Computer Vision and Digital Processing Techniques for the Concurrent Removal of artifacts in Acoustic Image Logs

Summary This study introduces a methodology leveraging Computer Vision (CV) and Digital Image Processing (DIP) to simultaneously detect and reduce artifacts in acoustic borehole image logs, enhancing reservoir characterization. Common artifacts, including breakouts and pad marks, were mitigated using advanced techniques applied together to the same images, ensuring consistent quality and accuracy. Both methods were applied concurrently to the same images, enabling simultaneous reduction of breakout and pad mark artifacts.Breakout removal combined noise reduction, edge detection, and the Segment Anything Model (SAM) to create precise masks. Pad marks, addressed using Fast Fourier Transform (FFT) and band-rejection filtering, were suppressed by isolating and removing interfering frequencies. Combined processing of both artifact types preserved geological features while improving image clarity. A z-score based method of normalization, corrected abrupt color variations caused by equipment anomalies, further enhancing data reliability. Results showed a 4.45% average reduction in vug porosity calculation error after filtering. Filtering pads first produced clearer images while maintaining accuracy, facilitating better analysis. The methodology improved both visual clarity and analytical precision, reinforcing the value of acoustic image logs in petrophysical property estimation.