A FAST METHOD FOR ANIMATED TV LOGO DETECTION Ersin Esen 1, 2 , Medeni Soysal 1, 2 , Tuğrul K. Ateş 1, 2 , Ahmet Saracoğlu 1, 2 and A. Aydın Alatan 2 1 TÜBİTAK Space Technologies Research Institute 2 Department of Electrical and Electronics Engineering, M.E.T.U. Balgat 06531 Ankara, Turkey ABSTRACT As a recent trend some TV stations prefer to use animated logos, therefore the detection of the presence of an animated TV logo emerges as a new requirement for certain applications. In this paper we present a novel method for the detection of animated television logos in real-time. The main idea is to handle all frames of the animated logo in a unified manner. For this purpose a unified logo boundary representation is utilized. In the training stage, the boundaries of the animated logo from each frame are placed in a single set. During detection, a voting-based decision scheme is performed in order to determine the presence of the trained logo. Furthermore robustness of the method is improved by incorporating negative clues regarding the existence of the animated logo obtained from the region of interest. Aforementioned clues are unified in order to reach a final decision by using effective combination rules. Finally, time windowing is used for eliminating false positives with short durations. The proposed method is examined through typical broadcast data and promising results are obtained. 1. INTRODUCTION Television (TV) broadcast stations use logos mainly as a visual aid to express their ownership of the aired media. These ownership indicating logos often differ in accordance with the type of the broadcasted material (program, commercial etc.). In addition, due to legal regulations in many countries, additional logos indicating the audience suitability rating of programs, shows and films are becoming widespread. These logos are displayed for a significant amount of time at the beginning of each program to inform the audience about its rating decided by experts according to some standards. Another usage of logos is for indicating the sponsor firms; sponsored programs, such as live games, involve the intermittent display of an overlaid logo of the sponsor. Logo detection has its roots in document analysis and data mining areas [1],[2]. In the document domain, the logos are used to infer some important properties, such as source of the document. However, utilization of the logos in TV broadcast resulted in a variety of reasons for their detection, such as video content understanding. Automatic classification of broadcast type can be easily performed utilizing logos [3]. However, in some situations, the user wants to remove the traces of the station's ownership from the video recorded from the TV. Such an aim is also possible with the techniques given in literature [4] by the prerequisite that the logo is automatically detectable. Another client-side application of logo detection is within personal video recorders (PVR). Commercial breaks that the most of common audience does not like can be removed from scheduled recordings by the exploitation of the fact that most TV stations change their logos, while switching between programs and commercials. In addition to this, for online viewers a zap-back alert can be set for a station which gave a commercial break. In this way, a viewer might zap through the other stations during the commercials and then be precisely alerted to return to the original station as soon as the commercial break ends. Another category of usage for logo detection in TV broadcast is publicity companies. These companies conduct the analysis of multiple TV stations' broadcasts in order to prepare reports for their clients about the durations and frequency of their advertisements and/or logos. Broadcast companies themselves also make use of logo detection in their own broadcast. Utilization of logo detection enables them to easily assess the quality of the broadcast at the receiver side [5]. Panel monitor producers also depend on logo detection in a fundamental problem [6]. Due to the long-time static scenes in the display, a ghostly appearance of these static pixels occurs even after the display is turned off [6]. This situation is called as the 'burn-in' problem and might cause permanent damage in the display. By the help of the detection of logo pixels, this damage can be prevented by regular alteration of color values. Finally, government institutions that are responsible for surveillance of activities of TV broadcast stations, take advantage of logo detection for the enforcement of legal regulations [7]. These regulations include ones that are based on the length and frequency of commercial breaks. These parameters can easily be determined automatically, provided that a robust, efficient and generic logo detection algorithm is at disposal.