Abstract
Block-matching motion estimation plays an important role in video coding. Various simple and efficient block-matching algorithms (BMAs) are developed, like some popular methods, exhaustive search or full search (ES, FS respectively), three step search (TSS), new three step search (NTSS), diamond search (DS) etc. In this paper, we propose an improved for the fast motion adaptive rood pattern search (ARPS), which consists of two sequential search stages: Initial Search and Refined Local Search. The key idea of this method is above different motion characteristics of different sequence images, we adopt different Thresholds in the mean absolute difference (MAD) between the current and reference block. The proposed algorithm are compared which some BMAs by tested at several video sequences, the excellent performance is verified and it is suitable for videoconferencing sequences. The algorithm that are evaluated in this paper are implemented in video codec and provide competitive computational speedup with similar distortion performance as compared with the popular Diamond-based Search algorithm in the MPEG-4 Simple Profile.