AUTOMATIC DETECTION OF RETINOPATHY

Abstract

This study takes into account the issue of estimating multi-pitch and direction-of- 
arrival (DOA) simultaneously for multichannel harmonic sinusoidal signals. The 
ESPRIT approach, which estimates the multi-pitch frequencies of multiple harmonic 
signals, is based on subspace methods that leverage the invariance feature in the time 
domain. It is initially applied to a spatio-temporal matrix signal model for a uniform 
linear array. We then provide the estimated pitch frequencies and, utilising the shift 
invariance structure in the spatial domain, the DOA estimates based on the ESPRIT 
approach. A computationally more efficient technique based on ESPRIT without 2-D 
searching is suggested, which performs comparable to the current state-of-the-art 
methods. On top of that, the suggested method's DOA and pitch estimate asymptotic 
performance analysis is given. Lastly, both synthetic and real-life recorded data are 
used to demonstrate the efficacy of the suggested strategy.