無鏡片屏下嵌入式光學指紋辨識

Lensless optical fingerprint sensors inside OLED displays

邱議鋒(Yi-Feng Chiu)

指導教授 ： 蘇國棟

國立臺灣大學/電機資訊學院/光電工程學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201903717

摘要

近年來因應電子商務消費習慣，手機搭載生物辨識系統已是一種趨勢。生物辨識中指紋辨識最普遍。光學式指紋辨識模組的原理，主要是藉由Fresnel 方程式計算，指紋中未接觸到保護玻璃的紋谷與接觸到保護玻璃的紋脊，造成光線在介面反射能量強弱的情形，再經由成像透鏡成像於感測器上，達到擷取指紋影像的功能。
光學設計之初，為了能有效、快速判斷設計架構之正確性。以及為了更真實模擬實際情形，視發生反射現象的「紋谷」為反射面。兩個紋谷間距為300μm。而在兩個紋谷之間的「紋脊」為吸收面。在Zemax光學軟體中建立此「指紋模型」。
在Zemax光學軟體中建立OLED手機模型。模擬並分析不同情形下光偵測器(photo-detector)簡稱PD，收到的光能量。此模型為「指紋辨識模型」。本文與一般光學指紋辨識最大的不同，我們沒有利用任何透鏡、稜鏡，只利用不同光源和光偵測器的位置分析。我們分析模型整體厚度等於1.104mm時降低玻璃厚度到0.40mm以及玻璃反射率0.1%時達到了指紋訊號和保護玻璃反射光的雜訊，訊雜比(S/N)約等於16.6dB。最後我們在OLED手機模型加入柔性薄膜封裝(TFE)結構，可以降低雜訊，訊雜比(S/N)約等於21.73dB。

關鍵字

Fresnel方程式 ； 指紋辨識 ； OLED

並列摘要

In recent years, in order to cope with the consumption habits of e-commerce, mobile phones equipped with biometric systems have become a trend. Fingerprint recognition is most common in biometrics. The principle of the optical fingerprint recognition module is mainly calculated by the Fresnel equation. The fingerprint does not touch the valley of the protective glass and the ridge that contacts the protective glass, causing the light to reflect the difference in energy at the interface, and then imaged on the sensor through the imaging lens to capture the fingerprint image.
At the beginning of optical design, it is necessary to effectively and quickly judge the correctness of the design architecture. In order to more realistically simulate the actual situation, the "valley" where the higher energy of the reflection phenomenon occurs is the reflective surface. The two valleys are spaced apart by 300 μm. The "ridge" between the two valleys is the absorption surface for reflecting lower energy. This "fingerprint model" was created in Zemax optical software.
In the Zemax, we build an OLED mobile phone model. Simulate and analyze the light energy received by the photo-detector referred to as PD in different situations. This model is a "fingerprint identification model." This article is the biggest difference from general optical fingerprinting. We do not use any lenses, prisms. We only use the position of different light sources and photodetectors for analysis. We found that when the overall thickness of the model is equal to 1.104mm, the thickness of the glass is reduced to 0.40mm and the reflectivity of the glass is 0.1%. We analyze the noise of the reflected light from the glass with the fingerprint signal, and the signal-to-noise ratio (S/N) was approximately equal to 16.6dB. Finally, we added a flexible thin Film Encapsulation (TFE) structure to the OLED mobile phone model to reduce noise, and the signal-to-noise ratio (S/N) is approximately equal to 21.73dB.

並列關鍵字

Fresnel equation ； Fingerprint ； OLED

參考文獻

[1] O. H. Charles Jr, S. Schmidheiny, and P. Watts, Walking the talk: The business case for sustainable development: Routledge, 2017.

[2] E. Henry, "Classification and Uses of Finger Prints.[Sl]: George Routledge and Sons," ed: London, 1900.

[3] F. Galton, "Finger Print Evidence," Nature, vol. 66, pp. 606-606, 1902/10/01 1902.

[4] H. J. Gross, H. Domdey, and H. L. Sänger, "Comparative oligonucleotide fingerprints of three plant viroids," Nucleic acids research, vol. 4, pp. 2021-2028, 1977.

[5] S. J. Hong, K. Y. Tam, and J. Kim, "Mobile data service fules the desire for uniqueness," Communications of the ACM, vol. 49, pp. 89-94, 2006.