Based On Biological Sonar System Of The Mobile Platform Target Detection Research

In the air, there are some marvelous animals-bats which can use their unique nature biosonar system to capture preys or navigate in darkness. The motivation of this work is to study the mechanism of bats'echolocation ability. An artificial biosonar system on mobile platform was designed for this work. Then this dissertation carried out some exploratory study of biosonar detection and classification.To simulate the function of bats'biosonar organs, this work prepared the hardware and software environment for the biosonar system on mobile platform. The hardware preparations include the hardware section determining, ultrasonic transducer selection, the signal condition circuits design for emitting and receiving transducer, power circuit design, the SNR of system enhance, micro motors selection and the motor controller design for the rotation of sonar head's "mouth". And the software for sonar chirp generating and emitting, program for receiving the echo, the host computer software and slave MCU program for controlling the rotation of "mouth" are implemented to construct the software environment.With the implemented biosonar system on the mobile platform, some efforts were performed for the bionic study of biosonar target detection and classification. Firstly,5 sensing strategy which can be used by the mobile platform in our work were designed to simulate the bats'natural sensing strategy. Then an echo data processing algorithm which imitates the animal hearing model was introduced. It combines Gammatone bandpass filter, half wave rectification, and low pass filter together. After that, an experiment was conducted to study the mechanism of fishing bats'indirect localization. The test result indicates that our artificial biosonar system can locate the fish's swimming trace successfully, and a preliminary localization error of current sensing system is calculated by error transfer function. At last, normalized cross correlation is used to estimate the similarity between target's sonar image and the stored templates. With this algorithm, a plant target is classified successfully with the sensing strategy adopted in this paper.