Combining 3D Robust Smoothing Filtering and Adaptive Clustering Analysis for Geophysical Signal de-nosing and Image Segmentation

Image segmentation is widely used to enhance images in many fields. However, the existence of noise interferences will distort the segmentation results. In this paper, we developed ahigh-dimensional robust smoothing filter algorithm based on M-estimation. The algorithm can remove both Gaussian random noise and peak pulse interference inone-dimensional (1D), two-dimensional (2D) and three-dimensional (3D) images. We combinedthe robust smooth filtering algorithm withan adaptive clustering analysis algorithm to segment geophysical image and extract the large-scale boundaries of anomalous bodies. By removing noise interferences and marking outliers of the 3D images in advance, the adaptive clustering results are more stable and the resolution is improved. The results of geophysical image segmentation under different noise interference demonstrate the effectiveness of the combined algorithm.