Corner detection using the point-to-centroid distance technique

What is it about?

Corners, a highly important local features of images and corner finding, play important a crucial role in computer vision and image processing, such as object tracking and vehicle detection. We present a new measure of corner sharpness termed as Point-to-Centroid Distance (PCD), and then examine its behaviors, which display beneficial characteristics that help distinguish corners from non-corners. Based on PCD behaviors, we propose a novel corner detector. Extensive experimental results demonstrate that the PCD technique is effective and simultaneously efficient for corner detection

Featured Image

Photo by Nico Smit on Unsplash

Photo by Nico Smit on Unsplash

Why is it important?

We propose a novel corner detector based on the Point-to-Centroid Distance (PCD) technique to enhance the performance of contour-based corner detectors.We exploit the characteristics of the PCD of contour to develop the corner response function. Since PCD utilizes a large neighborhood without using any derivatives, it is less sensitive to local variations and noise of the curve, which are weaknesses faced by CSS-based corner detectors. Furthermore, we adopt relative distance instead of absolute distance to make PCD more robust to the uniform scaling transform.

Perspectives