An Efficient Motion Estimation Schemes for Application Targeting High Quality Hevc Schemes
R. Kavitha, Sarath Vijayan · 2015
2 Abstract: In video coding the Motion Estimation (ME) is accomplished by two types of algorithm, namely, Full Search and Fast Search algorithm. The time and computational complexity of full search made them inefficient for motion estimation. On the other hand, the improved time efficiency of the Fast Search algorithm made them a popular choice for motion estimation. A novel approach to develop the fast motion estimation algorithm for High Efficiency Video Coding (HEVC) is proposed in this paper. The proposed algorithm is implemented in HEVC using the HM 15.0 reference software and the proposed algorithm is evaluated in terms of Peak Signal to Noise Ratio (PSNR), motion estimation time and Mean Square Error (MSE). The evaluation is performed using various standard videos such as Coast guard, Container, Harbour, Mother, Mobile, Bridge, Crew, City, Sea and Ice. The proposed algorithm outperforms the conventional UMHexagonS algorithm under various parametric evaluations. The proposed algorithm achieves an average PSNR gain of 0.4%, 1% and 0.52% for Quantization Parameter Q =20, Q =25 and Q =30 respectively. In similar, the plot for ME time PP P against the number of frames, shows that the motion estimation time is reduced in the proposed algorithm. Finally, the Mean Square Error (MSE) is computed and found to be reduced when compared to conventional algorithm.