--- a/README.md
+++ b/README.md
@@ -3,9 +3,7 @@
 * In guided wave-based SHM (GWSHM) of composite structures, directional dependent wave velocities, multiple and superimposed modes, mode-dispersion, mode-coupling, boundary reflections, unfiltered and noisy responses and additional frequency-dependent scattering due to damage makes dataset complicated difficult to handle directly using deep learning methods.
 * For GWSHM, we have used domain knowledge of (a) digital band-pass filter design and visualization, (b) cross-statistical feature engineering, (c) channel and frequency preferencing, (d) physics of ultrasonic guided wave propagation and Time Of Flight (TOF) based signal windowing, (e) signal augmentation with noise to preprocess the dataset before feeding into a network.  
 
-<p align="center">
-  <img src="images/pkaml.png" width="650" height="400" />
-</p>
+
 
 -------
 #### This repository contains codes accompanying the [paper](https://www.sciencedirect.com/science/article/abs/pii/S0041624X2100086X) "Combined two-level damage identification strategy using ultrasonic guided waves and physical knowledge assisted machine learning". The dataset accompying the paper is available [here](https://openguidedwaves.de/downloads/).