Adaptive LUM pre-filtering for affine invariant interest point detection

Interest point detection is required by several applications from fields such as computer vision, pattern recognition and content based image retrieval. The performance of the detectors is strictly dependent on the reliability and accuracy with which corresponding points are detected across a wide range of images that are the result of geometric and photometric transformations. This paper presents an adaptive LUM filter and suggests it as the basis of a pro-filtering stage for an affine interest point detection in order to improve the accuracy and reliability of the detector and to make it suitable for applications that do not require a matching operation