You are on page 1of 9

CLUSTER INNOVATION CENTRE

UNIVERSITY OF DELHI

Project Report

Image Warping
Supervisor Team Members:
Assist Prof Shobha Rai Ansh Raghuwanshi(12008)
Premchand Gurjar(12041)
Contents
➢Abstract

➢Introduction

➢Methodology

➢Result / Screenshot

➢ Conclusion

➢ References
Abstract
In this report we created warps of an image using several
warping methods like perspective warping ,reflection warping ,
rotation warping,transformation warping,projection
warping.Here we discuss the raster to numerical transformation
of pixel data so that this data can be used for mapping and
resampling the data.

Introduction
There are many platform which provide us to work
with image data in order to create warps of an image
But due to familiarity with python we chose to work
with openCV , Scikit-image and due to having certain
data structures we used pandas and numpy.
Methodology
To implement warping on an image we have to make
pixels of an image then map these and resampling.
For mapping we use complex functions
x=Fx (u,v)
y=Fy(u,v)

dst(x,y)=src(u,v)

For sampling we also use complex functions

iu=trunc(u+0.5)

iv=trunc(v+0.5)
src(x,y)=dst(iu,iv)

We iterate over these functions until every pixel is


mapped .

For resampling we use inverse complex functions

u=fx^(-1)(x,y),

v=fy^(-1)(x,y)

src^(-1)(x,y)=dst^(-1)(u,v)

For remapping we use

iu=trunc^(-1)(u+0.5)

iv=trunc^(-1)(v+0.5)

After resampling the data we get warp(specifically


depending on complex functions ).
Tools and techniques
Screenshots

Perspective transformed
Conclusion
An interactive image warping system is built based
on a surface fitting paradigm and fast multigrid
method with smooth transitions and flexible local
control. The system is able to perform close to
real-time warping operations on images with sizes up
to 256x256 with smooth transition. Moreover the
technique can be extended to create animations [
and perform three dimensional volume morphing .

You might also like