作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2012, Vol. 38 ›› Issue (15): 145-147. doi: 10.3969/j.issn.1000-3428.2012.15.040

• 人工智能及识别技术 • 上一篇    下一篇

改进的Gabor小波变换特征提取方法

尹 芳1,陈德运1,吴 锐2   

  1. (1. 哈尔滨理工大学计算机科学与技术学院,哈尔滨 150080;2. 哈尔滨工业大学计算机科学与技术学院,哈尔滨 150001)
  • 收稿日期:2011-10-08 出版日期:2012-08-05 发布日期:2012-08-05
  • 作者简介:尹 芳(1978-),女,讲师、博士研究生,主研方向:图像处理,模式识别,文字识别;陈德运,教授、博士生导师;吴 锐,讲师、博士
  • 基金资助:
    黑龙江省青年科学基金资助项目(QC2009C35);黑龙江省教育厅科学技术研究基金资助项目(12511098)

Improved Gabor Wavelet Transformation Feature Extraction Method

YIN Fang 1, CHEN De-yun   1, WU Rui   2   

  1. (1. School of Computer Science and Technology, Harbin University of Science and Technology, Harbin 150080, China; 2. School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001, China)
  • Received:2011-10-08 Online:2012-08-05 Published:2012-08-05

摘要: 针对自然场景中文本存在图像背景复杂、颜色多变、噪声强烈,图像存在变形、残缺、模糊、断裂等问题,提出一种基于方向预分类的Gabor小波变换特征提取方法。利用Gabor函数良好的频率选择性和方向选择性,同时考虑到笔划相对位置的偏移,方向预分类使得滤波器对笔划方向的选择更有针对性。实验结果证明,该特征提取方法对笔划变形和低分辨率字符具有较好的适应性,能有效解决低质量场景文本的识别问题。

关键词: Gabor小波特征, 方向预分类, 模糊笔划方向, 特征提, 场景文本, 文本识别

Abstract: Gabor wavelet transformation feature extraction method based on direction pre-classification is presented to resolve the recognition of scene text which is together with complex background, inconsistent of color, strong noise, deformation, incomplete, blur, fracture and so on. The method uses the frequency and direction selectivity of Gabor function, and fully considers the direction offset. Direction pre-classification makes the stroke direction selection more accurate. Experimental result shows the method is adaptive for character of stroke deformation or low resolution and can recognize low-quality scene text effectively.

Key words: Gabor wavelet feature, direction pre-classification, fuzzy stroke direction, feature extraction, scene text, text recognition

中图分类号: