Wu Yueying, Liu Pengyu, Gao Yuan, Jia Kebin
Beijing Advanced Innovation Center for Future Internet Technology, Beijing University of Technology, Beijing, China.
Beijing Laboratory of Advanced Information Networks, Beijing, China.
PLoS One. 2016 Nov 4;11(11):e0165698. doi: 10.1371/journal.pone.0165698. eCollection 2016.
High-efficiency video compression technology is of primary importance to the storage and transmission of digital medical video in modern medical communication systems. To further improve the compression performance of medical ultrasound video, two innovative technologies based on diagnostic region-of-interest (ROI) extraction using the high efficiency video coding (H.265/HEVC) standard are presented in this paper. First, an effective ROI extraction algorithm based on image textural features is proposed to strengthen the applicability of ROI detection results in the H.265/HEVC quad-tree coding structure. Second, a hierarchical coding method based on transform coefficient adjustment and a quantization parameter (QP) selection process is designed to implement the otherness encoding for ROIs and non-ROIs. Experimental results demonstrate that the proposed optimization strategy significantly improves the coding performance by achieving a BD-BR reduction of 13.52% and a BD-PSNR gain of 1.16 dB on average compared to H.265/HEVC (HM15.0). The proposed medical video coding algorithm is expected to satisfy low bit-rate compression requirements for modern medical communication systems.
在现代医学通信系统中，高效视频压缩技术对于数字医学视频的存储和传输至关重要。为了进一步提高医学超声视频的压缩性能，本文提出了两种基于使用高效视频编码（H.265/HEVC）标准的诊断感兴趣区域（ROI）提取的创新技术。首先，提出了一种基于图像纹理特征的有效ROI提取算法，以增强ROI检测结果在H.265/HEVC四叉树编码结构中的适用性。其次，设计了一种基于变换系数调整和量化参数（QP）选择过程的分层编码方法，以实现对ROI和非ROI的差异性编码。实验结果表明，与H.265/HEVC（HM15.0）相比，所提出的优化策略通过平均实现13.52%的BD-BR降低和1.16 dB的BD-PSNR增益，显著提高了编码性能。所提出的医学视频编码算法有望满足现代医学通信系统的低比特率压缩要求。
