一种保护图象轮廓细节的自适应亚抽样和插值方法

为在图象编码时减少原始图象数据，以提高压缩比和编码速度，提出了一种新的自适应亚抽样与插值的方法及可应用于多类分块图象编码的算法，该方法首先将图象划分成互不重叠的块，然后计算每个块的水平梯度与坚直梯度，再根据图象块的两个方向梯度值，将图象块分为平块、水平轮廊块、竖直轮廊块和高细节块等４类，同时对每类图象块采用不同的亚抽样与插值方法以减少原始图象数据。模拟结果表明：相对于其它的亚抽样与插值方法，该算法能够很好地保护图象中的轮廊及细节信息，从而极大地提高了重建图象的质量，尤其是图象的主观质量。另外，对于细节较多的图象，该算法在保持相当压缩比的同时，ＰＳＮＲ也提高了３．９ｄＢ；而对于细节较少的图象，压缩比与ＰＳＮＲ都略有提高。

A Novel Adaptive Decimation/Interpolation Method with Contour Preserving Property

A novel adaptive decimation/interpolation algorithm, which can be applied in all kinds of blocking image coding algorithm, is proposed. The image is first divided into non overlapping blocks. Then the horizontal and the vertical gradients of each block are calculated. According to the two gradient values, the image blocks are classified into four groups, namely smooth block, horizontal contour block, vertical contour block and high detail block. Then different decimation methods are applied to different class of blocks to reduce the original data quantity. Experimental results show, compared to other decimation/interpolation algorithms, the proposed algorithm can well preserve the contour and other detail information, thus substantially improve the quality of the reconstructed images, especially the subjective quality. The proposed method increases the PSNR by 3 9 dB while keeping comparable compression ratio for high detailed images. For low detailed images, compression ratio and PSNR are all slightly improved.