摘要
为精确地分割高分辨率无人机航拍图像中的不同地物,提出一种基于超像素和超度量轮廓图的无人机图像分割算法.首先对图像进行线性谱聚类,生成超像素;然后根据HSV颜色空间的直方图特征计算超像素区域间的不相似度;再结合层次分割思想得到可表示边缘强度的超度量轮廓图并将其归一化;最后利用合适的阈值删除边缘强度低于该阈值的轮廓,并将所对应的区域进行合并得到分割后的图像.与ISODATA,FCM和gPb-OWT-UCM算法比较的实验结果表明,该算法图像分割准确率较高,对初始参数的依赖性小,且计算复杂度低.
An image segmentation algorithm based on superpixel and ultrametric contour map is proposed to accurately segment different images captured in the high-resolution unmanned aerial vehicle(UAV).Firstly,the image was segmented into superpixels by linear spectral clustering.Secondly,the dissimilarity between superpixels was calculated according to the histogram features in the Hue,Saturation and Value(HSV)color space.The ultrametric contour map which can represent the strength of the corresponding contour was obtained with the idea of hierarchical segmentation and then normalized.Finally,using the appropriate threshold value,the contours with lower weights than the threshold were deleted and the regions with higher similarity were merged to obtain the segmented result.Compared with segmentation algorithms including Iterative self-organizing data analysis algorithm(ISODATA),fuzzy c-means(FCM)and globalized probability of boundary,oriented watershed transform and ultrametric contour map(gPb-OWT-UCM),the experimental results show that the presented algorithm has higher accuracy and lower computational complexity and less dependence on the initialized values.
作者
宋以宁
刘文萍
宗世祥
骆有庆
Song Yining;Liu Wenping;Zong Shixiang;Luo Youqing(School of Information,Beijing Forestry University,Beijing 100083;School of Forestry,Beijing Forestry University,Beijing 100083)
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2019年第8期1294-1300,共7页
Journal of Computer-Aided Design & Computer Graphics
基金
北京市科技计划(Z171100001417005)
“十三五”国家重点研发计划(2018YFD0600200)
中央高校基本科研业务费专项资金(2015ZCQ-XX)
关键词
无人机图像
图像分割
超像素
线性谱聚类
超度量轮廓图
unmanned aerial vehicle imagery
image segmentation
superpixel
linear spectral clustering
ultrametric contour map(UCM)
