D e t e c t io n A n d R e c o g n it io n O f N o r t h A t l a n t ic R ig h t W h a l e C o n t a c t C a l l s In T h e P r e s e n c e O f A m b ie n t N o ise
a b s t r a c t The problem of detection and recognition of contact calls produced by North Atlantic right whales, Eubalaena glacialis, is considered. A proposed solution is based on a multiple-stage hypothesis-testing technique involving a spectrogram-based detector, spectrogram testing, and featur...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Published: |
2008
|
| Subjects: | |
| Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.1035.1295 http://jcaa.caa-aca.ca/index.php/jcaa/article/download/1999/1746/ |
| Summary: | a b s t r a c t The problem of detection and recognition of contact calls produced by North Atlantic right whales, Eubalaena glacialis, is considered. A proposed solution is based on a multiple-stage hypothesis-testing technique involving a spectrogram-based detector, spectrogram testing, and feature vector testing algorithms. Results show that the proposed technique is able to detect over 80% of the contact calls detected by a human operator and to produce about 26 false alarms per 24 h of observation. i n t r o d u c t i o n Continuous monitoring of North Atlantic right whales (NARW) presence in large areas can be accomplished by passive acoustical methods using data recordings obtained from distributed autonomous hydrophone systems To reduce subjectivity and to decrease the labor costs, various NARW detection methods known from the literature can be used (see e.g., The goal of the research presented in this paper is to reduce the probability of false alarm in spectrogram-based detectors without negatively affecting the detection probability. The proposed technique is reduced to a multiple-stage hypotheses-testing process. In the initial stage, the spectrogram-based detector [11] is applied. The data segments accepted as signals in the initial stage are recognized using the proposed recognition technique. The hypothesis that the detected segment belongs to the known types of impulsive noise is tested in the second stage. If this hypothesis is rejected, a feature vector (FV) is extracted and tested in the final stage. Test results obtained using real data recordings are presented. d a t a m o d e l a n d p r o b l e m f o r m u l a t i o n We use the data model similar to that considered in We assume that the spectrogram-based detector is applied to the input data in the initial stage. For each 1 s data segment |
|---|