IPTV Parental Control based on Computer Vision

INTRODUCTION

IPTV (Internet Protocol Tele Vision)service has witnessed a huge acceptancethrough a good range of users of TV viewers. The development of communications and the wide usage of it gave intense internet services a good push through internet application. This development allowed online video services or as it is know now by IPTV services to have a good chance to be accepted and demanded by users. Users usually demand a good, stable, and reliable service through their mobile devices, such as phones, tablets, and laptops. With the available hotspots, and free internet providers inside stores, hospitals, malls… such services came available to users almost anyplace and anytime. In addition to mobile services there is always the traditional T.V located at Livingroom almost at every home. This development always comes with some challenges, one of these challenges is the parental control. Many offers have been provided to give a good solution for such essential option, especially through parents or guardians who may don’t have a good skill to manage such issue. Almost every device which has the ability to serve IPTV service has a parental control based on numerical password which usually consists of four decimal digits, which can be easily overcome [1]. The survey in [2] proves that 88% of parents/ guardians do not activate parental control option.In [3] the suggested system gave a new method to manage users who are screened depending on their registered age in the database and on the suggested age of the program requested by analyzing the information encoded inside EPG (Electronic Programming Guide) information. With the system suggested in [4], the authors provided a new algorithm as users go through a mathematical quiz depending on the age of the program included in the EPG then vocal answers provided by the users are used to test if those users are suitable to view the program or not.On the other hand image analyzing had a very good process, such advancement allowed computers to get cognitive services. A cognitive service will get any picture, analysis it, and then return all information regarding visual content which this service can find inside the picture. Microsoft one of the technology leaders revealed an online computer vision API (application program interface) [5]Fig (1).Many properties are delivered if this API detect a person face inside a picture, one of the important properties provided with this API is the age of the personas shown in Fig (2).This work will introduce a new authentication system depending on the age of users using information provided through the analyzing of the image after obtaining user’s picture, and other information related to the proper age of the user. Finally the suggested system will have the ability to decide to authenticate or de-authenticate users from viewing different IPTV programs.

Fig. (1)

Fig. (2)

METHODOLOGY

This work will test the suggested algorithm and find the efficiency for such IPTV authentication system. The experiment is conducted by checking the ratio of the corrected decisions and the ratio of the bad decisions. Correct decisions will be described as: 1. TP (True Positive)will express the ratio of users with ages that fit with the requested IPTV program and the system finds them eligible to view the requested program. 2. TN (True Negative) will express the ratio of rejected users who their ages are under the allowed age and the system finds them not suitable to view the program and reject them. While bad decisions are described with the following: 1. FP (False Positive), in this case users are under age and are not suitable to view the program but the system defines them as regular users, and allow them to view the program. 2. FN (False Negative), here users are on age or over age but the system finds them under the age and reject them from viewing the program. The Algorithm will go three stages. First the user or the viewer will request a specific IPTV service. In this stage the system will analyze the data included inside the stream and lock for the EPG data. Then the system will analyze the EPG data to find the program name. The second step is to capture a picture of the viewer by using the local camera available on the device, and then send this picture to the remote service [5] then the system will analyze all information answered from the remote service and the system will read the suggested age of the person in the picture provided. The third step will include requesting reference database to allocate the suitable age for the requested program [6]. Finally the fourth step, where the system will compare both results from both remote services [5] & [6], then a final decision will be made by verifying the age of the user, if the suitable age of the program is 5 years and under, then the system will pass users. If the suitable age is more than 6 years then the system will check the age that is given by service [5] to the age given by service [6] and decide if the user will be able to view or not the requested program. The suggested system is programmed by Python, PHP and HTML.Channel source is the terrestrial TV channels that are feed to the IPTV server. Web services are delivered under apache web server. In the viewer side all tests are done with different devices that have camera hardware, Wi-Fi connectivity, and webbrowser.

BIOMETRIC DATABASE

In this work, the database in [7] has been chosen to test our proposed system. This database provides data set of face photos. It includes the most of real-world conditions, were each set of images for each person is taken under actual variations such as pose, lightning, appearance, and other conditions. The images in this database are collected from Flickr [8], were most of photos are collected from smart phones such as IPhone 5 or later, this helped us choosing this database as same conditions for users’ photos can be gathered during everyday conditions. People in this data set are grouped into 8 groups, were each group presents a specific age range.The 8 groups are classified in the following groups: 0-2, 4-6, 8-13, 15-20, 25-32, 38-43, 48-53, 60- [7]. In this work the selected data set is divided into 3 groups: 0-5, 6-12, 12-. This age grouping provided a fair viewer classification as groups who watch IPTV service.

BLOCK DIAGRAM

Fig. (3) Presents the main block diagram of the proposed system. This block diagram starts when the viewer requests a specific IPTV channel. The next step is to analyze the Channel data. Then the system will go through deep analysis of the EPG data broadcasted within the channel data, then the system will use the extracted data like the name of the program to use the service in [5] to check suitable age that is allowed to view such program. The next step is tohave permission from the viewer to take his/ her picture to use it in the next step. Here the personal picture of the viewer will be sent to the remote service in [6] to perform the vision analysis, at this point this service is accomplished by using the computer vision web API provided by Microsoft. Once the API answers all the properties to the authentication system, the system will analyze all the data and extract the age property. Finally both ages are compared to check if the viewer is allowed or not to view the selected IPTV service.

Fig. (3)

Results:

  • Authentication classified viewers into 3 groups, this classification is based on the age range of the viewers.
  • If the requested program is suitable for ages 5 and under (group 1), then no authentication is considered.
  • For IPTV channels that have programs as there content is not suitable for ages of 5 and under, then authentication procedure was conducted.
  • 24 viewers of (group 2) 39% could pass the authentication, while 38 61% viewers could not pass, as shown in Fig. (4).
  • In group 3, 130viewers 69% of total viewers passed the authentication, whoever 58 viewers or 31% of the total viewers could not pass, as shown in Fig. (5).
  • Fig. (6) Provides success and failure cases of the second group. As it is shown there was 14 success cases of authentication TPwhich presents 23%, 10 success case of de-authentication TN16 %, while there was 2 failure authentication FP which presents 5%, and finally 36 failure case of de-authentication FN or 26% of total viewers.
  • Fig. (7) Provides success and failure cases for the third group. With this group there was 54TP, 76 TN, 9 FP, and 49 FN, that sequentially present29%,40%,5%, 26% of total viewers.

Fig. (4)

Fig. (5)

Fig. (6)

Fig. (7)

Future work:

This experiment presented a new IPTV authentication system. With this system computer vision was implemented to derive viewer’s age. The suggested system could distinguish between viewers and make decision to authenticate or de-authenticate them from viewing requested IPTV services. This system could bypass 100% of viewers who asked to view programs that are tagged with ages 5 and under. Viewers who requested programs that are designated for ages between (6&12) the system could correctly authentication 39% of viewers, and failed with 61% of the viewers. While the system could correctly authenticate 69% of viewers, and failed with 31% of the viewers. Microsoft vision API presented a good addition for such authentication system. This API presented better results with adult viewers than results with minors, but still this suggested solution is a good choice as all fail results were closer to reject other than acceptance which and with the IPTV system can be considered an accepted solution for unaccompanied minor viewers. To enhance this proposal there are some stages that can be modified or included in the client side, for example the training system can be transferred to the client device instead of using external service provided by exterior service. This modification in the system’s algorithm as shown in Fig (8) can provide many benefits such as building local vision training system which will reduce decision time as the system will be targeted to limited users instead of thousands of users’ samples. The second benefit is cutting the time of the request to the Microsoft external services and the answer time from this service, in other words there will not be request and answer times, instead this time will be shortened to the minimum as the service will be conducted locally. This will eliminate the wait period (2) Fig. (7) almost to the minimum.

Fig. (8)

References:

  1. Anne Adams, Martina Angela Sasse, and Peter Lunt, "Making Passwords Secure and Usable," Proceedings of HCI on People and Computers, vol. XII, pp. 1 - 19, 1997.
  2. Federal Communications Commission (FCC), In the Matter of Violent Television Programming And Its Impact On Children, FCC 07-50, MB Docket No. 04-261, April 2007.
  3. Këpuska, Veton Z., and Mahmood Alfathe. "Improving STB Devices’ Parental Control." International Journal of Engineering Research and Applications (2015): 133-36. Web.
  4. Alfathe, Mahmood and Veton Këpuska. "Speechmath IPTV’S Verification System". Journal of Networks and Telecommunication Systems , Vol. 2 (1) (2016): n. pag. Print.
  5. Microsoft Cognitive Services - Computer Vision API". Microsoft.com. N.p., 2016. Web. 1 Sept. 2016.
  6. Common Sense Media". Commonsensemedia.org. N.p., 2016. Web. 1 Sept. 2016.
  7. CS Research Lab". Cslab.openu.ac.il. N.p., 2016. Web. 1 Sept. 2016.
  8. Flickr, A Yahoo Company". Flickr. N.p., 2016. Web. 1 Sept. 2016.