Hausdorff Metric based Feature Extractor

In this paper we propose a new template based feature extractor called Hausdorff Metric Feature Extractor (HMFE). This feature extractor makes use of Hausdorff distance metric for calculating the corner strength. The working of HMFE is justified using the concept of derivatives and gradients. We have also analysed the performance of HMFE under the five criteria: detection, localisation, repeatability, complexity and sensitivity to noise. The performance of HMFE is comparable with Harris and KLT and in some cases it even outperformed them. We have also extended HMFE for colour domain- CHMFE.