An iterative image registration technique with an application to stereo vision-英文文献.pdf

From Proceedings of Imaging Understanding Workshop , pp. 121-130 (1981). An Iterative Image Registration Technique with an Application to Stereo Vision Bruce D. Lucas Takeo Kanade Computer Science Department Carnegie-Mellon University Pittsburgh, Pennsylvania 15213 Abstract 2. The registration problem Image registration finds a variety of applications in The translational image registration problem can be computer vision. Unfortunately, traditional image registration characterized as follows: We are given functions F(x) and techniques tend to be costly. We present a new image G(x) which give the respective pixel values at each location registration technique that makes use of the spatial intensity x in two images, where x is a vector. We wish to find the gradient of the images to find a good match using a type of disparity vector h which minimizes some measure of t he Newton-Raphson iteration. Our technique is faster because difference between F(x + h) and G(x), for x in some region of it examines far fewer potential matches between the images interest R. (See figure 1). than existing techniques. Furthermore, this registration technique can be generalized to handle rotation, scaling and shearing. We show