Image registration

Abstract

In the current Third Industrial Revolution, Image Registration is often used in the area such as X-ray Images, Satellites Images (Earth, Planets and Monitoring of Space) and Real Time Motion Tracking to enhance life by Research and Development (R&D). Basically, Image Registration is a process whereby multiple images which overlapping areas are being “combined” or “stitched” into one. Therefore, the quality of the result produced depends on the feature detection. Although there are many detection techniques available, each has its advantages and disadvantages for different type of features. For this project, a detection technique called the Speeded Up Robust Features (SURF) will be used. It is a modified technique much faster than Scale Invariant Features Transform (SIFT). Furthermore, M-Estimator will be used for detection of inliers and outliers. Similarly, the M-estimator is a modified technique of the Random Sample Consensus (RANSAC).