小波变换及其在影像表示中的应用和其方向选择性算法的改进

小波理论是研究多分辨率信号处理的理论,目前被公认为有前途的新理论。金字塔结构的影像处理、影像理解,现代GIS处理多比例尺的要求以及地形、场论中的逼近均需有关这一理论的支持。本文利用小波变换分解原始影像以实现影像的小波表示,并发展为任意方向的小波变换、逆变换,利用它检测航片中建筑物及其阴影的公共边缘。这种表示方式可为某些影像特征或识别目标提供较为经济的分辨率,该表示方式使影像中具有一定层次的结构更易于发现和利用。实验结果证明了此方法是有效的。

Wavelet Transform with Its Applications to Image Representation and The Improvement of Algorithm on Orientation Selectivity

The Wavelet theory which concentrates on the multiresolution signal processing is nowdays considered to be such a greaty potential theory. Image Processing with the pyramid structure、image understanding and a modern GIS which should have the multiscale function need this theory to support. This paper proposes a new kind of image representation——Wavelet Representation by using Wavelet Transform,improves Wavelet Transform with an arbitary direction,and employs it to extract the building/shadow edges from the aerial im-ages. This wavelet representation in computer vision and image analysis typically helps in two ways:1)allowing ths most economical resolution to be used for each of several image features or objects to be treated; 2) making certain hierarchical structure in an image more explicit and easier to use. The experimental results illustrate this approach is feasible.