Detection of human echolocator mouth-click signal using bio-inspired processing

Echolocation is a process to perceive the surrounding by analysing reflected echo from space via an active emission signal often utilize by animals such as bat, dolphin, and whale as their main sensing modality mainly for navigation purposes. For almost a century, the fundamental and understanding of development radar and sonar application have been inspired by animal echolocation until today. What is lesser known is that there is a group of humans (often blind people) who have developed amazing skills and proficiency in using echolocation in their daily life. They have converted their hearing sense into visual perception mainly for navigation. In recent years, revisit studies have revealed that the human echolocator mouth-click is preferable for echolocation process. However, technical knowledge on the detection scheme for human echolocator mouth-click has yet to be discovered. What is more, the detection of human mouth-click using matched filter leads to poor outcome due to having higher multiple local maxima and lower side-lobe level values of less than 3 dB. Thus, the aim of this thesis is to propose an improved detection process of human echolocator mouth-click by implementing a Bio-Inspired processing strategy. The process is inspired by the human auditory system (using Gammatone-Filter) in synthesizing sound signal into tonotopic (sub-band channel) prior relaying it to the human brain. Then, the individual channel is correlated and is absolute to obtain correlation output. Next, correlated outputs are summed and normalized to get a single correlation output, used for decision making upon the detection phase. Improved detection output is achieved by, i) significantly reducing the presence of multiple local maxima, and ii) higher side-lobe level (SLL) exceeding 10 dB for static and moving target scenarios in different frequency range experiments surpassing output from matched filter. Finally, the Bio-Inspired signal processing proposed in this thesis can be considered as reliable detection scheme for of human echolocator mouth-click signal.

Text FK 2020 36 ir.pdf Download (1MB)

Actions (login required)

View Item