Line extraction with the use of an automatic gradient threshold technique and the Hough transform

We present a method to detect roads and building boundaries on high resolution aerial photographs. These image features are found through the use of a line detection algorithm that is based on the three following steps: first, pre-smoothing, to remove noise and possibly irrelevant small scale details; this is followed by an automatic threshold of the gradient magnitude, thus generating a binary edge map; and in order to detect straight lines, this edge map is then used as the input of the Hough transform. We describe these steps with particular emphasis on the gradient threshold and show typical results obtainable with this method