多小波基的设计与地震图像并行融合实现

Design of a Multi Wavelet and Research of Parallel Seismic Image Fusion

面向地震图像解释中的具体问题,开展图像融合算法研究,增强目标识别的准确性,提高对地震图像的解译能力.设计了一种具备对称性、短支撑性、二阶消失矩和正交性的高性能多小波基,提出了其并行融合算法实现.算法考虑了对配准误差的适应性问题,提高了图像空间频率.从实验结果来看,图像的整体和细节特征比单小波融合方法有较大的提高.通过将地震图像融合技术与并行程序设计相结合,在多核处理器下采用虚拟节点技术搭建多机并行图像融合平台.实验结果表明:针对多核系统的并行融合方法省去了并行处理的数据寻址时间,相比并行化前,算法执行时间随着节点数增加而减少,并且各并行操作更为规范、简洁,符合算法操作规则性强的实现特点.

A type of symmetry, short support, two vanishing moments and orthogonally high performance multi wavelets was designed, and the parallel fusion algorithm that was beneficial for both the smooth regions of seismic image signal and that of the edge, texture component was put foreworded. The experiments and the quantitative measurement of the fusion results show that the fusion performance is better than signal wavelet and suitable for seismic image fusion application. By combining the seismic image fusion technology with the parallel program design, using the virtual node technology, the multi machine parallel image fusion platform in the multi-core processor was built. Results show that parallel fusion method based on multi-core system saves the data addressing time. Compared with parallel, the execution time of this algorithm decreases with the increase of the number of nodes, and the parallel operation is more standardized, concise, and in accordance with the characteristics of the strong algorithm operating rules.