基于条件随机场的多时相遥感影像分类*

多时相遥感影像分类方法通常使用人工设置的转移矩阵作为时间上下文信息，这样不仅难以获得准确的转移矩阵，而且没有充分利用时间上下文信息。针对多时相遥感图像中的时间与空间上下文信息难以构建的问题，提出了一种基于条件随机场模型的多时遥感影像分类方法。首先运用最大期望算法生成用于描述时间上下文信息的时间势能，然后结合空间以及时间上下文信息构造了条件随机场模型，最后使用该模型对多时相遥感影像进行分类。一系列的实验结果表明，该方法可以有效提高遥感影像的分类精度。

A New Multitemporal Remote Sensing Image Classification Method based on Conditional Random Fields

The conventional methods for classifying multitemporal remote sensing images applied artificial transfer matrix as the temporal context information. However, it is difficult to obtain the accurate transfer matrix and fully use the temporal context information. To handle difficulty in constructing the temporal and spatial context information in multitemporal remote sensing images, this paper proposed a novel multitemporal remote sensing image classification method based on conditional random fields. Firstly, this paper applied Expectation maximization algorithm to create the temporal potential that describes the temporal context information. Then, it constructed the conditional random field model by means of combing the spatial and temporal context information. Finally, this paper used the model to classify remote sensing images. Extensive experiment results demonstrate that the proposed method effectively improves the classification performance on remote sensing images.