Abstract :
The ´randomised´ Hough transform (RHT) saves computational resources compared to the standard Hough transform (SHT) by sampling its input data. Computer simulations using randomly-generated images show that the RHT can give lower detection rates and higher false-alarm rates than the SHT when used to locate straight-line segments, owing to its different parametric representation. A modified RHT is presented that achieves similar performance to the SHT, while retaining almost all of the speed advantage of the original.
