基于MFCC和GMM的昆虫声音自动识别

昆虫的运动、 取食、 鸣叫都会发出声音, 这些声音存在种内相似性和种间差异性, 因此可用来识别昆虫的种类。基于昆虫声音的昆虫种类自动检测技术对协助农业和林业从业人员方便地识别昆虫种类非常有意义。本研究采用了语音识别领域里的声音参数化技术来实现昆虫的声音自动鉴别。声音样本经预处理后, 提取梅尔倒谱系数（Mel frequency cepstrum coefficient, MFCC）作为特征, 并用这些样本提取的MFCC特征集训练混合高斯模型（Gaussian mixture model, GMM）。最后用训练所得到的GMM对未知类别的昆虫声音样本进行分类。该方法在包含58种昆虫声音的样本库中进行了评估, 取得了较高的识别正确率（平均精度为98.95%）和较理想的时间性能。该测试结果证明了基于MFCC和GMM的语音参数化技术可以用来有效地识别昆虫种类。     

Automatic recognition of insect sounds using MFCC and GMM

昆虫的运动、取食、鸣叫都会发出声音,这些声音存在种内相似性和种间差异性,因此可用来识别昆虫的种类.基于昆虫声音的昆虫种类自动检测技术对协助农业和林业从业人员方便地识别昆虫种类非常有意义.本研究采用了语音识别领域里的声音参数化技术来实现昆虫的声音自动鉴别.声音样本经预处理后,提取梅尔倒谱系数(Mel-frequency cepstrum coefficient,MFCC)作为特征,并用这些样本提取的MFCC特征集训练混合高斯模型(Gaussian mixture model,GMM).最后用训练所得到的GMM对未知类别的昆虫声音样本进行分类.该方法在包含58种昆虫声音的样本库中进行了评估,取得了较高的识别正确率(平均精度为98.95％)和较理想的时间性能.该测试结果证明了基于MFCC和GMM的语音参数化技术可以用来有效地识别昆虫种类.     

基于颜色直方图及双树复小波变换（DTCWT）的昆虫图像识别

为了给生产单位害虫管理的普通技术人员提供简便易操作的昆虫鉴别方法, 本文提出了一种新颖的基于图像颜色及纹理特征的昆虫图像识别方法。鳞翅目昆虫翅面图像经过预处理, 确定目标区域, 再进行特征提取。首先将彩色图像从三原色(red-green-blue, RGB)空间转换至色调饱和值（HSV）空间并提取有效区域内的色度、饱和度直方图特征, 然后经图像位置校准, 提取灰度图的双树复小波变换（DTCWT）特征; 匹配首先计算两颜色直方图特征向量之间的相关性, 将相关性大于阈值的样本再进一步用DTCWT特征匹配; DTCWT匹配通过计算Canberra距离实现, 从通过第一层颜色匹配的样本中取出最近邻作为最终匹配类别。算法在包含100类鳞翅目昆虫的图像库中进行试验验证, 取得了76%的识别率, 其中前翅识别率则达92%, 同时取得了理想的时间性能。试验结果证明了本文方法的有效性。     

基于颜色名和OpponentSIFT特征的鳞翅目昆虫图像识别

【目的】本研究旨在探索使用先进的计算机视觉技术实现对昆虫图像的自动分类方法。【方法】通过预处理对采集的昆虫标本图像去除背景,获得昆虫图像的前景蒙板,并由蒙板确定的轮廓计算出前景图像的最小包围盒,剪切出由最小包围盒确定的前景有效区域,然后对剪切得到的图像进行特征提取。首先提取颜色名特征,把原来的RGB（Red-Green-Blue）图像的像素值映射到11种颜色名空间,其值表示RGB值属于该颜色名的概率,每个颜色名平面划分成3×3像素大小的网格,用每格的概率均值作为网格中心点的描述子,最后用空阈金字塔直方图统计的方式形成颜色名视觉词袋特征;其次提取OpponentSIFT(Opponent Scale Invariant Feature Transform)特征,首先把RGB图像变换到对立色空间,对该空间每通道提取SIFT特征,最后用空域池化和直方图统计方法形成OpponentSIFT视觉词袋。将两种词袋特征串接后得到该昆虫图像的特征向量。使用昆虫图像样本训练集提取到的特征向量训练SVM（Support Vector Machine）分类器,使用这些训练得到的分类器即可实现对鳞翅目昆虫的分类识别。【结果】该方法在包含10种576个样本的昆虫图像数据库中进行了测试,取得了100%的识别正确率。【结论】试验结果证明基于颜色名和OpponentSIFT特征可以有效实现对鳞翅目昆虫图像的识别。     

基于稀疏编码和SCG BPNN的鳞翅目昆虫图像识别

【目的】为了给林业、 农业或植物检疫等行业人员提供一种方便快捷的昆虫种类识别方法, 本文提出了一种新颖的鳞翅目昆虫图像自动识别方法。【方法】首先通过预处理对采集的昆虫标本图像去除背景, 分割出双翅, 并对翅图像的位置进行校正。然后把校正后的翅面分割成多个超像素, 用每个超像素的l, a, b颜色及x, y坐标平均值作为其特征数据。接下来用稀疏编码(SC)算法训练码本、 生成编码并汇集成特征向量训练量化共轭梯度反向传播神经网络（SCG BPNN）, 并用得到的BPNN进行分类识别。【结果】该方法对包含576个样本的昆虫图像的数据库进行了测试, 取得了高于99%的识别正确率, 并有理想的时间性能、 鲁棒性及稳定性。【结论】实验结果证明了本文方法在识别鳞翅目昆虫图像上的有效性。     

松毛虫性信息素的固相微萃取及质谱和触角电位分析

松毛虫性信息素成分在腺体中含量很低。本研究利用固相微萃取（SPME）技术近距离长时间采集单头云南松毛虫Dendrolimus houi处女雌蛾释放的性信息素成分, 并进一步利用气相色谱-质谱（GC-MS）和气相色谱-触角电位联用（GC-EAD）技术分析SPME采集的性信息素成分。 结果显示: 在云南松毛虫处女雌蛾求偶召唤期间SPME聚乙二醇/二乙烯基苯（CAR/DVB）萃取头吸附了大量的反5, 顺7-十二碳二烯醇（E5, Z7-12:OH）和反5, 顺7-十二碳二烯乙酸酯（E5, Z7-12:OAc）成分及微量的反5, 顺7-十二碳二烯醛（E5, Z7-12:Ald）成分, 这3种腺体成分均能激起云南松毛虫雄虫触角电位反应。与溶剂浸提昆虫性信息素成分的方法比较, SPME活体采样技术和GC-MS及GC-EAD分析方法联用研究昆虫的信息素成分具有采集的样品代表性强、样品量能满足色谱分析需要和无溶剂干扰等方面的优点。最后还着重讨论了SPME技术的应用价值、效果及实际应用中应该注意的一些问题。      

实蝇科果实蝇属昆虫数字图像自动识别系统的构建和测试

针对双翅目实蝇科果实蝇属昆虫的自动识别,本文提出利用翅及中胸背板图像的局部二进制模式（local binary pattern, LBP）特征,采用Adaboost算法, 设计和开发“实蝇科果实蝇属昆虫数字图像自动识别系统”（Automated Fruit fly Identification System-Bactrocera, AFIS-B）。该系统包括图像采集、图像裁剪、预处理、特征提取、分类器设计、识别和显示,共7个模块。研究结果表明： LBP特征可以有效鉴别实蝇科果实蝇属昆虫;在对实蝇科果实蝇属8个种的测试中, 该系统表现出较高的准确性和稳定性,平均识别率可达80%以上。此外,还对果实蝇属昆虫翅膀及中胸背板图像在光照不均匀、姿态扭曲、样本受损及样本量大小等不同条件下的识别率进行了试验测试。结果表明, 该系统对测试样本的光照不均匀、 姿态扭曲和样本受损都表现出良好的鲁棒性,正确识别率与训练集样本各个种数量在一定条件下明显正相关,与训练集样本物种总量负相关。该项研究为实蝇科有害昆虫自动识别系统的构建及实际应用提供了理论、 方法及基础数据的支撑, 亦可为其他昆虫自动识别系统的研究和构建提供有益借鉴。 关键词：     

不同方法保存的蜜蜂基因组DNA提取的比较

目的:找出适合DNA提取的昆虫标本保存方法。方法:用几种常用的昆虫标本保存方法对蜜蜂处理不同时间后,用蛋白酶K法对其基因组DNA进行提取和纯化,然后对提取产物做琼脂糖凝胶电泳及紫外吸收分析。结果:75%乙醇及冻存处理材料的基因组DNA得率较高,为7.13～8.85μg/g,电泳条带较亮;甲醛处理材料的基因组DNA得率较低,为1.50～3.21μg/g。结论:用75%乙醇及冻存处理蜜蜂较适合于其基因组DNA的提取,不宜用甲醛。     

基于近红外光谱技术鉴别植食性金龟子的初步研究

【目的】植食性金龟子是我国的重要农林害虫,探索一种快速而准确地鉴别植食性金龟子的新方法,为将此法推及至其他鞘翅目昆虫的识别来建立研究基础。【方法】利用近红外光谱法对金龟子进行鉴别,提出了用支持向量机(Support vector machine,SVM)算法对15种植食性金龟子近红外光谱图(数据)进行分析,经过噪声波段去除后,用平滑求导与标准化法对的光谱进行预处理,选取金龟子标本150个,针对不同分类阶元和分类单元将66%样本谱图作为校正集,用SVM建立鉴别模型并对模型进行自身检验,用剩余样本图谱作为预测集对这些模型进行验证。【结果】模型的自身检验显示在金龟科4个亚科的鉴别模型中,鳃金龟亚科正确识别率为86%,其他样本的识别准确率均大于95%,在亚科不同属和属下不同种的鉴别模型中,除疏纹星花金Protaetia cathaica(Bates)外,其他样本的识别准确率均为100%;模型的预测集验证结果显示,在不同分类阶元和分类单元的鉴别模型中,由于云斑鳃金龟Polyphylla laticollis Lewis样本较少未能正确识别,其他样本的识别准确率均为100%。整体试验结果较为理想,说明模型性能较好。【结论】基于已定金龟子建立的模型能够很好地鉴别大部分样本,采用近红外光谱扫描技术结合支持向量机得到的植食性金龟子鉴别模型具有很强的推广能力。     

基于鸣声模型的云南柳莺地理种群关系分析

将录自北京、甘肃、四川和陕西7个地区共117段云南柳莺(Phylloscopus yunnanensis)的鸣唱样本分别提取短时特征Mel倒谱系数(MFCC),利用系统聚类方法,构建云南柳莺不同地理种群鸣声特征之间的树状关系图,并对云南柳莺地理鸣声差异产生机制的可能因素(地理距离、海拔等)进行探讨。这是基于鸣声短时特征的物种识别在研究同一物种不同地理种群关系中的首次尝试。结果显示,其鸣声地理差异与距离之间没有显著相关性(Pearson,r=﹣0.036,P=0.762,n=117),但与海拔存在显著的相关性(Pearson,r=﹣0.836,P0.001,n=117)。     

Mapping of soil insect infestations sampled by excavation and acoustic methods.

Geostatistical analysis was used to map traditionally and acoustically sampled populations of soil invertebrates at two different times in a hayfield at Grove Hill, AL, and once in an Auburn, AL, hayfield. The distributions of nearly all the soil invertebrates and their sounds were nonrandom in all three mapping studies. The maps constructed by excavation and acoustic sampling methods were compared by correlating the estimated (kriged) soil invertebrate counts with the estimated (kriged) counts of sounds per minute (pulse rate). Acoustic and traditional estimates were positively correlated in the Auburn study. Kriged estimates for green June beetle grub counts overlapped significantly with kriged estimates of sound pulse rate (R2 = 0.47). Overlap with sound pulse rates increased slightly when other soil organisms were counted along with green June beetle grubs: estimates of sound pulse rates were significantly correlated with counts of all white grubs (R2 = 0.50), all white grubs with earthworms (R2 = 0.52), all white grubs with earthworms and earwigs (R2 = 0.59), and total invertebrates (R2 = 0.59). The correlation between acoustic and traditional estimates was not significant at Grove Hill in either year, possibly because of a lack of experience in signal analysis or because the soil invertebrates may not have generated enough sounds to be detected above the background noise levels. These results suggest that acoustic technology is a promising tool for detecting insect pests in soil, but that further study and additional analysis are needed to improve interpretation of acoustic data obtained in the field.     

在电子耳蜗CIS方案中语音增强算法的实现

目的：电子耳蜗是一个帮助聋人恢复听觉的装置。它根据人耳的仿生学原理,用有限个电极刺激神经以恢复聋人听觉。目前实际应用的电子耳蜗技术已经能够在安静环境下帮助聋人恢复一定的听觉。本文在使用GIS方案的基础上,采取了频谱增强的方法,以提高电子耳蜗的在噪声环境下的性能。另外采用计算机仿真及声音合成的方法,以评估耳蜗植入者听到的声音。本实验获得了比较好的试听效果。其中提出的方法对耳子耳蜗的研究和工程现实具有一定的意义。     

Sound Modulation in Singing Katydids Using Ionic Polymer-Metal Composites （IPMCs）

Many insect families have evolved to produce and detect complex singing patterns for the purposes of mating, display of dominance, predator escape, and other needs. While the mechanisms of sound production by insects have been thoroughly studied, man-machine exploitation of such mechanisms has remained unreported. We therefore describe a method to modulate the frequency spectrum in the chirp call of a singing insect, Gampsocleis gratiosa （Orthoptera： Tettigoniidae）, a large katydid indigenous to China and commonly known as Guo Guo or Chinese Bush Cricket. The chirp modulation was achieved through the contact of a ribbon of lonic Polymer-Metal Composite （IPMC） against wing of the insect. The IPMC effectively served as an actuator when a small DC voltage was applied to the ribbon＇s faces. By applying a sequential on/off voltage waveform to the IPMC ribbon, the katydid＇s chirp was modulated in a corresponding manner. This configuration can be used as part of a broader application of using singing insects to harness their acoustic power to produce and propagate machine-induced messages into the acoustic environment.     

Low-Frequency Sounds Repel Male Mosquitoes <Emphasis Type="Italic">Aedes diantaeus</Emphasis> N.D.K. (Diptera,Culicidae)

We experimentally demonstrated that tonal acoustic signals with a carrier frequency of 140–200 Hz had a repellent effect on male mosquitoes (Culicidae). Swarming males of Aedes diantaeus were concentrated in a small space near the auxiliary attracting sound source which simulated the flight sound of conspecific females (carrier frequency 280–320 Hz). Then, the resulting cluster of attracted mosquitoes was stimulated with test signals of variable amplitude and carrier frequency from a second loudspeaker. The direction of mosquito flight from the source of test sounds and a decrease in their number above the attracting sound source were used as the criteria of behavioral response. Pronounced avoidance responses (negative phonotaxis) of swarming mosquitoes were observed in the range of 140–200 Hz. Most of the mosquitoes left the area above the attracting sound source within one second after the onset of the test signal. Mosquitoes mostly flew up, sideways, and backwards in relation to the test acoustic vector. We presume that mosquitoes develop defensive behavior against attacking predatory insects based on analysis of auditory information. The range of negative phonotaxis is limited at higher frequencies by the spectrum of the flight sounds of conspecific females, and in the low frequency range, by the increasing level of atmospheric noise.     

Auditory Sketches: Very Sparse Representations of Sounds Are Still Recognizable

Sounds in our environment like voices, animal calls or musical instruments are easily recognized by human listeners. Understanding the key features underlying this robust sound recognition is an important question in auditory science. Here, we studied the recognition by human listeners of new classes of sounds: acoustic and auditory sketches, sounds that are severely impoverished but still recognizable. Starting from a time-frequency representation, a sketch is obtained by keeping only sparse elements of the original signal, here, by means of a simple peak-picking algorithm. Two time-frequency representations were compared: a biologically grounded one, the auditory spectrogram, which simulates peripheral auditory filtering, and a simple acoustic spectrogram, based on a Fourier transform. Three degrees of sparsity were also investigated. Listeners were asked to recognize the category to which a sketch sound belongs: singing voices, bird calls, musical instruments, and vehicle engine noises. Results showed that, with the exception of voice sounds, very sparse representations of sounds (10 features, or energy peaks, per second) could be recognized above chance. No clear differences could be observed between the acoustic and the auditory sketches. For the voice sounds, however, a completely different pattern of results emerged, with at-chance or even below-chance recognition performances, suggesting that the important features of the voice, whatever they are, were removed by the sketch process. Overall, these perceptual results were well correlated with a model of auditory distances, based on spectro-temporal excitation patterns (STEPs). This study confirms the potential of these new classes of sounds, acoustic and auditory sketches, to study sound recognition.     

氨基酸和微量元素在虫草菌感染虫草蝠蛾幼虫中的影响

本文收集了虫草蝠蛾2—6龄五个龄期组幼虫,四组处于不同发育阶段的虫草菌丝体、被虫草菌感染后的僵虫体、全虫草体及子实体共计12组样品的氨基酸,微量元素分析资料,应用聚类分析方法对各样品变量的分类单元及其贡献值进行初步探讨。结果显示:各样品间的氨基酸、微量元素组成水平各不相同,5、4龄幼虫的总体组成水平在比例关系上更接近全虫草体,虫草菌丝体愈培养至晚期就愈接近子实体的组成。分析认为,5、4龄虫易被感染与其氨基酸、微量元素的组成背景值有关;以培养至晚期的菌丝体或有性生殖孢子更易感染幼虫。     

梨小食心虫化学通信中的信息物质

梨小食心虫Grapholitha molesta Busck是我国北方果区发生的重要害虫。当前对该虫的防治主要依赖化学农药, 但引起的害虫抗药性、 杀伤天敌和环境污染等问题非常严重。食心虫自身的信息素、 寄主/非寄主的他感化学物质对于调节其配偶选择和寄主定位起着至关重要的作用, 基于信息化学物质的害虫管理策略为果园食心虫的治理提供了新的途径。本文综述了国内外有关梨小食心虫化学信息物质研究进展, 包括雌蛾释放的性信息素组分及对雄、 雌两性的引诱, 雄蛾释放的信息素, 利用性信息素的迷向研究, 寄主植物挥发性引诱物质的鉴定和筛选, 以及梨小食心虫寄主转换机制等方面的研究现状和存在的问题。具体来说, 雌蛾的性信息素包括顺-8-十二碳烯醋酸酯、反-8-十二碳烯醋酸酯、顺-8-十二碳烯-1-醇和十二碳-1-醇, 各个组分的比例在地理种群间存在变异。在室内, 通过行为试验证实两种醋酸酯对雄蛾的引诱是必不可少的, 微量的顺-8-十二碳烯-1-醇对二元组分起到增效作用。在田间, 上述3种物质组成的诱芯具有较强的活性; 由此开发的性信息素迷向技术（人工迷向丝、 可喷施的微胶囊和蜡滴）被用于梨小食心虫交配干扰, 取得了很好的效果。梨小食心虫最主要的寄主植物桃梢挥发物包括22种化合物, 其中绿叶挥发物占到50%, 行为生测证实6~8个碳原子的物质是主要的活性化合物。顺-3-己烯丁酸酯、顺-3-己烯醇、反-2-己烯醛、苯甲醛和苯甲腈的五组分混合物, 其引诱力与天然桃梢挥发物相当。通过钙成像试验证实, 尽管苯甲腈在桃梢天然挥发物中仅占0.14%, 但雌蛾对含有该物质的混合物有显著趋性, 该物质对梨小食心虫成功识别寄主具有重要意义。最后对梨小食心虫信息化学物质下一步的研究和应用前景进行了探讨。