小波变换在图像去歪斜中的应用

提出一种基于小波变换的图像去歪斜方法。图像去歪斜的关键在于边缘检测和边框提取。边缘检测采用由粗到细的方法，利用小波变换的多分辩率特性，得到的结果较传统的方法有更好的去噪和定位性能，并能保留必要的细节。边框提取则是用改进的Ｈｏｕｇｈ变换寻找直线，将参数空间二次量化，在一定程度上解决了常规Ｈｏｕｇｈ变换抗干扰性和精度之间的矛盾。

Application of Wavelet Transform in lmage Deskewing

A new methed based on the wavelet hansform for deskewing the imags is presesnted in this paper. Two most important parts of deskewing are edge detection and extracting the image fiames. For edge detection, we put forward a coarse - to - fine method, using the multiresolution propelty of wavelet transform, whose localizing and denoising results are much better comparing to conventional metheds. For extracting the irnag frams, an improved Hough transform is presentecd to detect long lines in the image. The two level paramenter quantized Hough transform soves the contradiction between the computing complexity and noise performance of the standard Hough transform to a certain extent.