摘要
桥梁水域分割是桥梁目标识别的关键,首先对Markov随机场理论K-M分割方法进行改进,进而提出一种桥梁目标识别算法,对经过预处理的红外桥梁图像利用K-M分割方法进行桥梁水域分割;然后定义桥梁模板,寻找可能的桥梁点,并用Hough变换合并、连接;最后运用先验知识除去假目标,得到检测结果。整个算法解决了传统分割方法桥梁水域分割不清,导致目标识别准确率低的弱点。仿真结果表明,目标识别算法具有很高的准确性,可靠性,且计算效率高,时间性能好,可用于实时性处理。
Segmentation of Bridges waters is the key of bridge recognition. Firstly, this paper improved the K- Means segmentation method based on Markov random field theory. Thus a bridge target recognition algorithm was pro- posed. The algorithm for preprocessing infrared bridge images uses a Markov random algorithm in field bridge waters segmentation. After the use of bridges template for possible bridge point, Hough transform was used for merging and connecting. Finally, the priori knowledge was used to remove the false target and obtain the final detection result. With high accuracy, reliability and calculation efficiency, the time performance is good, so it can be used for real time processing.
出处
《计算机仿真》
CSCD
北大核心
2012年第11期299-303,共5页
Computer Simulation