A fast algorithm for super-dimension reconstruction of core images based on multithreading
To tackle the limitation of the super-dimension algorithm based on adjacent block matching,where only allows one block to be reconstructed at a time during the reconstruction process,and the time-consuming dictionary search significantly hampers its efficiency,this paper proposesed two methods to opti-mize the computational speed of super-demension algorithms.Firstly,the sub-region reconstruction approach was applied to the super-dimensional algorithm based on adjacent block matching,introducing a region-based parallel reconstruction algorithm that enables simultaneous reconstruction of multiple blocks;Secondly,in the process of searching for each dictionary element,a dictionary parallel search matching method is em-ployed to accelerate its process;Finally,these two algorithms are combined,and 3D reconstruction based on 2D images was preformed for multiple times through the dictionary generated by the three different porosity training images.Through comparing the reconstruction results of the proposed algorithm,the adjacent block matching-based super-demension algorithm,and some traditional reconstruction algorithms,along with statis-tical feature parameters of the real three-dimensional structure of rock cores,and by comparing the reconstruc-tion time-consuming of different algorithms,the computational effectiveness of the improved super-demension algorithm presented in this paper was validated.
3D reconstructionSuper-dimension algorithmPorous mediumParallel algorithmMulti-threading
国家科技期刊平台
NSTL
万方数据
