Structure- and Texture-Based Fullbore Image Reconstruction

Abstract

Borehole image logs are produced by tools lowered into a well. Such logs provide oriented electrical and acoustic maps of rocks and fluids encountered in the borehole. Electrical borehole images, acquired in either water-based (conductive) or oil-based (nonconductive) muds, are generated from electrodes arranged in fixed patterns on pads pressed against the borehole wall. Depending on borehole diameter, gaps nearly always occur between pads. Because of these gaps, it is common to have nonimaged parts of the borehole wall. The existence of gaps in pad-based borehole images hinders efficient geological interpretation and accurate formation evaluation. A novel method to generate fullbore images combines an inpainting technique with FilterSim: a continuous-variable multipoint statistical approach. The inpainting algorithm detects dips and captures the trend of borehole image logs. The extracted smooth trend maps are fed into FilterSim to guide the construction of high-resolution textures that honor the original borehole image data, leading to seamless reconstruction of 360$^{\circ }$ fullbore images. The proposed method has been tested using various borehole image patterns and proves to be a reliable and robust way to perform fullbore image reconstruction. The reconstructed fullbore images facilitate improved visualization and interpretation of borehole image logs in various ways, including automated dip picking for fractures and bedding planes, thin-bed analysis in deepwater formations, complex heterogeneity analysis, and accurate porosity estimation in carbonates.