Research And Realization Of Acoustic Target Recognition Based On DSP

The acoustic target recognition is key technology in ground target probing system, it has roused prodigious attention in home and abroad in recent years, relevant research academies and institutes of our country colleges and universities have consecutively researched and discussed of theory in the field of sound target signal characteristic,sound sensory technology,signal processing technology and artificial intelligence,and have got a series of result of scientific research.However, the technology has not reached maturity owing to the insufficient research in the critical problem such as the feature extraction of unstable target signal and system's synthesis and integration.This paper mainly researches many theoretic and technical problems of acoustic target recognition technology ,also it researches in system hardware, practicality, miniaturization as well as the real-time recognition.This paper research the basic theory and methods of the sound target identification system by the modern signal processing method firstly, including sound signal preprocessing, characteristic extraction and classified recognition.Firstly, carrying on the detailed theoretical analysis and the computer simulation to the traditional linear filter method and the modern wavelet denoising technology;Secondly, using the AR parameter model, the wavelet energy and the FOBW three symbols to conduct the characteristic extraction research;Finally, using the nearest neighborhood method and the K-NN method to carry on the classification to the target. The computer software simulation experiment indicated that chose FIR,the wavelet energy and K-NN method as the main arithmetic can make the classified result accuracy be high and the classified effect is better。In system realization,this article took TMS320F2812 as the core and designed hardware circuitry system for sound target identification , including signal recuperation module, A/D gathering module, memory expand module, power source administration module and communication module and so on; completed acoustic signal gathering,processing, characteristic extraction and recognition software design, realized various modules DSP algorithm transplant in the recognition system and optimized the partial algorithms, enhanced the running rate in the DSP. Obtained the satisfactory effect by using the system to carry on gathering, recognition experiment for the vehicles and the personnel sound signal.