隨著科技的發展,多核心處理器的技術被廣大的應用在生活中的3C 產品中,如何縮短程式處理執行的時間變成非常重要的問題。本論文 主要是探討利用異質雙核心嵌入式系統進行影像處理的效率,並希望 在未來能將之應用於無人飛行器上。一般在無人飛行器的視訊導航 中,若不是將影像傳至地面站處理,再將結果上傳,就是在單板電腦的 CPU 上直接運算。本研究透過比較ARM+DSP 的運算時間,與直接 用ARM 做運算的時間,來比較兩種模式的運算效率。在研究的方式, 首先在我們的開發主機上建構OpenEmbedded 的開發環境及TI(Texas Instruments,德州儀器) 所開發的DSP/BIOS LINK,之後安裝在嵌入式 開發板BeagleBoard-xM Rev.C 上,並改寫LOOP 範例程式,完成簡單 的影像處理,再比較其程式執行時間。本研究用不同大小的圖片進行測 試,發現使用ARM+DSP 來進行影像處理,比單純用ARM 進行影像處 理節省不少時間。此研究成果可為日後直接在機載電腦上進行視訊導航, 進行鋪路。 With the development of technology is widely applied to 3C products. How to shorten the processing time of program becomes a very important issue.This thesis investigates the efficiency of image processing using DSP collaborating with ARM for potential applications to onboard vision navigation of unmanned aerial vehicles (UAVs). Conventionally, a UAV with vision navigation either transmits the images to the ground station for processing, or process the images in the CPU. By comparing the processing time of ARM+DSP and of ARM only, we are able to conclude the which method has better efficiency. Specifically, this thesis illustrates step by step how to install operation systems in the ARM, and how to do image processing using DSP collaborating with ARM. Sample programs are provided for readers. Images of various sized are provided as examples to show the efficiency of our proposal. This thesis potentially contributes to the vision navigation of UAVs in Tamkang University.