Digital Validation Code Recognition Algorithm Research and Design
Putting forward a method of extracting the image of validation code base on the amount of information1.First and foremost,the image of validation code is processed by Otsu Law,then removes the isolated point in the image and makes use of Median Filtering to denoise.After split the image up into several small digital images,we use the method of Finite element to divide the split digital image into a lot of basal unit.We should compute the amount of information in each basal unit,at the same time;the connecting relationships of each basal unit are recorded.We sort amount of information in each basal unit according to descending order.We choose a suitable threshold value so that we can remove the basal unit that amount of information is below the threshold value.At last we construct the relationship map of the digital image according to the connecting relationships of each basal unit in order to extract the image outline.Then we carry out pattern recognition 23 by invariant moments so that we can recognizethe digital validation code.
Amount of Information Invariant Moment Threshold Value Connecting Relationship
Peng Cui Qingping Guo Pengpeng Duan
School of Computer Science and Technology,Wuhan University of Technology,Wuhan,Hubei 430063,China
国际会议
大连
英文
300-304
2008-07-27（万方平台首次上网日期，不代表论文的发表时间）