【作者】 王岩；

【导师】 苏健民；

【作者基本信息】 东北林业大学 ， 计算机应用技术， 2008， 硕士

【摘要】 随着近年来人为的滥捕、乱伐及污染等,使野生动物的数量大幅下降,保护野生动物的呼声也在不断提高。但在日常生活中,人们通常只能听见动物的叫声而看不见它们,为物种保护带来了困难。因此,基于动物叫声的物种识别技术的研究对保护和测定当地的物种有着实际的意义,同时对生物研究和环境监测也有着重要的意义。动物一般通过叫声与同一物种的其它成员进行交流,因而基于动物叫声的物种识别是有一定事实基础的。分类学家通过研究,已把动物的叫声信号作为物种分类的一种指标。因此,基于动物叫声的物种识别技术将成为生物特征识别领域的一个热点问题。本文通过对现有声音识别技术进行研究,在掌握识别原理与系统结构的基础之上,对其进行借鉴和改进,然后将这些技术应用于动物叫声物种识别中,作为实现基于动物叫声的物种识别系统的理论基础。小波分析是近年来出现的新的时频局部化分析方法,具有其它方法无可比拟的诸多优点,能够更精细地描述声音信号,捕获其中重要的非平稳信息。因此,本文主要对小波分析理论在动物叫声物种识别中的应用进行了研究。主要内容如下:(1)研究了基于静态离散小波变换的对叫声信号进行去噪的原理,提出了一种新的去噪方法,并探讨了此方法的去噪效果及其在动物叫声物种识别系统中的具体应用;(2)讨论了基于小波变换的端点检测方法。对通过小波变换后的低中频信号进行端点检测,不但可以去除高斯噪声对检测效果的影响,而且可以去除高频脉冲噪声对系统的影响;(3)通过对叫声信号中不同频段信号的重要程度进行分析,以及对现有参数在动物叫声物种识别系统中所存在的问题的讨论,本文提出了基于小波变换和MFCC参数的小波Mel倒谱参数,即WTMFCC参数。此特征参数具有很好的灵活性和鲁棒性,更适合对动物叫声信号进行特征提取。在此基础之上,本文采用MATLAB语言,建立了基于动物叫声的物种识别系统的实验平台。通过测试表明,本系统可以很好地实现通过动物叫声进行物种识别的功能,获得了令人满意的识别效果。本文通过对基于动物叫声的物种识别系统的研究和测试,为进一步开发实用性的物种识别系统做了基础和探索性的工作。更多还原

【Abstract】 In recent years, the number of wild animals has sharply declined because of human activations, such as over-hunting, denudation and pollution. And people begin to place paramount importance on the protection of them. But in our daily life, we often hear the animal vocalizations rather than see the animal, which makes animal protection in trouble. Therefore, species recognition technology based on animal sounds is particularly true for detecting and locating animals, and this is valuable for applications such as biological research and environmental monitoring. In general, the animals generate sounds to communicate with members of the same species and thus the animal sounds have evolved to be species recognition. Taxonomists have taken the sound signal of animals as a standard of classification of animals through investigation. To be more noticeable, species recognition technology based on animals sound will become a hot topic in the field of biometrics.In this paper, the theoretical basis of species recognition system is established based on the recognition technology, which is proposed depending on investigation and improvement of modern acoustics recognition technology widely used by researchers.Wavelet analysis is a new time-frequency localization method with many advantages that can not be found in other methods, being able to describe acoustics signal more subtly and to capture the unstable information in the sound signal. Therefore, this paper mainly studies the application of wavelet analysis theory in species recognition system based on animals sound. The details are as follows:(1) The principle of sound signal de-noising based on static discrete wavelet translation is expatiated and anew arithmetic is presented, with the effect and application in species recognition system based on animals sound;(2) Wavelet-based endpoint detection is discussed. Using low-IF signal to implement endpoint detection, which come from wavelet translation, could remove influence of Gaussian noise and high frequency pulse noise to detection effect;(3) Based on analysis on importance of different band signal and discussion about issue of present parameters used in the species recognition system, this paper presents the WTMFCC parameter, which is based on derived from wavelet translation and MFCC parameter. The feature coefficient of WTMFCC has well flexibility and robust performance, which is very suitable for feature extraction from animal sound signal.By using MATLAB language, the paper establishes an experimental platform of species recognition system based on animals sound. The test result shows that, this system has satisfactory recognition effect and could use animals sound to recognize the specie of the animals belonged.This paper makes a basic and preparatory study for exploring further practical sound recognition system by investigating and testing species recognition system based on animals sound.更多还原