Abstract The problem of object perimeter protection by human steps detection using autonomous seismic activity sensors is solved in this paper. The results of the CME (consecutive mean excision) and FCME (forward consecutive mean excision) algorithms to calculate these sensors threshold level are presented. The algorithms are tested on real experimental data. The number of false alarms for CME and FCME algorithms is equal to 23 10, respectively, operating with the envelope of seismic signal. The possibilities to reduce the number of false alarms for algorithms are explored. The signal trend neutralization allows us to reduce the probability of false alarms to 16 and 7 for the CME and FCME algorithms, respectively. The signal amplitude normalization in the analysis interval makes it possible to reduce the number of false alarms to a negligible value. The results allow us to choose the algorithm depending on the input data specifics. The CME algorithm with preliminary normalization of the signal amplitude is the most appropriate to solve the problem of human steps detection by autonomous sensor due to less computational complexity.
展开▼
