Abstract
The urgency of the discussed issue is caused by the need to improve methods for early detection of objects by their noise. The main aim of the study is to reduce the "dead time" in the problem of detecting objects by their noise, i.e. to reduce the implementation time of conventional sequential computation steps, i.e. the computation of the continuous wavelet transform and the integrated wavelet spectrum, as well as the detection of maximums in the threshold device (by modifying the algorithms and overlapping computational step). The methods used in the study include mathematical transformation to compute concurrently the overlapped discretized wavelet transform and integral wavelet spectrum. Computer modeling is carried out to detect objects with different test noises. The results obtained show a decrease in the total computation time when detecting narrowband signals by means of a successive correction of the wavelet spectrum due to the overlap of its computation and the continuous wavelet transform. An example of mathematical modeling of two test signals detection in a discrete white noise with a negative signal-to-noise ratio of –3 to –6 dB is given indicating that the overlap integral wavelet spectrum and continuous wavelet transform calculation makes it possible to detect both signals in the noise already at the 160th time step out of 256 time steps of a full cycle.
Keywords: continuous wavelet transform, integrated wavelet spectrum, narrowband noise, detecting, overlap calculations, wavelet, signal-to-noise ratio, mathematical modeling