基于Mumford-Shah模型的水上桥梁目标分割与识别

Segmentation and recognition of bridge over water based on Mumford-Shah model

在桥梁知识的基础上,从河流和陆地的区域特性出发,提出了一种基于MumfordS-hah(MS)模型的水上桥梁分割算法。首先对MumfordS-hah模型中的区域权重进行相关性定义,实现对水域的分割,然后根据桥梁与河流区域边界的几何位置关系实现对桥梁目标的提取和识别。实验表明,该算法能够实现对水上桥梁目标的提取,尤其对远距离小目标水上桥梁及灰度梯度较弱图像的桥梁分割更有效。

A method of segmentation and recognition of bridge over water based on Mumford-Shah model and the bridge knowledge was proposed. The region-correlating weighted Mumford-Shah model was used to segment the water region on the smoothed image, and then the bridge was detected and recognized by the segmentation result and the geometrical knowledge of bridge and water region. Experimental results show that the method is effective and exact for bridge detection and recognition, especially when the bridge is a small object or the intensity gradient is not strong.