本論文提出一個可以加快影像擷取與顯示之多主從系統架構，並且提出一個依據SoPC (System on a Programmable Chip)技術之軟硬體共同設計方法來實現即時的人體膚色追蹤。在加快影像擷取與顯示之多主從系統架構上，本論文設計五個硬體電路：（1）CMOS 擷取模組（CMOS Capture Module），（2）RAW2RGB 模組（Raw to RGB Module），（3）CMOS主控端模組（CMOS Master Module），（4）LCM 主控端模組（LCM Master Module）及（5）LCM 控制模組（LCM Control Module）。從實驗結果可以驗證所提之多主從系統架構確實可以加快影像擷取與顯示並且有效減少Nios II 處理器的負擔。此外，在即時的人體膚色追蹤之設計與實現上，本論文結合C 語言與硬體描述語言之軟硬體共同設計方法來設計七個影像處理模組：（1）影像濾波器（Image Filter），（2）影像擷取器（Image Grabber），（3）人體膚色偵測（Face Skin Detection），（4）影像增強處理（Image Enhancement），（5）邊緣偵測（Edge Detection），（6）目標物分割（Object Segment）及（7）目標物顯示（Object Display）。從實驗結果可以驗證所提之軟硬體共同設計亦確實可以用較少的影像處理時間來達到即時的人體膚色目標之辨識與追蹤。 In this thesis, a multiple master-slave system architecture is proposed to accelerate the process speed of image capture and display. Furthermore, a software and hardware co-design method based on a SoPC (System on a Programmable Chip) technique is proposed to implement a real-time target image tracking. In the proposed multiple master-slave system architecture, five hardware accelerating circuits: (1) CMOS Capture Module, (2) Raw to RGB Module, (3) CMOS Master Module, (4) LCM Master Module, and (5) LCM Control Module are design and implemented. Some experiment results illustrate that the proposed multiple master-slave architecture can actually accelerate the process speed of image capture and display. Moreover, the loading of Nios II processor can be reduced effectively. In the design and implementation of real-time face skin tracking, a software and hardware co-design method which integrates C language and Hardware Description Language (HDL) is applied to design seven image processing modules: (1) Image Filter, (2) Image Grabber, (3) Face Skin Detection, (4) Image Enhancement, (5) Edge Detection, (6) Object Segment, and (7) Object Display. Some experiment results illustrate that the image processing time is reduced effectively by the proposed method so that a real-time tracking of face skin object can be implemented.
