LIDAR – CloudCompare Change Detection

Table of Contents

1. Area of Interest

The area of interest is south of Christchurch, New Zealand in the seaside community of Diamond Harbor. The area is roughly 80 ha big, spanning a forested residential area and the ocean. The altitude ranges from 0m (sea level) to ca. 120m. Ideally, some changes like new garages or renovations will be visible. Additionally, it might be interesting to see different sea levels due to the tide changing.

Map 1: DSM of study area.

2. Importing Data

The data is then imported into CloudCompare and visually analyzed. Some false points can be seen in both cloud points, but especially the 2015 data has a lot of nonsenical data points that are removed.

Image 1: Point Cloud from 2015. Clearly visible is quite a bit of noise.
Image 2: 2021 Point Cloud.

3. Z-Coordinates as Scalar Field

Next up, the height information of each point is treated as a variable of the point itself. From the 3D description, the z coordinate is taken and treated as a scalar variable, enabling the visualization of points by their height.
Image 3: 2015 point cloud visualized by height.
Image 4: 2021 point cloud visualized by height.

4. Calculating Changes in Height

Next up, the difference in height between the two point clouds is calculated (“Cloud Distance”). For each point, the newer point is substracted from the older point, returning the difference in elevation.

Image 5: Cloud Distance.

In this study area, there were no points that are negative, meaning that the altitude was lowered. It does not seem like buildings were demolished, additionally the vegetation seems to be in “higher” bloom at the second acquisition date. Some minor alignment errors (upwards) might explain why there is no deviation downwards, but still the result seems to make sense.

Leave a Reply

Your email address will not be published. Required fields are marked *

Further Reading
Recent Updates
Machine Learning
Regression – Gradient Descent Algorithm

The following notebook performs simple and multivariate linear regression for an air pollution dataset, comparing the results of a maximum-likelihood regression with a manual gradient descent implementation.

Read More »