision and medical imaging. Several urban areas are found that are similar in social structure to the theoretical models developed earlier in the century. The urban areas are then compared to themselves to determine if there were any other groupings of modern British cities that can be made in terms of their social structure. Several such groups are discovered and will be briefly discussed.}, author = {Turton, Ian}, booktitle = {High Performance Computing}, citeulike-article-id = {4237774}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.43.4591}, keywords = {data-mining, geography, gis, patternrecognition}, pages = {467--478}, posted-at = {2009-03-30 19:08:07}, priority = {0}, title = {Application of Pattern Recognition to Concept Discovery}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.43.4591}, year = {1999} }

TY - CHAP ID - citeulike:4237774 L3 - citeulike-article-id:4237774 N2 - This paper explores the ways that pattern matching techniques can be applied to raster datasets to develop new concepts for geography. It is argued that it is becoming increasingly important for geographers to develop new ways of generalising datasets that will allow them to overcome the increasing data richness of the geocybersphere. The data set used for this study is a population density surface derived from the 1991 census of population by Bracken and Martin (1989). Using this data set the aim is to take the data-poor geographical theories of urban social structure of the first half of the century and make use of the data-rich environments of the 1990s to test the theories in a general and robust manner. To achieve this pattern matching techniques used in computer vision and other fields will be applied to raster data of population density and social and economic variables for Great Britain. Initially the raster data is segmented by the application of image analysis techniques to identify 129 urban areas in Britain. These urban areas are then compared to templates of the theoretical models of Burgess (1925) and Hoyt (1939) using methods developed in the fields of computer vision and medical imaging. Several urban areas are found that are similar in social structure to the theoretical models developed earlier in the century. The urban areas are then compared to themselves to determine if there were any other groupings of modern British cities that can be made in terms of their social structure. Several such groups are discovered and will be briefly discussed. EP - 478 TI - Application of Pattern Recognition to Concept Discovery T2 - High Performance Computing SP - 467 KW - data-mining KW - geography KW - gis KW - patternrecognition AU - Turton, Ian PY - 1999/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.43.4591 ER -