Displacement measurements using phase correlation and ENO interpolation image processing techniques

Abstract

The measurement of translational shifts has a vital role in image processing and has an impact on applications such as medical imaging, remote sensing and structural health monitoring. However, due to the physical limitations of photogrammetric equipment, a sufficient accuracy may not be achievable. Therefore this thesis presents two new displacement measurement techniques to improve accuracy when compared to previous methods. While existing sub-pixel edge detection methods may be used to overcome the limitation of one pixel spacing in photogrammetric equipment, they may produce unreliable results when presented with weak edge points. Therefore, this thesis presents a new sub-pixel edge detection algorithm namely, the Essentially Non- Oscillatory Laplace Difference of Gaussian (ENO-LDoG) method. This method incorporates a high-order ENO interpolation scheme and a Laplace of a difference of Gaussian to accurately determine edge points. Phase correlation is also a well-known method with the ability to measure displacement while remaining resilient to noise, illumination, occlusion and various 2D transformations. While this method has been extended to allow for sub-pixel shifts, the phase correlation matrix is often degraded by noise and aliasing. Therefore, this thesis presents a new robust phase correlation method namely, the Adaptive Masking Phase Correlation (AMPC) technique. In this method, an adaptive masking and an ENO level-set smoothing is performed to improve resilience to noise and aliasing in the phase correlation matrix. The proposed methods are tested using various images to show that a higher precision is achievable when compared to existing methods.

Date of Award	2013
Original language	English