Transcript NaughtonPresentation..

Enhancing the Enlargement of Images
Tara Naughton
Computer Systems Lab 2009-2010
Introduction

Typically two methods of image enlargement

Pixel replication


Interpolation


Replaces original pixel with group of pixels of the same
value
Replaces original pixel with group of expanded pixels
Produce results too jagged or too blurred
Fig 1. Pixel replication vs. interpolation
Similar Research

Focus on keeping defined edges

Ramp edge model


Fig 2. Step edge parameters
Locate edge pixels, derive parameters, derive
intensity, keep parameters the same in the
enlarged images
Step edge model

Divides image into “atomic areas,” derive step
edge parameters for each pixel, find intensity of
atomic areas, find step-likeness value to see if
near step edge, use step-likeness value as
weight to combine intensity values derived for
atomic area
Fig 3. Ramp edge parameters
Background

Coding in Python

Takes input of black-and-white .pgm images

Outputs enlarged black-and-white .pgm images
Discussion



Separates images into parts: edges and nonedges
Resize non-edges with interpolation method
Find intensity values for edges and resize
accordingly
Results



Produce images that are neither too blurry nor
too jagged
Compare between original, resized image, and
images resized by other methods
Currently produces image resized by pixel
replication
Fig 5. Original image
Fig 6. Image resized by pixel replication
(scaled to smaller size, but with
changes to image visible)