Research On Rock Image Preprocessing Method Based On Compressive Sensing Technology

Analysis on microscopic images for the rock thin section is one of the most common used methods for investigations on reservoir microstructures,which plays an important role in the petroleum and natural gas exploration and development.Wide view and high resolution can't be achieved simultaneously in acquisition process of the rock images.This difficulty can be overcome by image mosaic technology to some extent.Realization condition of the image mosaic is high and specific hardware for image acquisition is required.More important,related algorithms still need further improvement.These limit the application of image mosaic.Image super-resolution reconstruction method based on compressive sensing is studied and main works of the thesis are listed as follow:Characteristics of rock images and learning methods for over-complete redundant dictionary are further studied.Optimized K-SVD algorithm is selected as the method to train dictionaries in our researches.The 128 images of the core in Su 279 well in the depth of 3771.26 meters in Sugeli gas field acquired by Leica DFC450 C polarizing microscope are chosen as training samples.In these samples,the rock images with fractures,inter-crystalline holes or intergranular pores are included.Dictionaries for high resolution images and low resolution images are trained with these samples.In dictionary trainings,the convergence of K-SVD method and the dependence of calculation's amount on the size of the dictionary are investigated.Method on image super-resolution reconstruction method based on compressive sensing is studied and super-resolution reconstruction technology for single image is put forward.Super-resolution reconstruction method is based on the prediction model of sparse representations for high resolution and low resolution images.The proposed method is certified with rock images from Sugeli gas field and the reconstructed images are compared with results from cubic spline interpolation method and the method established by Yang et al.Numerical experiments indicate that the proposed method has advantages in large size rock images.Researches in this thesis provide an appropriate method to obtain high resolution rock microscopic images.Related results have high reference significance to the acquisition of wide view and high resolution rock images.Exploration works on petroleum and gas exploration and development may be promoted.