This paper proposes a new computation-constrained test zone search (CTZS) algorithm that includes dual pattern search (DPS) and recursive refinement pattern search (RRPS) to reduce the number of search points of the Test Zone Search (TZS) algorithm for HEVC. The DPS uses dual pattern including the octagon and diamond in the primary grid search of TZS. Moreover, the proposed RRPS employs the large cross search pattern (LCSP) and small cross search pattern (SCSP) in the refinement stage of the TZS algorithm.
The experimental results reveal that the proposed algorithm can reduce the total encoding time and motion estimation time about 40.51% and 60.12% respectively,c with compared to that of the original TZS implemented on HEVC HM reference software. Moreover, compared to the existing TZS algorithms [6]-[8], the proposed CTZS shows better performance in terms of the encoding time by up to 35.74%. The RD curves remain unchanged to that of the original TZS algorithm implemented on the HEVC HM reference software [4] while gaining a substantial reduction in terms of the encoding time of 35.74%.
