Detecting Fault Lines from Satellite Photos using Computer

Detecting Fault Lines from Satellite Photos using Computer Vision Ryan Meyer

Project Goals and Methods ● Purpose: Manually analyzing photos can be slow, and geologists may miss hard to identify features ● Goal: Given Landsat photos of the Earth’s surface, detect geological faults and lineaments ● Methods: image processing and computer vision algorithms 2

Example ● We want our program to detect the Garlock fault line in southern california from this Landsat image data. fault line manually painted 3

Process ● Load bands 7, 6, 1 into a false color image ○ Bands 7, 6 are recommended most for geological analysis true color false color (bands 1, 6, 7) grayscale conversion 4

Process ● Use Principal Component Analysis (a technique using matrix singular value decomposition) to reduce the dimensionality of image features. 5

Process ● Laplacian and sobel filtering to make edge detection easier Laplacian Sobel (X) Sobel (Y) Combined 6

Process ● Use the Canny edge detection algorithm to threshold edges in the filtered image. Canny edges 7

Process ● Apply the Hough transform and adjust hyperparameters until we get the most accurate result Hough lines 8

Process ● Overlay the lines on the original image Manual estimation of fault line Computer vision detected lineaments 9

Animated Process 10

References [1] Landsat-8 images courtesy of the U. S. Geological Survey [2] Anwar Abdullah, Shawki Nassr, and Abdoh Ghaleeb, “Remote Sensing and Geographic Information System for Fault Segments Mapping a Study from Taiz Area, Yemen, ” Journal of Geological Research, vol. 2013, Article ID 201757, 16 pages, 2013. https: //doi. org/10. 1155/2013/201757. [3] Farahbakhsh, Ehsan, Rohitash Chandra, Hugo KH Olierook, Richard Scalzo, Chris Clark, Steven M. Reddy, and R. Dietmar Muller. ” Computer vision-based framework for extracting geological lineaments fromoptical remote sensing data. ” ar. Xiv preprint ar. Xiv: 1810. 02320. 11