基于DCT域的系数分级预判与视觉优化算法研究

发布时间：2018-05-01 08:31

本文选题：H.264 + 零值量化DCT系数　；参考：《江西理工大学》2015年硕士论文

【摘要】：随着信息时代的发展,高清晰度的视频越来越受人们的青睐,视频编码技术也显得日益重要。H.264/AVC是VCEG和MPEG两大国际组织联合推出的视频编码标准,也是目前最流行的视频标准之一。该标准吸收了以往视频标准的优点,同时引进了多种先进的编码技术,使得其编码效率比以往标准至少提高了一倍以上。然而与此同时,也带来了大量的复杂编码运算量。因此,针对H.264/AVC视频编码算法进行性能优化,在保证编码质量的前提下降低编码的复杂度具有重要的现实意义和应用价值。本文围绕H.264/AVC标准中DCT域的编码算法,主要做了以下内容的研究工作:(1)针对运动补偿后的残差块经过4×4整数离散余弦变换(DCT)及量化过程后出现大量的零值量化系数(ZQDCT)而导致的冗余计算问题,在分析提取出最佳全零块判决阈值的基础上,结合高斯分布模型和DCT系数能量“Z”形分布特性,提出了基于该新阈值的零值量化系数分级预判算法,在残差块DCT变换之前按分级模式预判出ZQDCT,从而跳过其变换量化操作。实验结果表明,与标准算法相比,该算法明显降低了编码数据量,同时减少了最高达42%的离散余弦变换(DCT)、量化(Q)、反量化(IQ)以及离散余弦反变换(IDCT)过程计算时间。(2)结合人类视觉系统特性,将基于DCT域的恰可察觉失真(JND)模型应用于H.264标准算法中。由于人眼是视频图像的最终接收者,其对于DCT系数中低频分量具有更大的敏感性,而高频部分可能含有较多的视觉冗余系数。JND表示人眼不能察觉的最大图像失真,反映了图像中不可感知的视觉冗余信息。算法根据不同的4×4 DCT变换块计算出相应的JND阈值,对每个变换后的DCT系数进行过滤,其中对应位置比JND阈值小的DCT系数直接舍弃,从而去除视频图像中存在的主观视觉冗余。同时对舍弃后相应的DCT系数跳过其量化等操作,节省运算量。实验仿真结果表明,该JND模型能节省编码码率,同时对视频图像质量并不影响,对视频编码中的视觉冗余优化取得了良好的效果。本文通过对H.264标准中的DCT变换与量化过程的研究,利用分级预判算法将ZQDCT系数进行提前预判并跳过其DCT部分的编码,节省了运算量;另外从人类视觉系统的角度,对DCT域的视觉冗余进行了分析并提出算法进行过滤相关冗余系数,优化了编码的视频图像质量。
[Abstract]:With the development of the information age, high definition video is becoming more and more popular. Video coding technology is becoming more and more important..H.264/AVC is the video coding standard combined by two international organizations of VCEG and MPEG. It is also one of the most popular video standards at present. The advanced coding technology makes the coding efficiency more than twice as much as the previous standard. At the same time, it also brings a lot of complex coding operations. Therefore, it is of great practical significance to optimize the performance of the H.264/AVC video coding algorithm and reduce the complexity of the low coding in the premise of guaranteeing the quality of the coding. Based on the coding algorithm of the DCT domain in the H.264/AVC standard, the main research work is done in the following aspects: (1) the optimal total zero is extracted from the 4 x 4 integer discrete cosine transform (DCT) after the motion compensation and a large number of zero quantization coefficients (ZQDCT) after the quantization process. On the basis of block decision threshold, combining the Gauss distribution model and the "Z" distribution characteristic of DCT coefficient energy, a zero quantizing coefficient classification algorithm based on the new threshold is proposed. The ZQDCT is prejudged by the hierarchical mode before the DCT transformation of the residual block, and the transform quantization operation is skipped. The experimental results show that the algorithm is compared with the standard algorithm. It reduces the amount of coded data, and reduces the maximum 42% of discrete cosine transform (DCT), quantization (Q), inverse quantization (IQ) and discrete cosine inverse transform (IDCT) process computing time. (2) the exact perceptible distortion (JND) model based on the DCT domain is applied to the H.264 standard algorithm based on the characteristics of the human visual system. The human eye is a video map. The final receiver of the image is more sensitive to the low frequency component of the DCT coefficient, and the high frequency part may contain more visual redundancy.JND to represent the maximum image distortion that the human eye can't detect. It reflects the unperceived visual redundancy information in the image. The algorithm calculates the corresponding JND threshold based on the different 4 x 4 DCT transform blocks. Value, filtering the DCT coefficients after each transformation, in which the DCT coefficients with smaller corresponding positions than the JND threshold are abandoned directly, thus the subjective visual redundancy in the video image is removed. At the same time, the corresponding DCT coefficients skip their quantization and other operations to save the computation. The experimental simulation results show that the JND model can save the code rate and same code rate. It does not affect the quality of video image, and has good effect on visual redundancy optimization in video coding. Through the study of DCT transformation and quantization in the H.264 standard, the ZQDCT coefficient is prejudged in advance and skipped the coding of the DCT part by the hierarchical prejudgement algorithm, and the computation is saved. In addition, the human visual system is also obtained. From the point of view of the system, the visual redundancy in the DCT domain is analyzed, and the algorithm is proposed to filter the related redundancy coefficient and optimize the quality of the encoded video image.

【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN919.81

【相似文献】