基于邻域总变分和势直方图函数的高分辨率遥感影像建筑物提取

发布时间：2018-03-02 19:17

本文选题：高分辨率遥感影像　切入点：势直方图函数　出处：《计算机应用》2017年06期 　论文类型：期刊论文

【摘要】：针对现在的高分辨率遥感影像建筑物识别与提取方法存在的准确率低及数据要求严格等问题,提出一种基于邻域总变分(NTV)和势直方图函数(PHF)的方法。首先,计算遥感影像各像元的加权邻域总变分似然函数取值,并进行区域生长分割,将矩形度和长宽比作为约束条件提取候选建筑物;然后,进行阴影自动提取;最后,利用数学形态学对阴影进行处理,计算处理后的阴影和候选建筑物之间的邻接关系得到建筑物,并用最小外接矩形对其边界进行拟合。为了验证所提算法的有效性,选取深圳市PLEIADES影像中9幅具有代表性的子影像进行实验。实验结果表明,所提方法的平均查准率和平均查全率分别达到97.71%和84.21%,与水平集和基于颜色不变性特征两种建筑物提取方法相比,在总体性能F_1上具有10%以上的提高。
[Abstract]:In order to solve the problems of low accuracy and strict data requirement in the present building recognition and extraction methods of high resolution remote sensing images, a new method based on neighborhood total variation (NTV) and potential histogram function (PHF) is proposed. The weighted neighborhood total variational likelihood function of each pixel of remote sensing image is calculated, and the region growing segmentation is carried out. The rectangular degree and aspect ratio are used as constraints to extract candidate buildings. Then, shadow is automatically extracted. In order to verify the validity of the proposed algorithm, the shadow is processed by mathematical morphology, the adjacent relation between the shadow and the candidate building is calculated, and the boundary is fitted with the smallest outer rectangle. Nine representative sub-images of Shenzhen PLEIADES image are selected for experiment. The experimental results show that, The average precision rate and average recall rate of the proposed method are 97.71% and 84.21, respectively. Compared with the two building extraction methods based on the level set and color invariance feature, the proposed method has more than 10% improvement in overall performance.
【作者单位】：福建师范大学光电与信息工程学院;福建师范大学医学光电科学与技术教育部重点实验室;福建师范大学福建省光子技术重点实验室;
【基金】：教育部“长江学者和创新团队发展计划”创新团队项目滚动支持计划(IRT_15R10) 福建省自然科学基金项目(2017J01464)~~
【分类号】：TP751

【相似文献】