An Experimental Study on the TAT System for the HDTV Bandwidth Compression

对用于HDTV带宽压缩的TAT(Time-Axis Transform)系统进行了实验研究。4模TAT模式分配算法正确运行，通过一次预处理和TAT成功地将HDTV信号的带宽压缩到8.1 MHz，针对色度信号的带宽压缩，将其时间压缩到1/4，然后与亮度信号交替地在一行中进行多路复用。因此，用于模式分配的色度信号的块大小是亮度信号的8倍。在预处理过程中，对HDTV信号以64.8 MHz的采样频率进行采样，然后对场偏移量进行二次采样，去除一半的像素点。剩余的像素进一步减少到TAT的一半。通过将这些减少的采样数转换成采样频率为16.2 MHz的模拟信号，获得了压缩带宽为8.1 MHz的模拟信号。场偏移二次采样是一种很好的前处理方法，几乎不会引起可见的退化，2：1二次采样模式(I模式)的TAT与不使用I模式的TAT具有几乎相同的失真功率。差值仅为0.1-0.3分贝。然而，前者的重建画面在主观上略好于后者。帧间内插模式(F模式)对于整体结构非常精细的静止图像是有效的。因此，I和F模式的4模TAT适合于HDTV压缩。在本文开发的TAT系统中，分配模式是为了使重建图像中的失真功率最小。这项研究的下一步将是对模式分配进行精细控制，使其与人类视觉相匹配。

The TAT (Time-Axis Transform) system for the HDTV bandwidth compression was examined experimentally. Mode assignment algorithm of 4-modes TAT functioned correctly and the bandwidth of the HDTV signal was successfully compressed to 8.1MHz using a pre-processing and TAT.Concerning to the bandwidth compression of chrominance signals, they are time-compressed to 1/4 and then alternately multiplexed in one line with a luminance signal. Thus, the block size of the chrominance signals for the mode assignment is 8 times that of a luminance signal. However, no block distortion is observed in the reconstructed picture.In the pre-processing, the HDTV signal is sampled at a sampling frequency of 64.8MHz and then one half of the pixels are deleted by field-offset subsampling. The remaining pixels are further reduced to one half by TAT. By converting these reduced number of samples to analogue signal at a sampling frequency of 16.2MHz, an analogue signal with a compressed bandwidth of 8.1MHz is obtained. The field-offset subsampling is a good pre-processing and hardly causes visible degradation.TAT with 2:1 subsampling mode (I mode) has nearly the same distortion power as TAT without I mode. The difference is only 0.1-0.3dB. However, the reconstructed picture of the former is subjectively a little better than that of the latter. Interframe interpolation mode (F mode) is effective for still pictures with very fine structures as a whole. Therefore, the 4-modes TAT with I and F modes is suitable for the HDTV compression.In the TAT system developed here, the modes are assigned to minimize the distortion power in the reconstructed picture. The next step of the study will be a fine control of the mode assignment so that it is matched to human vision.