In order to provide personalized services in an intelligent digital TV (DTV), a fast and robust viewer detection algorithm is essential. However, conventional face detection algorithms are too costly or not robust for consumer use. Thus, in this paper, we propose a novel viewer detection algorithm for real-time viewer identification in the intelligent DTV. In the proposed algorithm, a viewer is detected by using our proposed modified codebook (MCB) method, and then the face region is extracted by using the normalized horizontal and vertical histogram from the detected person. Experimental results show that the proposed algorithm successfully performs at the speed of 20~25fps, which is suitable for real-time applications in the intelligent DTV.