The Study Of Pretratreatment Of The Audioe Signal Of Wheat

Detecting wheat using the acoustic technology can detect and analysis the quality of wheat using the signal processing technique without destroying the wheat samples. The method provides a technological means to the work of quality detection of grain procurement agency. The acoustic detection of wheat should process the collected quality signal of wheat through the technology of pretreatment firstly. This method can be helpful in further analysis and processing of quality signal of wheat.This paper focuses on the study of pretreatment of audio signal of wheat. The software and hardware devices of the acquisition device of wheat audio signal are selected and designed. The audio signal of wheat is collected and stored through the self-manufacture device. The audio signal of wheat is pre-processed by the methods of pre-emphasis, noise reduction, endpoint detection. In this paper, the audio signal of wheat is analyzed by the methods of short-time Fourier transform, wavelet transform and so on, which lay a foundation to the study of non-destructive detection of wheat based on acoustic.In this paper, the noise reduction method of bandstop filter, wavelet de-noising and spectral subtraction are studied to reduce the noise of the collected audio signal of wheat. The basic principle of several classic time-domain and frequency domain endpoint detection algorithm is studied. The existing algorithms are improved based on the basic analysis. Two-step endpoint detection algorithm based on variance, endpoint detection algorithm based on cepstral distance and the variance of frequency band, two-step endpoint detection algorithm based on wavelet transform are given in this paper. The results show that the endpoint detection accuracy can be improved with the algorithms presented in this paper compared with endpoint detection algorithm of single feature on the condition of certain SNE. The influence of the background noise can be reduced and the endpoint detection accuracy can be improved by the use of the two-step endpoint detection algorithm based on wavelet transform. This algorithm is more suitable for the need of endpoint detection of audio signal of wheat. The performance of the acoustic detection technology is also preliminarily discussed in this paper. 16 kinds of characteristic parameters in time domain and frequency domain are extracted. The relationship between these characteristic parameters and the wheat hardness index is analyzed. Experiments show that there is a certain correlation between the acoustic feature parameters of zero crossing rate and the hardness index of wheat. We can evaluate the quality of the wheat with the acoustic technology.