呙维, 彭旭, 刘异, 朱欣焰. 边缘约束下的分形网络分割算法[J]. 武汉大学学报 ( 信息科学版), 2019, 44(11): 1693-1699. DOI: 10.13203/j.whugis20170262
引用本文: 呙维, 彭旭, 刘异, 朱欣焰. 边缘约束下的分形网络分割算法[J]. 武汉大学学报 ( 信息科学版), 2019, 44(11): 1693-1699. DOI: 10.13203/j.whugis20170262
GUO Wei, PENG Xu, LIU Yi, ZHU Xinyan. Edge Restricted Fractal Net Evolution Approach[J]. Geomatics and Information Science of Wuhan University, 2019, 44(11): 1693-1699. DOI: 10.13203/j.whugis20170262
Citation: GUO Wei, PENG Xu, LIU Yi, ZHU Xinyan. Edge Restricted Fractal Net Evolution Approach[J]. Geomatics and Information Science of Wuhan University, 2019, 44(11): 1693-1699. DOI: 10.13203/j.whugis20170262

边缘约束下的分形网络分割算法

Edge Restricted Fractal Net Evolution Approach

  • 摘要: 分形网络演化算法(fractal net evolution approach,FNEA)是一种有效的多尺度影像分割算法,但对于具有斑点噪声、局部区域对比度低等特点的高分辨率合成孔径雷达(synthetic aperture radar,SAR)图像,直接应用FNEA算法得到的分割结果难以用于后续的面向对象影像分析。提出了基于边缘约束的FNEA(edge restricted FNEA,eFNEA)算法,通过加入边缘信息和构建异质性规则来为分割融入更多信息,提高分割效果。实验结果表明,对于微弱边缘和噪声污染严重等情形,eFNEA算法的分割结果均优于FNEA算法。

     

    Abstract: FNEA (fractal net evolution approach) is an effective multi-scale image segmentation algorithm, and is considered as the basis of object based image analysis. But it is difficult to use the segmentation result of FNEA for high resolution SAR(synthetic aperture radar) images due to speckle noise and low contrast. We propose the edge restricted fractal net evolution approach (eFNEA) which uses additional information including edge information, fractal feature, and aggregates by constructing heterogeneity rules to improve the segmentation effect. In this algorithm, exact edges are extracted using edge detection algorithm which is built in the edge detection and image segmentation(EDISON) system to restrict small scale region growing procedure. And the heterogeneity is computed by aggregating multiple features including edge regularity feature to remove broken edges and thus improve the segmentation effect. Two experiments are conducted to verify the validity of the algorithm. The results show that the algorithm performed reasonably well even when images contain weak edges or heavy noise. From this point of view, eFNEA is better than FNEA.

     

/

返回文章
返回