以FPGA為基底的即時車道偵測系統

Abstract

在本論文中，我們提出了一個簡單、快速又可靠的方法來偵測即時車道資訊，以及演算法之積體電路實現方式，並提出完整的硬體系統架構。我們以電腦視覺技術，根據道路標線影像特徵來偵測車道資訊。並以FPGA晶片作為演算法的實現方式，如此可縮短在晶片設計上的開發時程及費用。我們選用目前最普遍的PC系統作為前端影像擷取及應用平台，因此對於用戶端在影像擷取設備選用上或是車道資訊應用上有非常大的彈性。此外藉由PCI介面將FPGA晶片與PC系統連接，對於所開發的FPGA晶片將有高度的系統整合能力。總合以上特點，我們所提出的即時車道偵測系統不但在設備價格上相當便宜，在整合上也非常容易，因此相當具有實用性。

In this thesis, we propose a simple, fast, and reliable method to detect a lane information in real time.In addition ,we propose a complete hardware architecture to implement our algorithm.We use computer vision to detect a lane information according to the image of the lane’s mark. Using FPGA for implementation, we can reduce the time for development and the expenses in IC-Design. We choose the most common PC system to be a front platform for image processing and applications, so it is very flexile for various applications. By using PCI interface to link FPGA and PC system, our developed FPGA possesses a high ability of system integration. For here reasons, we proposed lane detection system is not only inexpensive but also very easy to implement, and thus show the applicability of the proposed method.