【作者】 王子亨；

【导师】 赵德斌；

【作者基本信息】 哈尔滨工业大学 ， 计算机科学与技术， 2009， 硕士

【摘要】 与现实世界类似,在计算技术领域同样存在数据量庞大和处理能力有限之间的瓶颈。近年来,由于数字影像技术的发展,每天都有大量的数字化视觉信息(图像、视频等)产生。然而,视觉信息数量的增长远远超过了现有的硬件能力,为应用领域带来了一定的压力。例如,图像和视频信息庞大的数据量为存储系统带来沉重负担;在传输时需要耗费大量的网络流量,同时受到网络带宽的限制,用户需要等待大量的时间。为解决这一瓶颈,有效利用现有的硬件条件,方便用户访问视觉冗余信息较为庞大的视频资料,成为近年来编码领域研究的热点。如果能在压缩系统中采用某种视觉技术来达到获取较低码率和较高视觉质量的目的,是很有意义的。为此,本文对HVS特性进行了研究,提出一种基于HVS“中心-环绕”特性、人眼亮度显著性、色度显著性、运动信息显著性等特性的视频压缩方法。视频压缩中,由于人眼实际得到的是解压缩后的图像,因此对重建图像质量的评价是人们非常关心的问题。本文在总结了各种视频质量评测方法的基础上,对基于结构失真的视频质量评测方法SSIM与传统评测方法PSNR进行了实验比较,并确定采用SSIM对本文的视频压缩方法进行评测。人眼的视锥细胞和神经细胞的分配是高度不均匀的,在小凹处密度很高,而周边区域的细胞密度则下降很快。因此HVS对视频图像的分辨率也是高度不均匀的。结合人眼的这种特性,本文对基于小凹的视频编码模型进行了研究,提出综合了以背景亮度与亮度变化梯度为基础的人眼亮度显著性、以色度变化梯度与时域色度差为基础的色度显著性、与以运动强度、运动对比度与运动矢量空间相位熵为基础的运动信息显著性三种特性的基于内容自适应foveation模型的视频编码方法。在假设关注点在视频图像中心的情况下,根据视频序列的内容对该视频不同区域的分辨率进行调整。实验结果表明,该方法能够获得较好的压缩效率。更多还原

【Abstract】 Similar to the real world, computing technology field also exists a bottleneck between amount of data and the limited processing capacity. In recent years, because of the development of digital imaging technology, every day a large number of digital visual information (images, video, etc.) is produced. However, the growth of the number of visual information has far exceeds the capacity of the existing hardware, it brought a certain amount of pressure for application areas . For example, a large amount of data of images and video take a heavy burden to the storage systems , in transmission need to spend a lot of network traffic, and be limited by the network bandwidth limitations, users need a lot of time to wait. To figure out this bottleneck, using of existing hardware effectively and user-friendly access to visual information, become a hot area of research in recent years. In this paper, author give a new method of video coding base on“center– surround”, luma salience, chorma salience, and motion salience.In video compression, because actually receipt of the human eye is the image after decompression, so the evaluation of the quality of the reconstructed image is much concerned .This paper take a summary of video quality evaluation method, and the SSIM video quality evaluation method is analyzed and realized, and use the method in this paper.The human eye cone cells and nerve cells is a highly uneven distribution, high density in a small hollow, and the surrounding areas of the cell density decreased rapidly. To combine this characteristics of the human eye, the article give a new method of video coding integrated with luma salience which integrated with the background brightness and the luma gradient, chorma salience which integrated with the chorma on the timeline and the chorma gradient, and motion salience which integrated with the motion intensity, motion contrast, and motion phasic entropy. The experimental results show that this method can have better compression efficiency.更多还原