Semi - Automatic and Automatic Vectorization of Raster Images

Abstract

An efficient and general software for converting a binary raster image (.tiff file) comprised of pixels into a vector image in .shp format is to be developed. Raster and vector are two methods of representing geographic data in digital computers. The software for GIS makes use of ESRI shape files. In a shape file, the geometry for a feature is stored as a shape comprising a set of vector coordinates. Shape files support point, line and area (polygon) features. In raster representation, space is divided into an array of rectangular cells. All geographic variation is then expressed by assigning properties or attributes to these cells. Shape files have the advantages such as: faster drawing speed & edit ability, Less disk space, easy to read & write, etc. Some of the images that are used by GIS are Raster images. Doing manipulation on these images by GIS software requires the conversion of these raster images to vector images. Existing algorithms suffer from the defects of being slow, memory intensive and producing vast quantities of very short vectors. The process of vectorizing raster maps is subject to major ambiguities. Therefore the main objective is to develop efficient software for the conversion of raster images into vector images both automatically & semi-automatically. The algorithm used in the software should be fast & should consume less memory. This dissertation presents an algorithm for converting raster maps in a rapid manner to accurate vector maps with a minimum of vectors.