Judgement And Recognition Of Text Area In Spam Image

Under the background of "Internet +",the emergence of mass image spam has become an important factor to damage the image and the safety of the network.It also influences net citizens to use the internet seriously.The Research on the technology of filtering image spam has a very important practical significance of the issue for saving the social running cost and maintaining the national security and stability.Spam images contain a large number of text regions.To judge and recognize the text areas,and to carry on the image classification based on the content will improve the system of filtering image spam and the effect of controlling spam.In addition,this will contribute to the maintenance of good order in the email and on the Internet.In this paper,we introduce three kinds of methods for judging the text regions firstly,which are based on the edge detection technique,the maximum between cluster variance and the character of the stroke width.A lot of tests and optimization of these methods have been carried out,and we have obtained many good results.Then,we can effectively identify the text regions in the image by the fine processing technology and recognition technology of the text regions.Finally,according to the keyword spotting of the text specific content,the labeling and classification of spam images are accomplished.In summary,we put forward the overall solution of judgment and recognition of text area in spam image,set up the test set of the spam image sample,and complete the test work of the whole solution.Satisfactory results are achieved.Through this study,we have achieved the recognition and identification of the text region in the spam image,finished the established target,and made an effort to maintain the normal order of e-mail using.