In this project, the problem of road centre-lines extraction on satellite images is addressed. A simple semi-automated methodology and a classical segmentation framework will be used to extract road centre-lines on satellite images. Unlike a fully automated road centre-line detection algorithm, the operator selects a region of interest on an image. We then apply a Gaussian filter on the image to smooth the image. Next, we perform a morphological area closing operation on the image to simplify the data in the image. After the data from the image is simplified, we perform watershed transform to obtain connected lines of a selected road network. Lastly, recursive iterations are implemented in order to eliminate some connected lines that do not belong to the selected road centre-line. Experimental results reveal that road centre-lines are successfully extracted, although the algorithm needs to be improved when dealing with roads with occlusions.
Copyright is held by the author.
The author has not granted permission for the file to be printed nor for the text to be copied and pasted. If you would like a printable copy of this thesis, please contact firstname.lastname@example.org.
Member of collection