根據 Google 委託易普索(Ipsos)公司所作調查，台灣地區 2013 年智慧手機普及率已高達 51%，如能將此資源有效運用在全台灣超過 2200 萬台機動車輛之監控與查察，應可對改善都市交通及穩定社會治安等方面發揮有效助益。結合雲端運算之智慧手機車牌辨識系統在人性化設計、提升辨識率及調閱車籍相關資訊等方面均具有開發優勢，唯有以 3.5G 無線通信傳輸數百萬像素(pixel)之彩色影像部份存有瓶頸。可行的解決方案是將AdaBoost 字元辨識演算法劃分為影像處理及字元辨識兩部份，並將影像處理部份於智慧手機上執行，使其產生僅有車牌文字區域之最小資料量正向灰階影像後才進行 3.5G 無線傳輸，藉以改善影像傳輸耗時的問題；至於字元辨識部分則將主機端接收到的正向灰階影像資料，運用雲端技術進行分散處理，以達成大量與高速運算的需求。本研究之驗證測試是以 Android 手機為前端攝像及影像處理平台，透過撰寫手機 APP 介面以進行全自動化作業。後端伺服主機內建 AdaBoost 字元辨識系統及雲端控制系統，可在系統完成字元辨識後再與車籍資料庫進行比對，藉以提高辨識率及調閱相關資訊後回傳手機應用。

According to the investigation Google commissioned Ipsos, the penetration of smart phone in Taiwan has reached 51% in 2013. If we used this resource effectively to monitor and to inspect over 22 million vehicles in Taiwan, it should be able to play an effective vote for improving urban traffic and stabilizing social order. The license plate recognition systems with smart phone and cloud computing have the advantages of development in user-friendly design, enhance recognition rates and access to information of vehicle registration, only exist some of bottleneck in transmitting megapixels of color images with 3.5G wireless communication. One of the possible solutions is to divide the complete processing of license plate recognition into image processing and character recognition processing two parts, and to take advantage of AdaBoost algorithm in both. The portion of image processing will run on smart phone to produce the minimum data with only positive grayscale image of license plate characters to transmit in 3.5G wireless communication, to effectively improve the time-consuming problems of image transfer. For the part of character recognition, the host server will subjected to dispersion treatment with cloud technology to achieve large demand data processing and high-speed operation for received forward grayscale images. Experimental tests of this study is to utilize an Android phone as the front imaging and image processing platform, and write APP interface to achieve fully automated operation. The back-end host server was built-in an AdaBoost characters recognition and clouding control system to conduct character recognition processing and to compare against vehicle registration database to improve recognition rate and return information to smart phone.