摘要
JM模型是JVT(joint video team)发布的H.264标准测试模型,对运动搜索算法的学习和研究有着重要的意义。本文结合JM10.1的源代码对UMHexagonS算法的提前终止策略进行了深入研究,对搜索模板进行象限分割来减少算法的搜索点数,在基本上保持了原有图象质量的情况下有效的节省了编码时间和运动估计时间。通过对5种测试序列的测试结果进行分析,优化后的UMHexagonS算法可以在图像信噪比和码率基本不变的情况下,编码时间节省8.87%,运动估计时间平均节省11.46%,有效的提高了编码的实时性。
JM is the standard test model published by joint video team （JVT）,which has great significance in learn-ing and research of motion search algorithm. In this paper,Combining with JM10.1 source code ,the early termination strategy of UMHexagonS algorithm （the Unsymmetrical-cross Multi-hexagon-grid Search algorithm） is optimized and the search points are reduced by the search template quadrant method of segmentation algorithm ,which could maintain the original image quality under the condition of saving the encoding time and motion estimation time. Through analyz-ing the test results of five kinds of test sequences, compared with the original algorithm, the optimized UMHexagonS algorithm saved 8.87% encoding time and an average of 11.46% motion estimation time under the condition of the im-age signal to noise ratio （SNR） and bit rate basically unchanged which effectively improve the real-time performance of the code.
出处
《长春理工大学学报（自然科学版）》
2014年第1期116-119,共4页
Journal of Changchun University of Science and Technology(Natural Science Edition)
