欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对无人车导航中立体视觉实时性的要求，提出了一种基于现场可编程门阵列（FPGA）的立体匹配方案及硬件结构．该方案采用非参数化的CENSUS算法进行立体匹配，克服了传统方法受亮度差异影响大的缺点，运算简单，适合硬件逻辑操作．基于FPGA的硬件系统利用合理的流水线设计提高了数据处理的并行性，可以实时地对大数据量进行处理．利用双目立体视觉硬件系统，进行了合成图对和真实图对的测试．实验结果表明，所提出的立体匹配方案和硬件结构能够以15ms的速度完成匹配，并且具有较强的鲁棒性．
英文摘要：
A solution of stereo matching and hardware architecture based on field programmable gate array （FPGA） was presented to meet the requirement of real-time stereo vision in the unmanned vehicle navigation. A non-parametric algorithm CENSUS was used for stereo matching, overcoming the drawback of traditional methods under the situation of illumination unbalance between stereo cameras. This algorithm requires only simple operations and is suitable for FPGA implementation. Proper pipelines were elaborately designed to improve the capability of the parallel data processing and large data flow. This solution was tested by the synthesized and real data on a stereo vision system. The matching process could be accomplished in 15 ms. The results show that the proposed implementation is fast and robust.
同期刊论文项目
同项目期刊论文
期刊信息
位置：