基於瞬態誘發耳聲傳射與同步自發性耳聲傳射特徵之身份辨識

Person Identification Based on features of Transient Evoked Otoacoustic Emission and Synchronized Spontaneous Otoacoustic Emission

連漢文(Han-Wen Lien)

指導教授 ： 劉奕汶

國立清華大學/電機資訊學院/電機工程學系所/碩士(2018年)

摘要

目前，耳聲傳射在臨床醫學的應用上，主要為聽力量測。然而，從過去瞬態誘發耳聲傳射(TEOAE)的分析結果，發現同一健康耳在不同量測時間具有一致性及穩定性，可能適合成為生物辨識之工具。本論文研究在能有效的量測健康耳之瞬態誘發耳聲傳射的情況下，分析脈衝響應之後半段訊號是否含有同步自發性耳聲傳射(SSOAE)，並從其頻譜及ConceFT觀察是否存在數個特定頻率且穩定震盪的SSOAE。初步觀察後發現不同日期的SSOAE頻譜似乎能保有每個人的獨特性。於是本論文收集19位受測者的耳聲傳射，其中20隻耳朵含有SSOAE，約52.6 %。所以，在特徵萃取上，本研究嘗試開發包含TEOAE及SSOAE之綜合判斷機制，並以K個最近鄰居法(KNN)應用於身分辨識。適當地增添SSOAE作為特徵後，本研究在僅量測15秒的耳聲傳射，其中5位之雙耳皆無SSOAE，身份辨識率約84 %；另外14位皆至少一耳含SSOAE，其辨識率達100 %。另外，利用主成份分析將特徵維度降至17維，整體身份辨識率仍可達97.89 %。未來期望同時量測雙耳之耳聲傳射，並完成為時五秒之量測辨識系統。

關鍵字

瞬態誘發耳聲傳射 ； 同步自發性耳聲傳射 ； 身份辨識

並列摘要

Currently, otoacoustic emissions (OAEs) are applied to hearing detection in the clinical medicine. However, the previous analyses show that the transient evoked otoacoustic emission (TEOAE) is also suitable for biometric identification thanks to the consistency and stability of TEOAE across each measurement from the same healthy ear. Under the circumstances that the TEOAE can be effectively measured from the healthy ears, we analyze the second half of the impulse response and examine whether the synchronized spontaneous otoacoustic emissions (SSOAEs) existed. SSOAEs are several weak signals continually oscillating at certain frequencies. We found that SSOAE spectrum seems to preserve their uniqueness at each measurement. Collected from 19 subjects, TEOAEs were measured from both ears of 19 subjects, SSOAEs existed in approximately 52.6% of them (20 over 38 ears). This research is dedicated to developing a general detection scheme that can extract TEOAE and SSOAE as features, and thus realize person identification by utilizing the k-nearest neighbors (KNN) algorithm. Under the condition that measurements only last 15 seconds for the individual, the person identification rate is about 84% for those who lack SSOAEs (5 over 19 subjects), while the person identification rate rises to 100% after adding SSOAEs for the rest (14 over 19 subjects) whose SSOAEs existed in at least one ear. Even if we reduce the feature size to 17 dimensions by principal component analysis (PCA), the person identification rate still remains 97.89 %. In the future, it seems possible to automate all the processes for binaural measurements so that personal identification can be completed in 5 seconds for certain applications.

並列關鍵字

Person ； Identification ； Transient ； Evoked ； Otoacoustic ； Emission ； Synchronized ； Spontaneous

參考文獻

[1] D. Purves, G. J. Augustine, D. Fitzpatrick, W. C. Hall, A. S. LaMantia, J. O. McNamara, and S. M. Williams. Neuroscience: Third edition. Sinauer Associates, Inc., pp. 287-293, 2004.

[2] R. Fettiplace and C. M. Hackney, "The sensory and motor roles of auditory hair cells," Nature Reviews Neuroscience, vol. 7, no. 1, pp. 19-29, 2006.

[3] D. T. Kemp, "Stimulated acoustic emissions from within the human auditory system," J. Acoust. Soc. Am. Vol. 64, no. 5, pp. 1386-1391, 1978.

[4] C. Bergevin, G. A. Manley, and C. Köppl, "Salient features of otoacoustic emissions are common across tetrapod groups and suggest shared properties of generation mechanisms," Proc. Natl. Acad. Sci. Vol. 112, no. 11, pp. 3362-3367, 2015.

[5] D. H. Keefe, M. P. Feeney, L. L. Hunter, and D. F. Fitzpatrick, "Comparisons of transient evoked otoacoustic emissions using chirp and click stimuli," J. Acoust. Soc. Am. Vol. 140, no. 3, pp. 1949-1973, 2016.