Distance Estimation of the Thames Water Megapig Using Image Sequences

📄 Project Abstract

The project is based on work currently being carried out by ThamesWater on their Megapig project. The Megapig is a condition assessment tool which Thames Water hope to use on their large trunk mains. The problem dealt with in this project involves the accurate distance and velocity estimation of the Megapig along the main using image sequences. This report outlines the production and evaluation of a system to cross correlate pairs of images in an image sequence, in order to obtain an accurate distance measurement. Following the correlation of the images, the system was extended to include a Kalman filter to estimate the velocity and enable the tracking of the Megapig. The results from this project have shown the use of the fast Fourier transform to be the best method of correlation, and that the addition of sub pixel accuracy, by applying a quadratic curve, can increase the accuracy of results a great deal. The use of a Kalman filter has also proved a successful method of estimating ve...