摘要：長微博圖片是近年來出現的微博信息形式，對其進行文字提取對社交網絡信息傳播規律的分析具有重要意義，但目前尚缺乏針對靜態低分辨率長微博圖片的文字提取效果的對比分析。本文針對6種主流長微博生成工具進行分析，將主流長微博圖片分為8種主要類型，并利用6種典型OCR工具進行測試。實驗結果表明，現有OCR技術對長微博圖片的平均識別率達到90%以上，但對于常見的復雜類型長微博識別率較低，亟待研究針對性的長微博圖片文字提取方法。
關鍵詞：長微博；文字提取
中圖分類號：TN919.81        文獻標識碼：A        文章編號：
Character Recognition Effect Analysis and Comparison of Static Low-resolution Long-microblogging Pictures
Zhao Lei1, Guo Jing2*, Wu Xiao2
1Renzixing Network Technology Co., Ltd.（100029）
2CNCERT/CC（100029）
Abstract：Character recognition from long-microblogging pictures, which is emerging recent years, is important for the analysis of information propagation of social networking. However, there is still a lack of analysis and comparison of character recognition effect of static low-resolution long-microblogging pictures. This paper analysis six pop generator of long-microblogging pictures, and divided the pictures into eight main types. Then, we use six typical OCR tools for testing. The experimental results show that the existing OCR technology to achieve more than 90% of the average recognition rate of long-microblogging picture, but for the less common type of complex long microblogging recognition rate urgent targeted long-microblogging image text extraction method.
Keywords：long-microblogging, character recognition

參考文獻 (References)
[1]李勇軍.在線社交網絡的拓撲特性分析[J].復雜系統與復雜性科學，2013，9(3)：22-37.
[2]熊熙,曹偉,周欣,等.社交網絡形成和演化的特征模型研究[J].四川大學學報:工程科學版,2012,44(004):140-144.
[3]Fan Z, Tse F, Campanelli M R, et al. OCR-guided text tokenization of digital images: U.S. Patent 8,180,164[P]. 2012-5-15.
[4]Fan Z, Francis K T, Zhou B, et al. Text vectorization using OCR and stroke structure modeling: U.S. Patent 8,244,038[P]. 2012-8-14.
[5] 李心潔, 王春恒. 基于稀疏表達的視頻文字檢測方法[J]. 計算機工程, 2011, 37(6): 145-147.
[6] 許鵬飛, 姚鴻勛, 紀榮嶸, 等. 基于LHBP 多尺度向性濾波的文字檢測算法[J]. 中國圖象圖形學報, 2010, 15(8): 1207-1213.
[7] Bernard M T, Gosselin C. Spatial and Color Spaces Combination for Natural Scene Text Extraction[C]//Proc. of 2006 International Conference on Image Processing. Atlanta, USA: [s. n.], 2006:985-988.
[8] 蔣人杰, 戚飛虎, 徐立, 等. 基于連通分量特征的文本檢測與分割[J]. 中國圖象圖形學報, 2006, 11(11): 1653-1656.
[9]李敏花, 柏猛. 基于數學形態學的復雜背景圖像文字檢測方法[J]. Computer Engineering, 2012, 38(4).
[10]田其沖, 董恒強, 何新求. 靜態圖像提取文字技術[J]. 電腦編程技巧與維護, 2008 (9): 96-97.
[11]戴維, 張申生. 基于二值化聚類的圖像文字提取算法[J]. 計算機應用, 2009, 29(1): 57-59.