Abstract
A method for estimating the translation offset between a pair of images to subpixel accuracy is described and analyzed. This method assumes the extraction of a digital edge representing a straight line in the image and then computes the best real line which could have given rise to the digital line. The key difference between this method and other approaches to subpixel fitting is that our method uses subpixel line location to estimate a digital model such as the digital line and then fits the digital model to the image while other approaches directly fit a continuous model to the image or an image related object such as a cross correlation matrix. The present paper lays a theoretical foundation for this digital approach. This technique can be applied in a variety of remote sensing and industrial vision problems requiring subpixel registration accuracy.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
