一种新的基于率失真优化的感兴趣区域编码

New ROI coding algorithm based on rate-distortion optimization

根据EBCOT编码中率失真优化截取算法的特点,,结合最大位移法提出了以提升率失真斜率来实现任意形状ROI优先编码的方法(DRshift法)。该方法提升的对象是ROI的失真斜率,而非ROI系数,从而避免了最大位移法(Maxshift)中引入过多的位平面导致的编码效率降低的问题。实验结果表明,该算法可在接收到图像的全部编码数据之前,优先实现任意形状的感兴趣区域的高品质重建,提高了编码效率,并且,通过改变提升因子的大小可实现感兴趣区域与背景区域重建图像质量的灵活可调。

According to the Rate distortion optimization of EBCOT, and combined with the Maxshift method, an improved region-of-interest(ROI) method was proposed. By shifting the corresponding distortion-rate slopes instead of the corresponding ROT coefficients in Maxshift, this method avoided the coding efficiency defects caused by increased bit-plans in Maxshift. It could reconstruct the arbitrarily shaped ROI image with higher quality without receiving the entire image and improve the efficiency of coding. Besides, the new method enabled the flexible adjustment of compression quality in ROI and background. Experiments indicate its efficiency.