农业科教视频中文字信息提取算法  

Text Extraction Method Based on Agricultural Science and Education Video

作　　者：赵洁 罗丹 樊李行 曹梦琪 耿耀君[1] ZHAO Jie;LUO Dan;FAN Li-xing;CAO Meng-qi;GENG Yao-jun(School of Information Engineering,Northwest A＆F University,Yangling Shaanxi 712100)

机构地区：[1]西北农林科技大学信息工程学院,陕西杨凌712100

出　　处：《数字技术与应用》2018年第6期129-130,共2页Digital Technology & Application

摘　　要：农业科教视频中的文字内容包含了丰富的语义信息,视频文字自动提取对视频分析、索引和检索有着极其重要的作用。在农业科教视频复杂的背景下,现有提取方法难以高效准确地定位并提取文字。本文提出一种基于复杂背景下稳健的农业科教视频文字提取方法,该算法主要通过视频解码、MSER文本定位、投影分割及Tesseract文字识别实现视频文字提取。实验结果表明:该方法可以快速提取视频文本信息,提取精度较高,具有较好的稳健性。The text content in agricultural science education video contains rich semantic information.,and the automatic extraction of video text plays an extremely important role in video analysis, indexing and retrieval. Under the complex background of agricultural science and education videos,existing extraction methods are difficult to locate and extract texts efficiently and accurately. This paper proposes a robust agricultural science and education video text extraction method based on a complex background. It realize text extraction by decoding video, localization of text by MSER,projecting and segmenting the text area and text recognition by using Tesseract. Experiments show that this method can quickly locate the video text area,with high accuracy and good robustness.

关 键 词：文字提取 MSER 投影分割 Tesseract 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]