現今監視系統幾乎隨處可見,顯示人們對安全的重視,但大部分的系統仍舊停留在錄影的階段,為了使獲得的影像資訊即時發生用途。許多廠商已開發具有即時辨識功能的監視系統,為了達到即時處理資訊的目的,必需要在前端系統進行運算,因此需要具備影像處理能力的嵌入式裝置。本論文是以SOPC的方式建構影像處理平台,提供一個可進行軟硬體協同設計及系統層級規劃的驗證環境,以進行相關演算法的設計。我們在該平台實作一個支援即時影像處理的基礎框架,提供簡單的輸出入界面與影像處理系統進行連接,該介面具有協調頻寬之功能,提供穩定頻寬給周邊裝置進行操作。透過該介面,我們可以容易地達成對影像進行擷取、軟體運算及顯示的動作。利用這個方式,可使整個影像的輸出入及運算過程於同一平台中輕易地完成。經由本文中的執行範例,顯示該平台架構具備可行性,並可協助開發者了解相關的設計瓶頸。
Surveillance systems are widely used nowadays. However, many of the surveillance systems do not support real-time image identification features yet. As a result, many companies have started to develop systems with the capabilities of real-time image processing. To help process image in real-time, we need a device to analyze the image at the front-end of the system. This thesis presents a SoPC-based image processing platform. The platform supports a software/hardware co-design environment for the developments of image processing algorithms. We’ve implemented a framework to support real-time image processing. The framework provides simple video input/output interfaces to connect to the user provided image processing system. The interfaces were implemented so as to satisfy the bandwidth requirements for the incorporated peripheral devices. With the interfaces, we will be able to capture, compute, and display image as easily. In addition to support simple software image processing in the system, the platform also allow the software to be transformed into hardware quickly. In this thesis, serial examples based on the platform are also described to show that the framework is feasible.