Download presentation
Presentation is loading. Please wait.
Published byOscar Williams Modified over 9 years ago
1
Odessa national academy of telecommunications n. a. A.S. Popov Department of Automation and control of technological processes CONTENT IMAGES FILTERING Ph.D., docent V.I. Zagrebnyuk Lecturer V.Y. Kumysh Odessa 2011
2
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Ensuring the safety of children on the Internet relates to the problems of regulating the telecommunications market. In the problems of regulating the telecommunications market one should distinguish at least three aspects: -Economical; -Technical; -Social. It is known that in the Internet more than 30 billion web pages with pornographic material are registered. According to the announce of service providers representatives it is impossible to block the access to porn sites effectively without closing large segments of the Internet. The experience of China (Great Chinese Firewall) can be considered as an example, when Chinese population is deprived of access to the vast number of informational resources. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, The technical aspect of child safety providing on the Internet
3
Odessa national academy of telecommunications named after A.S. Popov Department of Automation and control of technological processes Appearance of huge amount of specialized applications, on the software tools market, realizing supervision under content access on the side of the user, tells us about actuality of Internet traffic selection problem. – ParentalControl is the addition for the Internet Explorer, that helps parents to prevent access for kids to the adult web pages; – KidsControl is the parental control instrument for kids activity on the Internet; – Tools for limiting access to applications, games and web pages on the basis of Windows Vista; – Tools for limiting access to the web pages in Norton Internet Security and Kaspersky Internet Security. While almost all of this applications were in English some years before today many of them are developed specially for russian- and ukrainianspeaking users, that, undoubtedly, tells about existence of the demand on such solutions. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, The technical aspect of child safety providing on the Internet
4
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes All these solutions are based on one principle: the program comes with a set of filters, that are configured on sexual materials, foul language, violence, online-games, casinos, forums, and other criteria, allowing parents to choose different filtering parameters for their child. The option of an interactive addition of filter by means of the user, so-called black list, is also present. Nevertheless, the problem of automatic blocking of socially dangerous content is far from a final decision: it is necessary to increase the accuracy of black list forming on the basis of objective estimation methods; it is necessary to attract large number of highly qualified experts in various fields for creation of black lists on the basis of subjective estimation methods; it is necessary to update filters sets and black lists continuously due to the rapid increase of the number of similar resources. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, The technical aspect of child safety providing on the Internet
5
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Existing methods assume, that the human skin color is determined by two major color components r (red) and g (green). Component r is the blood color influence and r + g is melanin color influence. In fact, due to the use of digital processing for increasing the quality and image compression, the color of human skin is significantly distorted and contains all three color components r, g and b. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Content images filtering
6
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes The method for blocking and automatic denying the access to web pages with pornography images and videos have been developed. This method is based on the concept of the content image retrieval by the sample. It suggests: Creating the two-level image as the database of the samples of human skin (the dominant color descriptors at the 1st level, histogram descriptors at the 2nd level). Preliminary analysis of the loaded by browser images in order to identify those images, that are likely to be pornographic. It includes the dominant color descriptors extraction from the images and the following content image retrieval at the 1st level of the database. Final analysis and, if necessary, blocking of the site and its inclusion in the "black list" with following access denying. Involves the histogram descriptor extraction from the images, taken at the preliminary analysis stage, and content image retrieval of the selected images at the 2nd level of the database. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Content images filtering
7
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, The dominant color descriptor extraction for content image retrieval Images, that are relevant by the color content Dominant color descriptors
8
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Histogram descriptor extraction for content filtering of images Histogram descriptorOriginal image “Body color” segment
9
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Histogram descriptor extraction for content filtering of images Histogram descriptor Original image “Body color” segment
10
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Histogram descriptor extraction for content filtering of images Histogram descriptorOriginal image “Body color” segment
11
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Histogram descriptor extraction for content filtering of images Histogram descriptor Original image “Body color” segment
12
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh Histogram descriptor extraction from the images with limited color range Histogram descriptorOriginal image “Body color” segment
13
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes The method for blocking and automatic denying the access to web pages with pornography images and videos suggests: Image or independent video raster color quantization to decrease the color redundancy. Dominant color extraction. Preliminary analysis of the loaded by browser using content image retrieval at the 1st level of the database. Histogram descriptor extraction. Final analysis using content image retrieval of the selected images at the 2nd level of the database. Definition of “Body color” part of the image. Taking the decision on blocking the image. Taking the decision on blocking the site. Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Proposed method
14
Odessa national academy of telecommunications named after A.S. Popov Department of automation and control of technological processes Content Images Filtering V.I. Zagrebnyuk, V.Y. Kumysh, Thank You for Your attention!
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.