欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对区域变化检测受分类器精度影响大、无法探测出内部细微变化这一问题，本文提出了基于热含量不变量的合成孔径雷达（Synthetic aperture radar，SAR）图像点特征变化检测．该方法利用热核特征，具有计算简便、矩阵扰动性小的特点，且有效地降低了噪声的干扰．由热核不变量的统计特性，采用期望极大化（Expectation maximization，EM）算法解决了SAR图像的自动变化检测．同时通过对权的讨论，给出了适用于SAR图像的权函数定义．对单波段单极化SAR与多极化SAR图像，本文算法相比于基于像素和似然比的方法，能够更快速更精确地检测到变化区域．
英文摘要：
The result of the methods of image change detection are affected by classifier precision and can hardly detect the image subtle change. To address this issue, this paper proposes a point signature change detection algorithm based on heat content invariants. This method, which uses the characteristics of heat kernel on graphs, is convenient to compute and has small matrix perturbation. It also can effectively reduce the noise disturbance. Based on the statistical characteristics of heat kernel invariants, the changes about synthetic aperture radar （SAR） image can be automatically detected by the expectation maximization （EM） algorithm. In addition, a weight function which is suitable for SAR image is put forward. Compared with the pixel-based and likelihood ratio method, our method can more quickly and accurately detect the change area of single band and single-polarization SAR image and multi-polarization SAR image.
同期刊论文项目
同项目期刊论文
Local discriminant non-negative matrix factorization feature extraction for hyperspectral image clas
期刊信息
位置：