论文部分内容阅读
针对现有降噪采用时空域滤波、去块效应采用环路滤波/后处理这种“分而治之”方案的缺点,通过分析降噪的基本技术、低码率视频应用中块效应产生的原因和去块效应的常用方法,从滤除不需要或相对不重要的高频离散余弦变换(DCT,discrete cosine transform)系数角度提出了降噪和去块效应可同时处理的设想。新提出的预处理算法是根据图像纹理特性、运动情况和码率约束等,在图像的不同区域采用不同强度的自适应双边滤波,这样不仅去噪,滤除不重要的细节以便于高效压缩,避免块效应的产生,同时也可起到一定程度的码率控制效果。
For the existing noise reduction using spatio-temporal filtering, deblocking effect using loop filtering / post-processing of this “Divide and conquer ” the shortcomings of the program, through the analysis of the basic technology of noise reduction, low bit rate video application block effect causes And the common method of deblocking effect, the idea that noise reduction and deblocking can be dealt with simultaneously is put forward from the point of filtering the unwanted or relatively unimportant coefficient of DCT (Discrete Cosine Transform). The new preprocessing algorithm is based on the image texture characteristics, motion and bit rate constraints, in different regions of the image using different intensity adaptive bilateral filtering, so not only denoising, filtering out unimportant details in order to facilitate efficient compression, To avoid the block effect, but also can play a certain degree of rate control.