Pattern recognition by means of the Radon transform and the continuous wavelet transform

It is known that the detection of segments in digital pictures can be effec tively performed by means of the Radon transform (RT), which concentrates t he information about linear features of an image in few high-valued coeffic ients (i.e. peaks) in the transformed domain. We describe a post-processing method based on the continuous wavelet transform (CWT), which provides inf ormation to be used in the recognition task by means of the analysis of the wave shapes of these peaks. We show that this RT- and CWT-based technique can also be successfully applied to the detection of rectangular shaped obj ects without modifications, and is able to provide information on their wid th; a further generalization is described in the case of the detection of s pots of circular and square shape. Several applications can be devised in t he field of land and sea monitoring. A case of interest occurs in aerial im ages of sea regions, when such patterns, suitably combined, can be thought of as 'V'-shaped wakes of sailing ships plus, in case, the ships themselves ; a technique for the detection of such objects is proposed through an exam ple of application. (C) 1999 Elsevier Science B.V. All rights reserved.