基于小波变换的地震勘探数据压缩的工程分析

Some practical aspects of seismic data compression based on wavelet transform

为满足对地震勘探数据压缩的特殊要求 ,论文研究了直接影响压缩质量的几个因素 :地震数据小波系数的分布情况、小波分解层数及边界处理。从实际处理结果可见 ,根据数据小波系数分布特点及结合零树量化编码对地震数据压缩可以满足限失真提高压缩比的工程要求。实验表明 ,压缩 1/ 10后其能量损失为 0 .2 %左右 ,由差剖面可以看出 ,损失的主要为噪声。小波分解层数与提高压缩比和压缩质量并不成比例关系 ,一般分解 3层即可满足工程要求 ,这样可大大提高压缩速度。对较大的地震剖面 ,通过一定的边界处理采用分块处理办法 ,可既不影响压缩效果、又提高压缩速度。改进后的算法更适于实际应用。

Seismic data compression must be done with few losses, so that the useful information is accuarately retained for later processing and interpretation. This paper discusses some key factors that affect the performance of seismic data compression based on wavelet transforms, such as the distribution of wavelet transform factors, proper levels of decomposition and the effect of boundaries. Practical experiment results indicate that seismic data compression considering the distribution characteristics of their wavelet coefficients and using zero tree quantization coding satisfies the engineering demands of improved compression ratio with minimal distortion. The energy loss in the experiment was about 0.2% after compressing 10 times and the loss was mainly undesired noise. Our analysis showed that three layer wavelet decomposition can satisfy engineering demands with good processing speed. For large sections, an efficient block partition method is proposed with both good compression performance and speed. Practical experiments showed that the proposed method works well.