| Traditional edge detection has been to locate the pixels which belong to an edge. On the other hand subpixel edge estimation locates the position of an edge within a pixel. Two new subpixel edge estimation methods are proposed in this thesis. They are called the Analytical Method and the Iterative Method.; The Analytical Method provides an explicit expression for calculating the parameters of a line. It is much simpler than the other existing methods. The Iterative Method uses a nonlinear least-squares procedure, where an iterative search is performed to determined the parameters of a line. It is demonstrated that this method always converges to the correct line. The performance of the two new methods has been studied extensively by way of experiments, and the results have been compared with other structural methods. It is found that the new methods are superior to other subpixel estimation methods. |
