基于立体块的压缩视频感知技术研究

发布时间：2018-04-03 09:04

本文选题：压缩视频感知　切入点：低采样率　出处：《西安电子科技大学》2014年硕士论文

【摘要】：压缩感知突破了奈奎斯特采样速率的限制,实现了对于信号信息的直接采样。压缩视频感知将压缩感知引入到视频编解码中,实现了一种新颖的视频编码方式。现有的压缩视频感知方案大都基于帧或者基于块对视频信息进行编解码,然而对于三维视频数据,基于立体块的压缩视频感知方案在空间和时间上跨越了全部信息内容,是一种更为直观的处理多维数据的方式。可惜的是,在进行同时多帧测量操作中需要复杂且昂贵的空间时间光调制,使其在硬件实现方面被认为是不切实际的。本文改变视频立体块中同时处理多帧数据的方式,提出了一种适用于无线多媒体传感器网络的基于立体块的压缩视频感知编解码系统。编码端采用低复杂度的逐帧测量方式,使其适用于底层节点能量受限的无线网络;利用采样率的重新分配处理增加了首尾两帧的测量值信息,提高了首尾两帧的重建质量,并且以质量有所改善的首尾两帧作为辅助信息对中间帧进行重建,使其重建质量也有所提高,改善了系统的整体性能。解码端通过对逐帧测量得到的测量值进行全局置乱,使其置乱后的测量向量接近于同时多帧测量所得到的测量向量,有效地提高低采样率下系统重建质量。仿真结果表明,系统具有高质量的解码性能,尤其在低采样率情况下,相比于逐帧测量多帧重建方案提高了近4~5dB;相比于多帧测量多帧重建方案也提高了0.4~0.9dB左右。可见,所提方案在具有硬件可实现性的前提下,解码性能得到有效改善。
[Abstract]:Compression sensing breaks through the limit of Nyquist sampling rate and realizes direct sampling of signal information.Compressed video perception is introduced into video coding and decoding, and a novel video coding method is implemented.Most of the existing compressed video sensing schemes are based on frames or blocks to encode and decode video information. However, for three-dimensional video data, the compressed video sensing schemes based on three-dimensional blocks span all the information content in space and time.Is a more intuitive way to process multidimensional data.Unfortunately, complex and expensive space-time light modulation is required for simultaneous multi-frame measurement, which is considered to be impractical in hardware implementation.This paper changes the way in which multi-frame data is processed simultaneously in a video stereo block, and proposes a compressed video perceptual coding and decoding system based on stereo block for wireless multimedia sensor networks.The coding end adopts the low complexity frame by frame measurement method, which makes it suitable for the wireless network with limited energy at the bottom node, increases the measurement value information of the first and last frames by using the rescheduling of the sampling rate, and improves the reconstruction quality of the first and the last frames.And the first and last frames with improved quality are used as auxiliary information to reconstruct the middle frame, so that the reconstruction quality is improved, and the overall performance of the system is improved.At the decoding end, the measured values obtained by frame by frame are global scrambled, which makes the measurement vector close to the measurement vector obtained by simultaneous multi-frame measurement, which effectively improves the quality of system reconstruction at low sampling rate.Simulation results show that the system has a high quality decoding performance, especially in the case of low sampling rate, compared with the frame-by-frame measurement multi-frame reconstruction scheme, the improvement is nearly 4db, and compared with the multi-frame measurement multi-frame reconstruction scheme, the 0.4~0.9dB is also improved.It can be seen that the decoding performance of the proposed scheme is improved effectively on the premise of hardware realizability.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN919.81

【相似文献】