摘要
针对现代城市道路易受周围背景环境噪声干扰以及算法本身非理想状态而出现的“虚峰”、“虚检”及“漏检”现象,提出了一种道路中心线半自动提取的改进型高斯滤波的预处理算法。为了减轻在道路中心线提取过程中噪声对道路识别所造成的影响,首先,该预处理算法对获取的遥感图像进行伽马(Gamma)增强,增大了道路与背景信息的灰度差;通过改进的高斯掩膜滤波模板平滑去噪,用最大熵(KWS)及区域生长模型对图像进行分割生成二值图像,降低了道路识别的难度;二值图像结合霍夫(Hough)变换对图像分割后的道路进行半自动化提取。实验结果表明,该预处理算法有效的规避了复杂背景因素的干扰,相较于传统的城市道路中心线提取效果在精度和效率方面都有了明显的改善。
In view of the phenomenon of"false peak","false detection"and"missing detection"of modern urban road louie due to the interference of background noise and the non-ideal state of the algorithm itself,an improved gaussian filtering pretreatment algorithm for semi-automatic extraction of road center line was proposed.In order to reduce the impact of noise on road identification in the process of road centerline extraction,firstly,the pre-processing algorithm Gamma enhances the acquired remote sensing images,increasing the gray difference between road and background information.The improved gaussian mask filter template is used for smoothing and denoising,and the maximum entropy(KWS)and the region growth model are used for image segmentation to generate binary images,which reduces the difficulty of road recognition.The binary image is combined with Hough transform to semi-automatic extraction of the road after image segmentation.Experimental results show that the pretreatment algorithm can effectively avoid the interference of complex background factors,and has obvious improvement in accuracy and efficiency compared with the traditional urban road center line extraction.
作者
张立恒
王一
高力
徐明远
王晓
ZHANG Li⁃heng;WANG Yi;GAO Li;XU Ming⁃yuan;WANG Xiao(School of Information Engineering,Chang’an University,Xi’an 710064,China;Xi'an Zhongkexingtu Space Data Technology Co.,Ltd,Xi’an 710000,China;Information Engineering University,Zhengzhou 450000,China)
出处
《电子设计工程》
2020年第2期12-16,共5页
Electronic Design Engineering
基金
十三五国防预研（301020603
41412010401)
关键词
现代城市道路
高斯滤波
伽马增强
最大熵模型
区域生长模型
霍夫变换
modern urban road
gaussian filtering
gamma enhancement
maximum entropy model
region growth model
hough transform
