适应移动智能设备的目标跟踪器

针对增强现实技术在移动智能设备上应用需求,设计了一种适应移动智能设备嵌入式系统的目标跟踪器。在特征描绘阶段采用亮度信息进行二值特征的快速分割并建立强显著性的二值特征描绘器;在特征选择阶段提出一种稀疏跟踪搜索模板进一步提高跟踪算法的执行效率。然后,在跟踪器中建立了存储目标初始信息的静态库与不断更新目标外观或运动变化信息的动态库,通过对比静、动态库与搜索模板区域中的信息确立跟踪目标位置。对跟踪器执行时间进行了对比,结果表明:在保证较高的跟踪精度条件下,采用稀疏搜索模板能明显改善算法的执行时间,将搜索半径设为10可在绝大多数情况下满足跟踪器的实时要求。对跟踪器的有效能力亦进行了对比,结果显示:在3组不同搜索半径下,提出的DBRISK描绘方法的平均跟踪误差相对于BRISK(Binary Robust Invariant Scalable Keypoints)方法分别下降了16%、28%和29%。实验表明:提出的方法能够明显改善跟踪器的信息提取准确度,适用于计算能力有限的移动智能设备。     

基于运动特征的HEVC快速帧间预测编码研究

为了降低高效视频编码(HEVC,high efficiency v ideo coding)的帧间预测复杂度,提出了一种基于运动特征的HEVC快速帧间预测新方 法。首先利用视频相邻帧的时域相关性,通过计算 每个待编码单元(CU)及其子块的帧差离散度(FDD)确定该CU的最佳编码深度d ;再依据该深度下CU的区 域运动特征(RMF_d)将待编码CU划分为3类运动区域,进而确 定该CU的候选帧间预测模式,减少 不必要的帧间预测模式遍历过程。试验结果表明,本算法可以在保证编码性能损失不大的前 提下显著提高编码效率;与标准算法相比,在低延时和随机访问两种编码结构下,同等客观 质量下码率(BDBR)分别增加0.89% 和0.83%,同时节省了51.6%和48.5%的编码时间。     

Region-of-interest based rate control for UAV video coding

To meet the requirement of high-quality transmission of videos captured by unmanned aerial vehicles (UAV) with low bandwidth, a novel rate control (RC) scheme based on region-of-interest (ROI) is proposed. First, the ROI information is sent to the encoder with the latest high efficient video coding (HEVC) standard to generate an ROI map. Then, by using the ROI map, bit allocation methods are developed at frame level and large coding unit (LCU) level, to avoid inaccurate bit allocation produced by camera movement. At last, by using a better robustness R-λ model, the quantization parameter (QP) for each LCU is calculated. The experimental results show that the proposed RC method can get a lower bitrate error and a higher quality for reconstructed video by choosing appropriate pixel weight on the HEVC platform.     

Fast prediction unit selection method for HEVC intra prediction based on salient regions

In order to reduce the computational complexity of the high efficiency video coding(HEVC) standard, a new algorithm for HEVC intra prediction, namely, fast prediction unit(PU) size selection method for HEVC based on salient regions is proposed in this paper. We first build a saliency map for each largest coding unit(LCU) to reduce its texture complexity. Secondly, the optimal PU size is determined via a scheme that implements an information entropy comparison among sub-blocks of saliency maps. Finally, we apply the partitioning result of saliency map on the original LCUs, obtaining the optimal partitioning result. Our algorithm can determine the PU size in advance to the angular prediction in intra coding, reducing computational complexity of HEVC. The experimental results show that our algorithm achieves a 37.9% reduction in encoding time, while producing a negligible loss in Bjontegaard delta bit rate(BDBR) of 0.62%.     

红外激光车载云台去抖动设计

车辆行驶产生的颠簸、晃动,会严重影响红外激光车载云台采集视频的质量,不利于信息的观察和判读。为了改善视频质量,提出一种实时有效的车载电子稳像方法。首先,用非线性扩散滤波建立尺度空间,使视频帧在无损精度的前提下突出边缘信息;然后,采取图像亮度信息与梯度信息相结合的特征提取策略,利用图像亮度信息快速搜索潜在特征点并记录其位置,并通过分析其梯度信息与预判条件的对比结果来选取优质特征点,在特征描绘阶段,通过增强二值特征描绘器之间的显著性和差异性,提升特征量的辨别力;最后,通过位置验证辨别相似特征,提高全局运动矢量估计精度。时间比对实验表明提出的算法能够满足实时处理的要求,在分辨率为720 P时也能达到每秒处理帧频大于60帧;有效能力对比实验中,新方法的重复度均高于65%,明显提高了特征探测能力,并且在加入去抖动设计后6组实验的帧间转换精度分别提升了46.8%、30.8%、28.44%、28.1%、33.9%和53.4%,表明该方法极大地改善了抖动视频的稳定性和视频信息提取的准确度。