Environment-Anomaly-Detection-Using-Satellite-Images

Environment Anomaly Detection using standard machine learning techniques.

Introduction

The technique uses PCA and KMeans to detect anomaly. The code is based on the paper: Click Here

Blog link: Click Here

---

detect_anomaly.Anomaly (first_image_path = None, second_image_path = None, block_size = 5, return_diff = False)

Parameters

first_image_path: String

Image path of the first image of the particular location

second_image_path: String

Image path of the second image of the same location

block_size: Integer

Size of the window that will be used to create eigen space and vector space

• Default: 5 if not set
• Choose it to be an odd number

return_diff: Bool

If you want function to return difference image as well

Returns

change_map: matrix

Image matrix depicting the change.

• Black region showing no change
• White region showing change

diff_image: matrix

Difference image is the absolute difference between two given image.

Use

• import detect_anomaly
• create object using detect_anomaly.Anomaly() and pass required parameters.
• call method of the obect: detect() and store the results.