Burge R E, Ali S M
Department of Physics, King's College London, U.K.
Scanning Microsc Suppl. 1988;2:191-212.
An image model is defined based on the boundaries between image regions with different textures and series of descriptions of those textures. Six models of texture are studied under the categories of pixel-based and region-based models. Several techniques for the determination of the unit-cell of textures are presented. The model is applied to the consideration of (a) image correction, (b) the classification of image texture, (c) image enhancement including averaging of detail in periodic specimens, and (d) image data compression. A floating point format, which provides a significant simplification for the Huffman code, is also introduced.
基于具有不同纹理的图像区域之间的边界以及这些纹理的系列描述来定义图像模型。在基于像素和基于区域的模型类别下研究了六种纹理模型。提出了几种确定纹理单位晶胞的技术。该模型被应用于以下方面的考虑：(a) 图像校正，(b) 图像纹理分类，(c) 图像增强，包括对周期性样本中的细节进行平均，以及 (d) 图像数据压缩。还引入了一种浮点格式，它为霍夫曼编码提供了显著简化。
