Feature detection with automatic scale selection

The fact that objects in the world appear in different ways depending on th e scale of observation has important implications if one aims at describing them. It shows that the notion of scale is of utmost importance when proce ssing unknown measurement data by automatic methods. In their seminal works , Witkin (1983) and Koenderink (1984) proposed to approach this problem by representing image structures at different scales in a so-called scale-spac e representation. Traditional scale-space theory building on this work, how ever, does not address the problem of how to select local appropriate scale s for further analysis. This article proposes a systematic methodology for dealing with this problem. A framework is presented for generating hypothes es about interesting scale levels in image data, based on a general princip le stating that local extrema over scales of different combinations of gamm a-normalized derivatives are likely candidates to correspond to interesting structures. Specifically, it is shown how this idea can be used as a major mechanism in algorithms for automatic scale selection, which adapt the loc al scales of processing of th local image structure. Support for the proposed approach is given in terms of a general theoretica l investigation of the behaviour of the scale selection method under rescal ings of the input pattern and by integration with different types of early visual modules, including experiments on real-world and synthetic data. Sup port is also given by a detailed analysis of how different types of feature detectors perform when integrated with a scale selection mechanism and the n applied to characteristic model patterns. Specifically, it is described i n detail how the proposed methodology applies to the problems of blob detec tion, junction detection, edge detection, ridge detection and local frequen cy estimation. In many computer vision applications, the poor performance of the low-level vision modules constitutes a major bottleneck. It is argued that the inclu sion of mechanisms for automatic scale selection is essential if we are to construct vision systems to automatically analyse complex unknown environme nts.