高空间冗余图像DCT压缩产生的几何衰变

Geometric degradation produced during DCT compression of images with high spatial redundancy

离散余弦变换 (DCT)压缩图像的客观质量评价通常采用均方误差 (MSE)和峰值信噪比 (PSNR) ,而缺乏对图像几何衰变的考虑 ,这正是基于位置信息的图像测量系统的关键。针对这一问题 ,对两幅高空间冗余图像进行DCT压缩实验研究 ,结果表明 :用 4× 4分块 ,1:2压缩比时 ,MSE和PSNR变化不大 ,但有 1或 2个像元的几何衰变。与一般情况下光学测量系统总体误差只有 1～ 2个像元比较 ,在因压缩而产生像元级的位置变化时 ,将会严重影响测量精度 。

The objective image quality evaluation of DCT(discrete cosine transform) compressed images is usually achieved using MSE(mean square error) and PSNR(peak signal to noise ratio) with the geometric degradation of images out of consideration, which is the very key to the optical measurement system based on position information. DCT compression experiemnts made with two images with high spatial redundancy show that MSE and PSNR do not change much under 4×4 processing block and 1∶2 compression ratio, but there is a geometric degradation of 1 or 2 pixels; considering the total error of the optical measurement system is generally only 1 or 2 pixels, a change in position of more than 1 pixel resulting from DCT compression will have a significant influence on the accuracy of the measurement system. A good deal of attention should, therefore, be paid to geometric degradation during DCT compression.