姜继春,王晓红,许秦蓉.H-Cb混合颜色模型下快递单手写体提取算法研究[J].包装工程,2014,35(19):114-118.
JIANG Ji-chun,WANG Xiao-hong,XU Qin-rong.Handwritten Text Extraction of the Express List Based on H-Cb Mixed Colour Model[J].Packaging Engineering,2014,35(19):114-118. |
| H-Cb混合颜色模型下快递单手写体提取算法研究 |
| Handwritten Text Extraction of the Express List Based on H-Cb Mixed Colour Model |
| 投稿时间：2014-05-03 修订日期：2014-10-01 |
| DOI： |
| 中文关键词: H-Cb 混合颜色模型 手写体文字 识别率 提取算法 |
| 英文关键词: H-Cb mix color model handwritten text recognition rate extraction algorithm |
| 基金项目:上海市研究生创新基金项目（JWCXSL1302） |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 在不受光照条件的影响下， 利用 H-Cb混合颜色模型， 提取快递单底单图像手写体文字信息。 方法 首先将图像从 RGB 颜色空间分别转换到 HSI 颜色空间和 YCbCr颜色空间 ； 然后将改进的YCbCr颜色空间的 Cb 颜色分量与 HSI 颜色空间的 H 颜色分量进行信息融合； 最后对提取出的手写体文字信息进行阈值和反相处理， 并将该算法提取结果与基于 YCbCr颜色空间 Cb颜色分量阈值分割方法和基于 Lab颜色空间的手写文字聚类算法的提取结果， 在分割效果、文字识别率上进行对比。 结果利用 H-Cb混合颜色模型检测出的手写体文字更准确， 具有更高的识别率， 在理想文字切分条件下识别率达96%。 结论 使用 H-Cb混合颜色模型提取手写文字受光照条件影响小， 提取出的图像噪声小、识别率高，算法简单可行，为彩色图像的检测与判定技术提供了支撑。 |
| 英文摘要: |
| Objective To extract handwritten text information from the express lists using the H-Cb color mixing model without the effect of illumination conditions. Methods Firstly, the image was converted from RGB color space to HSI color space and the YCbCr color space. Then the Cb color component of the improved YCbCr color space was combined with the H color component of the HSI color space. Finally, threshold processing and inverse processing were conducted on the extracted handwritten text information. The extracting results of this algorithm were compared with the threshold segmentation using Cb color components of the YCbCr color space and handwritten text clustering algorithm based on Lab color space in terms of segmentation results and character recognition rate. Results The handwritten text detected using the H-Cb mixed color model was more accurate, and had a higher recognition rate. Under the conditions of ideal character segmentation, the recognition rate reached up to 96% . Conclusion Using H-Cb mixed color model to extract handwritten text could avoid the effect of illumination conditions, and the extracted image had few noises and high recognition rate, and this algorithm was simple and feasible, which provided support for the detection and decision technology of color image. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
渝公网安备 50010702501716号