Structure and Texture-Based Fullbore Image Reconstruction

This paper presents a novel method to generate fullbore images by incorporating an advanced inpainting technique with Multi-point Statistics approach. Inpainting is a general term used in image processing that refers to filling in gaps or repairing damaged images. The inpainting technique is used to detect dips and capture the trend of the borehole image logs. The extracted smooth-trend maps are then fed into the statistical simulation to guide the construction of high-resolution textures while honoring the original borehole image data, leading to a seamless reconstruction of 360° fullbore images. Testing of the proposed method on various borehole image textures has demonstrated that it is a reliable and robust way to perform fullbore image reconstruction in deepwater sediments, which are dominated by laminated or high-angle features, carbonate environments with typical vugs and fractures, and fluvial systems with channel-fill deposits. 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, and complex heterogeneity analysis and accurate porosity estimation in carbonates.