Research on Object Shape Detection from Image with High-Level Noise Based on Fuzzy Generalized Hough Transform

Hough transform has been applied abroad in object shape detection. However, the traditional generalized Hough transform may not make the vote focus to one point when the image has a high-level noise. As a result, the object positioning is not very precise, or even wrong. It makes the Hough Transform can't be used in strong noisy image or complex object background on this condition. In this paper, we apply fuzzy set theory to generalized Hough transform and use a new method to process strong noisy image. The method regards the unfocused area not just as some simple point but a "fuzzy voting point"-a fuzzy area. Consequently, the fuzzy set theory can be used to describe the "fuzzy voting point". By constructing a new subjection function, we can calculate a cut set and use it as weight to optimize the position of the reference points. The experiments show that this method can get more accurate and robust object position than traditional method in shape detection from high-level noise image.