TY  - BOOK
AU  - Fernandez, joselito
TI  - Automatic image stitching
U1  - T F3912 2012 c.1 
CY  - 3
N1  - 000-099
N2  - The problem considered in this paper is an automatic construction of panorama images which mainly applies to a pair of images captured in different directions. Fundamentally, this problem requires machine vision; as we need which of the panorama join up.

The system is implemented by using Microsoft Visual C# 2008 and it uses a method based on Accord.NET built-in features to realize fully automatic image stitching, in which it includes two main parts: image matching and image blending. As the noises images have large differences between the other images, when using cross-correlation and RANSAC (Random Sample Consensus) algorithm to realize correct and robust matching, it supplies a probabilistic model to verify the panorama image sequence. In addition, to have a more satisfied panorama image, it uses a simple and fast blending method which is linear gradient alpha blending method.

When two images that are to be stitched are taken from different angles, the line where the two images overlapped will be more prominent. To address this, we have to work on an image blending code that we can incorporate to the existing cross-correlation and homography estimation code. This is to ensure a more seamless image stitching program the next time around. Also, we can improve still the speed of image processing. One way of doing so may be combining different algorithms to a single code. This might help in speeding up processing time. Finally, the experiment results confirm the feasibility of our methods
ER  -