The Pattern of Echoes in Feature Space

Xiukun Li,Zhi Xia
DOI: https://doi.org/10.1121/1.4708990
2012-01-01
Abstract:In order to increase the performance of recognition in underwater bottom object detection, the distribution pattern of different kinds of echoes is researched in this paper. The past literatures more focus on the characteristics of object echoes in feature space, but the research of joint distribution pattern of object echoes, reverberation and other interference is few. In this paper, reverberation and fake object echoes are taken as the mainly interference for recognizing object echoes, and they are assumed have steady characteristic in time-frequency feature space, respectively. There are two problems are discussed: the separability between different kinds of echoes in feature space, and the stability of them. To deal with these two problems, feature compression and cluster analysis are adopted. And the feature space is generated by FDWT. The data acquired from a lake experiment is processed in this paper, and the processing results prove that object echoes, fake object echoes and reverberation within limited time scale have steady distribution pattern in FDWT feature space.
What problem does this paper attempt to address?