A Research on Logo Recognition in the E-business
It is one of the urgent problems to retrieve and catalog E-goods images according to the product iogos in many E-business application. This paper puts forward an algorithm based on the combination of Fourier transform and information entropy, guiding the research on the logo recognition problem. Correlation ratio threshold and entropy difference ratio threshold were used to evaluate the matching result. In the experiment, the accuracy ratio and recall ratio in the logo recognition were statistically analyzed with the application of a large quantity of E-goods images. The results show that our algorithm works effectively on the logo recognition under the condition of proper correlation threshold and entropy difference ratio threshold.
Logo recognition image matching Fourier transform information entropy
Hui Li Youzhi Zheng
College of Software Shenyang Normal University Shenyang, China Department of Computer Science & Technology Tsinghna University Beijing, China
国际会议
长沙
英文
3260-3264
2009-10-10（万方平台首次上网日期，不代表论文的发表时间）