电子稳像的分层位平面全局运动估计算法

提出一种分层位平面匹配的快速全局运动估计算法.算法充分利用多分辨率思想,在每一层选用不同位平面进行匹配以找到运动矢量|利用层内块运动的空间相关性和层间矢量传递性,来自适应地选择搜索起始点|同时设定阈值,对运动矢量满足准确度的子块,直接结束搜索|将所得到的各子块的运动矢量作为集合,根据其统计规律,找到出现概率最大的样本作为正确的全局运动矢量.实验表明,该算法计算速度快,且与块匹配全搜索相比,其误差小于半个像素,能够达到快速准确进行全局运动估计的能力.     

超高速高性能门控型三分幅相机

提出了一种改进块匹配宏块分布排列的快速传感器电子稳像算法,通过陀螺传感器测量摄像系统的抖动,利用小范围快速块匹配算法估计局部运动矢量,再运用最小二乘法解算全局运动矢量.小范围快速块匹配算法得到的局部运动矢量准确度高,仅需部分局部运动矢量即可准确解算出全局运动矢量.基于此在保证运动矢量准确度情况下,对块匹配宏块的分布排列进行了改进,从而减少匹配宏块数量加快算法速度.通过对宏块网格模型的分析,得出对小范围快速块匹配算法进行宏块分布改进的方案,进而设计出快速传感器电子稳像算法.仿真及实验表明:运算时间提高89%左右,且算法准确度略高于改进前算法.     

基于时间序列预测的电子稳像算法研究

块匹配电子稳像算法是一种稳定性好、准确度高的电子稳像算法.块匹配算法在目标区域中从起始点到匹配点进行搜索时,需要对图像块进行反复匹配,计算量大、实时性差成为限制其应用的主要问题.本文从缩小块匹配算法搜索范围的思想出发,提出了一种利用时间序列预测来确定最优搜索起始点的电子稳像算法.根据图像序列全局运动矢量的内部统计特性,选择合适的时间序列模型;采用AIC准则和Durbin-Levinson递推算法估计模型的阶次和参量,并通过残差检验对模型进行检验和更新.利用建立的时间序列模型和历史数据对当前时刻全局运动矢量进行最优预测,并将其作为搜索起点来进行下一步精确搜索.实验结果证明,时间序列预测方法有效缩小了块匹配算法的搜索范围,使计算速度得到较大幅度的提高,并可直接推广到其它电子稳像算法中.     

改进块匹配宏块的快速传感器电子稳像方法

提出了一种改进块匹配宏块分布排列的快速传感器电子稳像算法,通过陀螺传感器测量摄像系统的抖动,利用小范围快速块匹配算法估计局部运动矢量,再运用最小二乘法解算全局运动矢量.小范围快速块匹配算法得到的局部运动矢量准确度高,仅需部分局部运动矢量即可准确解算出全局运动矢量.基于此在保证运动矢量准确度情况下,对块匹配宏块的分布排列进行了改进,从而减少匹配宏块数量加快算法速度.通过对宏块网,格模型的分析,得出对小范围快速块匹配算法进行宏块分布改进的方案,进而设计出快速传感器电子稳像算法.仿真及实验表明:运算时间提高89％左右,且算法准确度略高于改进前算法.     

一种基于自适应补偿的快速帧速率上转换算法

提出一种基于自适应补偿的快速帧速率上转换算法.算法在塔型结构数据上进行运动估计并利用相邻块运动矢量对上层传递矢量进行修正,减少计算量的同时获得了平滑的运动矢量场.在匹配搜索过程中采用动态调整搜索窗策略,避免了过搜索和搜索不足的问题.运动补偿克服了传统的补偿算法仅采用一种插值方法的不足,根据运动矢量的可靠性分别采用了3种不同的插值方法.为了减少块边缘的失真,采用了重叠块运动补偿的插值方法.在遮挡区域,设计了加权多候选运动矢量插值方法,对前后两帧补偿结果分别赋予不同的权值以减少失真.实验结果表明,该算法与传统算法相比不仅可以大幅度降低计算量,且插值图像的质量有所提高.     

用于像移探测的位平面匹配改进算法

像移探测作为电子稳像的重要的步骤,直接影响稳像的精度和速度。针对传统位平面匹配算法的局限,对各阶位平面的探测效果进行了综合测试,并提出了一种改进方案,即对位平面4,5经过下采样后组成一个合成位平面来进行匹配。研究了匹配块大小变化对位平面算法的影响,比较了相邻帧和固定帧两种参考帧选取方法的优劣。实验表明,改进的位平面匹配算法提高了位平面算法的稳健性,在没有增加计算量的前提下,算法精度达到了与全搜索块匹配法相当的程度。     

MPEG-4视频中运动背景下的目标检测算法

针对由运动摄像机捕获的MPEG-4视频流中的运动目标检测问题,提出了一种直接利用压缩视频码流进行全局运动估计的新算法.算法从全局运动估计的基础出发,利用背景宏块运动相似性的特点快速建立背景宏块集合并采用常用的四参数全局运动估计模型估计运动参数.最后,计算运动矢量残差,通过对运动矢量残差的筛选检测运动日标.算法利用MPEG-4码流中蕴含的运动信息.不需要对压缩流完全解码,较大地提高了检测效率;进一步改善了检测效果.实验验证了提出的全局运动估计算法的检测效率和检测效果.     

一种视频编码的运动估计的快速算法

在活动图象名称编码中,广泛采用了块匹配运动补偿技术,其中全局搜索法（ＦＳ）性能最好,但速度最慢。传统的运动估计快速算法如三步法、四步法、较分搜索运动估计法而言减少了搜索次数。其于低分辨率图象的运动估计能大大降低计算能大大降低计算量。文章将二者结合起来,不仅减少计算量,而且信噪比ＳＮＲ介于全局搜索法和三步法之间。     

一种基于特征块匹配的电子稳像方法

提出一种基于特征模块匹配的图像稳定的算法.该算法自适应地搜索图像特征模块进行后续稳像操作.在稳像过程中,提出一种图像特征匹配的快速算法,匹配得到局部运动矢量,然后对匹配结果进行一致性检验,最后采用可变窗口宽度的均值滤波法对图像序列运动轨迹滤波.试验表明,该算法有效地稳定了图像序列,并且满足实时性要求.     

基于多帧参考的整帧丢失错误掩盖算法

为了解决视频传输中整帧图像丢失的误差扩散问题,提出了一种基于多参考帧运动矢量外推的整帧丢失错误掩盖算法.通过对多个参考帧的运动矢量外推得到丢失块的候选运动矢量集;采用多参考帧边界匹配准则进行丢失帧和其后续帧的误差估计,并选择丢失块的最优运动矢量;最后采用自适应的重叠块运动补偿方法重建丢失帧.实验结果表明,该算法在恢复整帧图像的同时,有效地降低了视频序列中整帧图像丢失的误差扩散影响,与已有算法比较,该算法恢复的视频序列平均峰值信噪比提高了0.5～1dB,具有更好的错误掩盖效果.     

The application of improved block-matching method and block search method for the image motion estimation

Image block matching is one of the motion estimation methods for video inter-frame coding and digital image stabilization. The methods used for matching and searching will greatly affect the accuracy and speed of block matching. The block matching method based on the oblique vectors is suggested in this paper where matching parameters contain both horizontal and vertical vectors in the image blocks at the same time. Improved matching information can be obtained after making correlative calculations in the oblique direction. A novel search method of matching block based on the idea of simulated annealing is presented in this paper to improve the searching speed, accuracy and robustness in the fast operation of the block-matching motion estimation. The simulated annealing algorithm can easily escape from the trap of local minima effectively. With the two methods the block matching can be used for motion estimation at the real-time image processing system and high estimation accuracy can be achieved. An image stabilization system based on DSP (Digital Signal Processing) system is developed to verify this algorithm. Results show that both the matching accuracy and the search speed are improved with the methods presented.     

基于红外图像编码的运动预测相关算法

针对运动预测过程中单一匹配准则的不足提出了一种联合匹配准则，能较好地处理小目标的灰度渐变。基于运动预测过程中不断进行的像素块匹配运算提出了一种新的像素抽样算法，该算法使得运算量大大减少，匹配精度逼近全像素匹配算法。这两种技术的综合应用使得整个运动预测效率有较大改善。     

A Hybrid Rao-NM Algorithm for Image Template Matching

This paper proposes a hybrid Rao-Nelder–Mead (Rao-NM) algorithm for image template matching is proposed. The developed algorithm incorporates the Rao-1 algorithm and NM algorithm serially. Thus, the powerful global search capability of the Rao-1 algorithm and local search capability of NM algorithm is fully exploited. It can quickly and accurately search for the high-quality optimal solution on the basis of ensuring global convergence. The computing time is highly reduced, while the matching accuracy is significantly improved. Four commonly applied optimization problems and three image datasets are employed to assess the performance of the proposed method. Meanwhile, three commonly used algorithms, including generic Rao-1 algorithm, particle swarm optimization (PSO), genetic algorithm (GA), are considered as benchmarking algorithms. The experiment results demonstrate that the proposed method is effective and efficient in solving image matching problems.     

A coarse-to-fine kernel matching approach for mean-shift based visual tracking

Mean shift is an efficient pattern match algorithm. It is widely used in visual tracking fields since it need not perform whole search in the image space. It employs gradient optimization method to reduce the time of feature matching and realize rapid object localization, and uses Bhattacharyya coefficient as the similarity measure between object template and candidate template. This thesis presents a mean shift algorithm based on coarse-to-fine search for the best kernel matching. This paper researches for object tracking with large motion area based on mean shift. To realize efficient tracking of such an object, we present a kernel matching method from coarseness to fine. If the motion areas of the object between two frames are very large and they are not overlapped in image space, then the traditional mean shift method can only obtain local optimal value by iterative computing in the old object window area, so the real tracking position cannot be obtained and the object tracking will be disabled. Our proposed algorithm can efficiently use a similarity measure function to realize the rough location of motion object, then use mean shift method to obtain the accurate local optimal value by iterative computing, which successfully realizes object tracking with large motion. Experimental results show its good performance in accuracy and speed when compared with background-weighted histogram algorithm in the literature.     

Improvement of a new rotation function for molecular replacement by designing new scoring functions and dynamic correlation coefficient

A previously published new rotation function has been improved by using a dynamic correlation coefficient as well as two new scoring functions of relative entropy and mean-square-residues to make the rotation function more robust and independent of a specific set of weights for scoring and ranking. The previously described new rotation function calculates the rotation function of molecular replacement by matching the search model directly with the Patterson vector map. The signal-to-noise ratio for the correct match was increased by averaging all the matching peaks. Several matching scores were employed to evaluate the goodness of matching. These matching scores were then combined into a single total score by optimizing a set of weights using the linear regression method. It was found that there exists an optimal set of weights that can be applied to the global rotation search and the correct solution can be ranked in the top 100 or less. However, this set of optimal weights in general is dependent on the search models and the crystal structures with different space groups and cell parameters. In this work, we try to solve this problem by designing a dynamic correlation coefficient. It is shown that the dynamic correlation coefficient works for a variety of space groups and cell parameters in the global search of rotation function. We also introduce two new matching scores: relative entropy and mean-square-residues. Last but not least, we discussed a valid method for the optimization of the adjustable parameters for matching vectors.     

基于Harris角点全局运动估计的场景锁定算法

场景锁定技术是视频跟踪领域的一个关键技术,需要对图像的全局运动进行估计,常用的运动估计算法由于计算量大、对噪声敏感等因素很难得到实际应用。为了减少运动估计的计算量,提高全局运动估计的精度,提出了一种基于Harris角点全局运动估计的场景锁定方法。将图像分成4×4的16个块,选取每个块中响应值最大的角点,以参考图像角点周围矩形块与待匹配图像进行匹配,然后利用RANSAC算法对角点进行一致性检测,利用最小二乘法解算全局运动参数,最后计算图像之间的累积运动。实验结果表明,该算法运动估计精度高,稳定性好,能较好地实现场景锁定。