Robust hough transform for noisy and cluttered images

Finding arbitrary shapes within image data is a problem with applications ranging from Internet searching to intelligence data processing to analyzing zoning maps. This task is further complicated when the images are not pristine but contain noise. The effect of noise on the Generalized Hough Transform is analyzed using idealized images with variable amounts of noise. The ability of the algorithm to detect the desired shapes is reported as a function of the amount of noise. Time performance degradation is considered as are methods for increasing the ability of the algorithm to detect objects under various scale and rotation variations. As part of this thesis, a modular software platform was developed to support custom image processing algorithms including filtering, gradient transformations, edge detection, and implementations of the Hough Transform