Abstract
Many sports enthusiasts, from
novice photographers to professional publishers, rely on manual image
segmentation with tools like Photoshop to combine multiple images of a bike
trick or basketball dunk into a single image by cutting out the foreground of
each image and overlaying it onto the background of one image. The goal of this
project is to develop an algorithm that can automatically combine multiple
images generated from burst shots of an
action into a single image that clearly shows the full action. This requires
three main tasks for each set of images, including background alignment between
images in the cases when the camera is moving (using feature detection and
matching), segmentation of the foreground and background components of each
image even in cases when portions of the background might be moving, and
finally cleanly combining the foreground image segments all of the images onto a
single background image. The algorithm described in this paper successfully
compiles a variety of image sets, including those where the foreground object overlaps
between images or sets with multiple objects, but fails to compile sets where
multiple objects cross paths during the action.
Domain: Image
Processing
Language: MATLAB
Contact Detail
Ph.no: 7200555526
e-mail:
info.nanosoftwares@gmail.com
No comments:
Post a Comment