Robust Image Corner Detection In Computer Vision using Machine Learning

Abstract

The vision of the machine relies on the processing of images to acquire the information needed to
interpret, grasp and measure the world around the machine vision systems. The key local features of
the image are the corners. They are usually nothing more than points with a high curvature and
appear at the intersection of various image intensities. A corner detection supporting vector machine
and an algorithm based on the artificial neural network is supported. No complex differential
geometric operators can be involved with a help vector machine and artificial neural network
algorithms. All approaches are having hidden capability of learning, resulting in excellent results for
a wide spectrum of images. The empirical outcomes shows that the proposed methods of corner
detection outperforms the current strategies even in terms of robustness in corner detection even after
we include input noise with different noise levels. Image de-noising is done with the help of median
filter system. The task of comparison and evaluation of feature detectors have also been presented in
this paper.