基于内容的图像过滤技术的研究-通信与信息系统专业论文.docxVIP

肇毒强}工大学礞士学攮论文Abstract 肇毒强}工大学礞士学攮论文 Abstract With the development of Internet，irifermation l宅sources have lnereased largely． All kinds of bad information，especially pornographic information，are spreat by this beyond-region，beyond-national boundaries and open COmmunication means．To stop the Internet pornographic information becomes an important task of the country and the scienty。Recently people have done a lot of research and put forward a series of schemes．Because the Internet pornographic information includes mainly pornographic images and videos，filtering the pornographic images is an effective method to stop the pornographic information。In order to filter pornographic images， this paper takes a deep research On several key technologies in the area of content_based image filtration and proposes an implementation scheme for all image filtration used in Email filtration system． This paper discusses in detail about tWO key technologies，which are color and texture’S technology in the area of content—based image filtration．Based on a completed study on the feature of pornographic image’S color，We put forward new model————the statistic color space model for skin—color detection．Comparing with other skin—color detection models used in image filtration，we found that the statistic color space model had obtained better segmentation effect and higher segmentation efficiency．Then，in this paper，we study several common models of texture description，and propose tWO models for skin—texture detection．During the test，we used three standards respectively named Precision，Recall and efficiency to estimate the performance of the two models，and analyzed their advantages and disadvantages． According to the comparison，we can make proper choice in real application． This paper also presents content—based image filtration module solution for Email filtration system．And the image filtration module has taken some effect．We hope it will provide some practical and academic references for the fu