A Comparison Framework for Spectrogram Track Detection Algorithms

This paper presents a method to facilitate performance comparisons between different spectrogram track detection algorithms tested upon different data sets. There is no standard test database which is carefully tailored to test the different aspects of such algorithms and real data is often proprietary in nature. This naturally hinders the ability to perform comparisons between a developing algorithm and those which exist in the literature using the same test database. The method presented in this paper will allow researchers to present, in a graphical way, information regarding the data on which they test their algorithm while not disclosing proprietary information.