论文部分内容阅读
目前，我们进行汉字模式识别所使用的汉字是以图象的方式输入计算机。要将其转换为讨算机所能识别的字符，其关键在于输入汉字图象特征的提取。作为图象的汉字有其自身的特点，它是由较简单的笔划所组成的，每种笔划又有其较固定的方向性（即空间分步的固定性），所有的汉字都是由几种简单的笔划所组成的。而小波变换为我们提供了一个十分有效的分析图象信息的多分辨率方法，它可以将原始图象分解为模糊子图象和水平方向、垂直方向、斜方向上的子图象。因此，小波变换为我们分析汉字图象信息提供了一个十分有效的手段。本文正是基于汉字的上述特点，并利用小波对空间频率的多分辨率分析方法，对汉字图象处理而得到汉字图象特征的。
At present, the characters we use for pattern recognition of Chinese characters are input into the computer in the form of images. To convert it to a character recognized by the computer, the key lies in the extraction of the image features of the input Chinese characters. As the image of the Chinese characters have their own characteristics, which is composed of simple strokes, each stroke has its more fixed direction (that is, the space of the step-by-step fixed), all Chinese characters are by a few Composed of simple strokes. The wavelet transform provides us with a very effective multi-resolution method for analyzing image information. It can decompose the original image into fuzzy sub-images and sub-images in horizontal, vertical and oblique directions. Therefore, wavelet transform provides a very effective means for analyzing the image information of Chinese characters. This article is based on the above characteristics of Chinese characters, and the use of multi-resolution wavelet analysis of spatial frequencies, Chinese characters and image processing to get the image features.