...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>International journal of engineering research and industrial applications >A NOVEL APPROACH FOR EXTRACTING TEXT FEATURES IN IMAGE INDEXING SYSTEM USING SAD
【24h】

A NOVEL APPROACH FOR EXTRACTING TEXT FEATURES IN IMAGE INDEXING SYSTEM USING SAD

机译：利用SAD提取图像索引系统中文本特征的新方法

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In Image indexing system, retrieval of digital image is significant of image databases. One fundamental indexing is text appearing in color images. This paper introduces a novel method for detecting, extracting and inpainting. The proposed algorithm in this paper, deals with the problem of inpainting. The method for detecting text blocks in image, we utilize stroke filter, and an efficient segmentation verification algorithms. Then, we estimate text and background color using color histogram obtained in each sub-blocks of the candidate text blocks. Finally, by implementing the missing hole-filling though texture synthesis and matching based on pixel by pixel approach, number of pixels restored at each inpainting step. We use SAD approach to find best matching pixel for damaged pixel around their neighborhood region. The matching process is based on priority for inpainting of pixels. This proposed technique makes faster than other inpainting methods. Experimental results on texture synthesis algorithm shows effectiveness of proposed method.
机译：在图像索引系统中，数字图像的检索对图像数据库很重要。一种基本的索引编制方法是在彩色图像中显示文本。本文介绍了一种新的检测，提取和修复方法。本文提出的算法解决了修复问题。该方法用于检测图像中的文本块，我们利用笔画过滤器和有效的分段验证算法。然后，我们使用在候选文本块的每个子块中获得的颜色直方图来估计文本和背景色。最后，通过纹理合成实现缺失的孔填充，并基于逐像素方法进行匹配，可以在每个修复步骤恢复像素数量。我们使用SAD方法为邻近区域周围的损坏像素找到最佳匹配像素。匹配过程基于像素修复的优先级。提出的技术比其他修复方法更快。纹理合成算法的实验结果表明了该方法的有效性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号