【摘要】 该文提出了利用小波变换获取纹理结构子图像能量参数,并用这些参数进行高分辨率图像纹理结构分类的新方法。由阐述遥感影像纹理结构识别原理入手,提出影像纹理结构特征抽取的小波变换方法,构造了有明确的数学和物理意义的参数来描述影像纹理信息,在此基础上利用这些参数进行影像纹理结构分类。试验结果表明,小波变换方法适用于具有规则和较强方向性的纹理结构影像分类。更多还原

【Abstract】 This paper discusses decomposition in multi-scales of the texture and structure information in a high resolution image based on wavelet transform,describes how to build the parameters of sub-image blocks and their application in the classification of the texture and structure in a satellite image.The texture and structure are very helpful in image classification because they indicate the recognizing features of objects in a high resolution satellite image. With the high resolution satellite images being available, the details of the ground object are much clearer than before so that there are lots of clear textures and structures in the images which provide important information for image classification.On the basis of the research of wavelet transform and its meanings in both mathematics and physics,the authors first derived average energy of sub-image blocks from wavelet decomposition. In theory, the average energy of sub-image blocks at the same decomposition level is an indicator of the vertical weight, the horizontal weight and the diagonal weight of the image textures at a specific resolution. At the same time, the average energy of sub-image blocks at the different decomposition level show the change among these levels. As a result, the curves describing the average energy of the sub-image blocks indicate the different characteristics of different objects in an image.Then a new method to classify regular textures having rather strong orientations is provided. The method makes use of the energy parameters derived from the sub-image blocks to identify texture features. Results of the experiments are satisfying. This new method has particular effects in classifying regular textures and structures having rather strong orientations in high-resolution satellite images.更多还原