基于树状分解隐马尔可夫树的纹理分类模型研究

遥感图像纹理信息丰富，要准确地进行纹理特征描述，必须尽可能地抓住其本质特征和属性。以小波域隐马尔可夫树模型为基础，并结合遥感图像特点，提出在全树状小波分解的基础上建立隐马尔可夫树模型，在子图选择上用图像熵作为判据，使分解更有针对性，并使计算效率得以提高。

Tree Structure Decomposition-based Hidden Markov Tree Model Used in Texture Classification Model Research

Remotely sensed images have rich texture information,and their characteristics and attributes must be grasped so as to get the texture features. This paper,based on the wavelet-domain hidden Markov tree(HMT) model taking the characteristics of the remotely sensed images into consideration, establishes the HMT model on the basis of tree-structured transform. It uses the entropyas the rule of transform to make the transform more relevant and the calculation more efficient.