Detection of corners and smooth joins using wavelet transform

In this work a new computationally efficient algorithm is presented to detect both corners and smooth joins which is robust and simple. It consists of two stages: the first stage detects corners or dominant points, and the second stage detects point of inflection and smooth joins. Here a wavelet based technique is presented to detect all these features from the orientation space of the object. Since wavelets are usually suitable for the detection of sharp changes, the detection of inflections and smooth joins are realized by artificially distorting the orientation space. This distorted orientation space is exploited to detect arcs along the boundary of the object. We have also tested our algorithms under a noisy environment