基于最小二乘自适应分解插值的图像编码

Image Coding Based on Least Square Adaptive Splitting and Interpolation

为了改善JPEG在低比特率条件下重建图像的质量,以及提高"下采样—JPEG—插值"方案的率失真性能,提出了利用最小二乘法确定双线性插值的最优插值顶点,然后对这些顶点值进行编码,解码的顶点值被用来进行双线性插值以重建原始大小的图像;在预定保真度的情况下,根据局部图像的插值误差是否满足要求来进行插值图像块的自适应分解。实验结果表明,该方法改善了JPEG低比特率条件下的重建质量,率失真性能整体优于"下采样—JPEG—插值"方案。

In order to improve the reconstructed image quality at low bit rate and the rate -distortion performance of ＆quot;down-sampling-JPEG-interpolation＆quot;schemes, an image coding algorithm was proposed that determines the optimal interpolating ver -tices of bilinear interpolation using least square method .Then the vertices were coded;the reconstructed image was obtained by bilinear interpolation using the decoded vertices .In addition, the interpolation image block will be split further adaptively accord-ing to whether the interpolation error of the local image meets the predefined fidelity requirement .Experimental results show the algorithm can improve the reconstructed image quality of JPEG at low bit rate;the whole rate-distortion performance outperforms that of the ＆quot;downsampling -JPEG-Interpolation＆quot;schemes.