Description: 本文的工作是实现一个视频监控系统的软件核心部分。在内容上首先对
H. 263视频编码标准进行叙述，接下来讨论实现快速运动估计的方法，针对块
匹配运动估计给出了一种改进的钻石搜索算法代替原有的三步搜索法。实验
结果证明，经算法改进优化的新钻石搜索法优于原先的三步攫索法，大大提
高了程序运行的效率。随后简单介绍了C64xx系列DSP和视频数据的AVI存
储方式，研究了用软件的方法实现 H. 263编码标准，以及将其移植到 DSP后
的优化问题，最终达到了在 DSP上实时编码的要求。-This article is the realization of a video surveillance system in the core of the software. First of all, the contents of H. 263 video coding standards are described, followed by discussion of fast motion estimation methods, for block-matching motion estimation is given an improved search algorithm in place of the original diamond three-step search method. Experimental results show that by the algorithm to improve the optimization of the new diamond search method is superior to the original three captured Sofa, greatly improve the efficiency of the program is running. Then briefly introduce the DSP Series C64xx and AVI video data storage, research using the software method H. 263 coding standard, and its transplantation to the DSP after the optimization problems, and ultimately reached in the DSP real-time encoding requirements. Platform: |
Size: 2752512 |
Author:liu |
Hits:
Description: 在快匹配运动估计算法中,它的计算复杂度可以描述成:运算量=搜索点×块匹配准则.该表达式可以这样解释,在快匹配准则确定的情况下,如果一个匹配点的单位运算量确定了,这时动动估计的动算量就有直接取于搜索的点数,也就是取决于运动估计的搜索算法，实现了三步搜索法-Match the fast motion estimation algorithm, its computational complexity can be described as: Operational × volume = search block-matching criteria points. The expression can be interpreted in such a way, the fast matching criterion to determine the circumstances, if a match point of the unit operator determine the amount, and then moving moving moving estimate of the volume operator has direct access to the points in the search, that is, depends on the motion estimation search algorithm, to achieve a three-step search method Platform: |
Size: 1024 |
Author:罗鹏 |
Hits:
Description: 本论文首先系统地介绍了视频压缩编码的基本原理和方法，以及各国际标准
化组织提出的压缩标准；其次分析了JPEG和MPEG-4的主要技术，据此提出
MPEG-4简单框架编码器的软件实现方案，给出流程图，在VC++环境下用C语言
程序实现了JPEG压缩及MPEG-4简单框架的视频压缩功能，采用测试序列验证
无误，并对两种标准的压缩比进行试验对比，突出压缩的关键技术为DCT和运动
估计算法；接着深入研究了DCT变换、运动估计和补偿算法并进行了优化改进，
给出了用于快速匹配块的自适应十字模式搜索算法，并给出了对各种块匹配估计
算法的仿真结果。然后详细介绍了TI公司TMS320C6711DSP的软硬件特性，并
在此基础上提出系统的硬件实现方案。接着介绍了基于系统软硬件环境的移植问
题和优化方法，并针对算法移植和仿真过程中的问题进行了说明。
最后就课题设计过程中研究的结果予以总结，并阐述了以后课题研究的要点。-In this thesis,an introduction to video compression principal,methods and central
international video coding standards is done firstly.Then analyzing the core techniques
of JPEG and MPEG-4 and proposing a scheme of MPEG-4 SP encoder,the coding
function of JPEG and MPEG-4 is achieved in C successfully with test sequence Then
compare the compressed ratio of two standards,point out the key of compression is DCT
and Motion Estimation algorithm.Afterwards,studying the DCT/Motion
Estimation/Motion Compensation algorithm in detail and giving an adaptive rood
pattern search for fast Block-Matching motion estimation.Then the simulation of several
motion estimation algorithms based block-matching is given.Accordingly,a scheme of
a DSP-based video compression system is put forward,with the features of TI’s
TMS320C6711 DSP.At last,the transplant and optimization are considered on the basis
of soft and hard environments,also the simulation process.
In the end,the summarization about the Platform: |
Size: 1805312 |
Author:father2002 |
Hits:
Description: H.264中的运动估计中的BMA（块匹配算法），共有三种方法，三步搜索法，全局搜索法，二维对数搜索法-H.264 motion estimation in the BMA (fast matching algorithm), a total of three methods, three-step search method, global search method, two-dimensional logarithmic search method Platform: |
Size: 7340032 |
Author: |
Hits:
Description: 摘 要：基于块的运动估计是视频压缩国际标准中广泛采用的关键技术。在对目前运动估计快速块匹配算法研究的基础上，描述了运动估计
的原理；揭示 了在图像质量、搜索速度和压缩码率等方面提高算法效率时存在的3 类主要问题 ：初始搜索点的选择、匹配准则和搜索策略 ；
分另 q 阐述了目前常用的解决这3 类问题的方法，并进行了比较和分析 ；提出了对运动估计算法的一些展望。-Abstract: The block-based motion estimation is a video compression standard is widely used in international key technology. In the current fast block matching motion estimation algorithm based on the principles described in the motion estimation revealed in image quality, the search speed and compression rate and so improve the efficiency exists in the 3 principal issues: the initial search point selection, matching criteria and search strategies points q describes the other currently used to solve the 3 class problem, and compared and analyzed proposed motion estimation algorithm on some prospects. Platform: |
Size: 182272 |
Author:zjl |
Hits:
Description: 为提高基于块的运动预测的搜索精度,该文基于块匹配准则,提出最小绝对差分误差和(SADD)算法。SADD 以 SAD 为基础,利
用块匹配中相邻像素点误差之间相关性,将块内各像素误差按一定顺序进行差分,消除了相邻点误差之间的冗余。该算法更精确地表达了
块与快之间的相似程度。实验结果表明,该算法可有效降低码率,提高 PSNR 值,而且其鲁棒性好,可适用于各种序列。
-n order to improve the search accuracy performance of motion estimation based on block, this paper researches the block matching
algorithm, brings forward a new matching algorithm named the minimum Sum Of Absolute Different Differences(SADD). Based on SAD, the
algorithm SADD takes advantage of the relativity in the adjacent differences inside one block, adopts the difference operation in a certain order to
remove the redundancy, which enables the algorithm SADD to describe the similarity between two blocks much more accurately. Simulation result
shows that this algorithm can reduce the bitrate effectively and improve the PSNR performance well. And it can be adapted to kinds of video
sequences.
Platform: |
Size: 97280 |
Author:tom |
Hits:
Description: 视频编码中的运动估计，用于块匹配搜索算法，进行最佳匹配快的搜索。-Motion estimation in video encoding, for the block matching algorithm, the best match for fast search. Platform: |
Size: 2048 |
Author:刘海 |
Hits:
Description: 视频编码中的运动估计，用于块匹配搜索算法，进行最佳匹配快的搜索-Motion estimation in video encoding, for the block matching algorithm, the best match for fast search. Platform: |
Size: 2048 |
Author:刘海 |
Hits:
Description: UMHexagonS算法作为整像素的快速运动估计算法，该算法的运算量相对于快速全搜索算法可节约90 以上，同时保持较好的率失真性能。-UMHexagonS algorithm as fast integer pixel motion estimation algorithm computation with respect to the fast full search algorithm can save more than 90 , while maintaining a good rate-distortion performance. Platform: |
Size: 9216 |
Author:杨鸣人 |
Hits:
Search - Motion Estimation fast search algorithm