本文为西澳大利亚大学（作者：ROOZBEH ANVARI）的硕士论文，共74页。

图像处理技术在自动驾驶中的应用已经引起了国内外学者的广泛关注。然而，图像处理算法的要求很高，这给传统的实时实现带来了很大的负担。同时，FPGA的出现为ASIC的快速原型设计和实现带来了许多便利，因此，与传统方法相比，能够在相对较短的时间内设计、测试和合成图像处理算法。

本论文研究如何将所需算法进行最佳组合，以达到车道检测与追踪问题的最佳解决方案，同时使设计适合最小系统的目标。该结构实现了三种算法：方向可调滤波、Hough变换和Kalman滤波。对每个模块的理论背景进行了研究，详细描述了实现过程，分析了设计的成果和不足。

The application of the Image Processing toAutonomous Drive has drawn a significant attention in the literature andresearch. However the demanding nature of the image processing algorithmsconveys a considerable burden to any conventional real-time implementation.Meanwhile the emergence of FPGAs has brought numerous facilities toward fastprototyping and implementation of ASICs so that an image processing algorithmcan be designed, tested and synthesized in a relatively short period of time incomparison to traditional approaches. This thesis investigates the bestcombination of required algorithms to reach an optimum solution to the problemof lane detection and tracking while is aiming to fit the design to a minimalsystem. The proposed structure realizes three algorithms namely SteerableFilter, Hough Transform and Kalman Filter. For each module the theoreticalbackground is investigated and a detailed description of the realization isgiven followed by an analysis of both achievements and shortages of the design.

1.  引言

2. 文献回顾

3. 边缘检测

4. 车道检测

5. 跟踪与卡尔曼滤波

6. 结论

7. 未来展望

更多精彩文章请关注公众号：