本文關(guān)鍵詞： 組合鍵檢測(cè) 聲音識(shí)別 盲源信號(hào)分離 智能手機(jī)　出處：《深圳大學(xué)》2017年碩士論文　論文類型：學(xué)位論文

【摘要】：隨著聲音定位技術(shù)以及竊聽(tīng)技術(shù)的發(fā)展,擊鍵內(nèi)容的識(shí)別的研究已經(jīng)受到工業(yè)界和學(xué)術(shù)界的關(guān)注。隨著智能手機(jī)的發(fā)展,尤其是手機(jī)上各種傳感器性能的提升,智能設(shè)備已經(jīng)被廣泛地應(yīng)用在定位問(wèn)題中以實(shí)現(xiàn)厘米級(jí)別的定位。根據(jù)信息安全的原理,竊聽(tīng)技術(shù)的研究能夠有效地阻止識(shí)別用戶的擊鍵內(nèi)容�，F(xiàn)有用來(lái)解決鍵盤擊鍵內(nèi)容的識(shí)別主要集中在以下三類:基于WiFi信號(hào),基于計(jì)算機(jī)視覺(jué),基于聲音信號(hào)�；赪iFi信號(hào),利用商業(yè)無(wú)線路由設(shè)備和無(wú)線網(wǎng)卡來(lái)檢測(cè)用戶擊鍵的手勢(shì),主要原理是檢測(cè)手勢(shì)和無(wú)線網(wǎng)絡(luò)信道之間的關(guān)系;基于攝像頭和計(jì)算機(jī)視覺(jué)技術(shù)來(lái)識(shí)別擊鍵內(nèi)容,這種方案的缺點(diǎn)是受光照條件影響;基于聲音的鍵盤擊鍵內(nèi)容的識(shí)別,通過(guò)收集敲擊鍵盤的聲音并分析聲音的特征來(lái)識(shí)別不同的鍵�；诼曇粜盘�(hào)的方法,具有很強(qiáng)的分辨率,同時(shí),現(xiàn)有的鍵盤檢測(cè)的方法很少用于鍵盤組合鍵的檢測(cè)。為了解決上述問(wèn)題,本文提出了一種基于獨(dú)立主成分算法的組合鍵擊鍵內(nèi)容的識(shí)別的方法。本文進(jìn)行了一些實(shí)驗(yàn),來(lái)驗(yàn)證之前方法在組合鍵研究上的不適用,同時(shí)驗(yàn)證主成分分析方法在組合鍵研究的可行性。以下三個(gè)部分,是本文的主要研究工作:第一,閱讀相關(guān)論文,了解研究現(xiàn)狀。特別地,對(duì)聲音盲源信號(hào)分析以及獨(dú)立成分分析技術(shù)的學(xué)習(xí),了解傳統(tǒng)的基于WiFi定位方法策略以及聲源定位方法。第二,設(shè)計(jì)實(shí)驗(yàn),采集數(shù)據(jù),部署好實(shí)驗(yàn)場(chǎng)景。在比較安靜的環(huán)境中進(jìn)行實(shí)驗(yàn)并運(yùn)用技術(shù)手段對(duì)實(shí)驗(yàn)數(shù)據(jù)進(jìn)行降噪,盡可能地減少因環(huán)境或人員偶然操作等因素產(chǎn)生的噪聲。第三,特征提取,鍵盤擊鍵內(nèi)容檢測(cè)。利用FastICA算法對(duì)雙麥克風(fēng)接收到的混合信號(hào)進(jìn)行分離,并利用機(jī)器學(xué)習(xí)方法進(jìn)行分類,識(shí)別出擊鍵內(nèi)容。實(shí)驗(yàn)結(jié)果表明,本文采取的智能手機(jī)這一方案能夠有效地檢測(cè)組合鍵。實(shí)驗(yàn)結(jié)果顯示,平均準(zhǔn)確率78.4%。本文主要的創(chuàng)新點(diǎn)在于首次利用聲音信號(hào)盲源分離技術(shù)進(jìn)行組合鍵的研究,本文設(shè)計(jì)的系統(tǒng)簡(jiǎn)單,只需要手機(jī),容易操作且具有一定的實(shí)用性。
[Abstract]:With the development of sound location technology and eavesdropping technology, the research of keystroke content recognition has attracted the attention of industry and academia. With the development of smart phones, especially the improvement of the performance of various sensors on mobile phones. Intelligent devices have been widely used in positioning problems in order to achieve centimeter-level positioning. According to the principle of information security. The research of eavesdropping technology can effectively prevent the identification of keystroke content of users. The existing identification of keystroke content is mainly focused on the following three types: based on WiFi signal and computer vision. Based on sound signal and WiFi signal, commercial wireless routing device and wireless network card are used to detect user keystroke gesture. The main principle is to detect the relationship between gesture and wireless network channel. Keystroke content is identified based on camera and computer vision technology. The disadvantage of this scheme is that it is affected by illumination conditions. The recognition of keystroke content based on sound, by collecting the sound of the keystroke keyboard and analyzing the characteristics of the sound to identify different keys. Based on the sound signal method, it has a strong resolution and at the same time. The existing methods of keyboard detection are rarely used for keyboard key combination detection. In order to solve the above problems. In this paper, an independent principal component algorithm based on the key combination keystroke content recognition method. Some experiments are carried out to verify the previous method in combination key research is not applicable. At the same time verify the feasibility of the principal component analysis method in the study of key combination. The following three parts are the main research work of this paper: first, read the related papers, understand the research status. In particular. To the sound blind source signal analysis and independent component analysis technology learning, to understand the traditional localization method based on WiFi and sound source location methods. Second, design experiments, collect data. Deployment of the experimental scene. In a quiet environment to experiment and use technical means to reduce the noise of experimental data, as far as possible to reduce the environment or personnel accidental operation and other factors generated by noise. Third. Feature extraction, keyboard keystroke content detection. FastICA algorithm is used to separate the mixed signals received by two microphones, and the machine learning method is used to classify the mixed signals. The experimental results show that the proposed smart phone can effectively detect the key combination. The experimental results show that. The main innovation of this paper is to use blind source separation technology of sound signal for the first time to study key combination. The system designed in this paper is simple and only needs mobile phone. Easy to operate and practical.
【學(xué)位授予單位】：深圳大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TN912.3

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 許紅波;粟毅;黃春琳;王懷軍;;MIMO雷達(dá)的近場(chǎng)模型DOA估計(jì)[J];信號(hào)處理;2009年11期

，

本文編號(hào)：1492715