【作者】 姚春莲； 高丽华； 陈谊； 李炜；

【Author】 Yao Chunlian~1 Gao Lihua~1 Chen Yi~1 Li Wei~2 1 School of Digital Content,Beijing University of Technology and Business,Beijing,100037,China 2 Digital Media Laboratory,School of Computer Science and Engineering,Beihang University,Beijing,100083,China

【机构】 北京工商大学数字内容学院； 北京航空航天大学计算机学院数字媒体室；

【摘要】 帧内预测是提高Ⅰ帧图像压缩比的有效方法,现有预测算法多以块为预测单元,通过多种搜索模式,从邻近点中选择最佳匹配源,预测效果与图像特点密切相关且搜索过程运算复杂度极高, 在实时性要求较高的实际系统中很难被采用。本文提出了一种新基于图像插值的帧内预测算法。该算法将原始输入图像分割为若干子图,通过插值运算构造各子图的预测参考帧,并以子图为基本预测单元直接与参考帧进行差分编码,具有预测精度较高,计算复杂度低等优势。此外,该算法还可与其他许多编码方法相结合,以取得更好的压缩效果。更多还原

【Abstract】 Intra Prediction is a good method to improve compression ratio of I frame.At present,Block as the basic prediction unit is adopted in prediction algorithm,and best match source is achieved from adjacent point by multi search method,prediction result is relative to the character of image.Because of high computation complexity,existing algorithms are difficult to be used in real-time system.In the paper,a intra prediction method based on image interpolation is proposed.First the original input image is divided into several sub-pictures,and prediction frame is calculated by interpolation processing.Sub-picture as the basic prediction unit can be differential coded with reference frame.This algorithm has the advantage of high precision and low computation complexity.Combines with other method,the compression performance can be improved.更多还原