IMAGE PLUG-INS FOR GIMP - PowerPoint PPT Presentation

About This Presentation
Title:

IMAGE PLUG-INS FOR GIMP

Description:

IMAGE PLUG-INS FOR GIMP Author: Vera Kuki Supervisors: Shaun Bangay Adele Lobb – PowerPoint PPT presentation

Number of Views:108
Avg rating:3.0/5.0
Slides: 19
Provided by: Vera196
Category:
Tags: for | gimp | image | ins | plug | discrete | maths

less

Transcript and Presenter's Notes

Title: IMAGE PLUG-INS FOR GIMP


1
IMAGE PLUG-INS FOR GIMP
  • Author Vera Kukic
  • Supervisors Shaun Bangay
  • Adele Lobb

2
GIMP
  • Overview
  • Recap
  • GIMP
  • Image Plug-ins
  • Research
  • Algorithm
  • Colour Transfer Algorithm
  • Image Inpainting Algorithm
  • Phases
  • Questions

3
GIMP
  • GIMP is the GNU Image Manipulation Package
  • Open source
  • Well documented
  • A lot of reference material available
  • Easy to use

4
Image Plug-ins
  • Colour Transfer between Images
  • Transferring colour from a source colour image to
    a destination greyscale image
  • Image Inpainting
  • Touching up old photos
  • Removing unwanted objects/text from images

5
Research
  • Research based on
  • Transferring Color to Greyscale Images
  • Image Inpainting
  • Read the papers that were referenced by the
    authors of the original papers
  • Mathematics visited the maths department on a
    number of occasions
  • Developed strategy and algorithms

6
Colour Transfer Algorithm
  • Choose the source colour image and the
    destination greyscale image
  • Convert each image to the laß
  • Select colour from source image using swatches
  • Transfer the swatch colour to the selected region
    of the greyscale image

7
Colour Transfer Algorithm
  • Convert each image to the laß
  • The laß space was developed to reduce the
    correlation between the 3 coordinate axis of the
    colour space
  • l chromatic luminance channel
  • a yellow-blue channel
  • ß red green channel

8
Colour Transfer Algorithm
  • Convert each image to the laß
  • First convert RGB image to XYZ tristimulus values
  • Then convert XYZ to LMS space
  • From LMS space convert to laß space

9
RGB to LMS
  • // convert RGB to LMS.
  • void rgbtolms (double r, double g, double b,
  • double l, double m, double
    s)
  • l 0.3811 r 0.5783 g 0.0402 b
  • m 0.1967 r 0.7244 g 0.0782 b
  • s 0.0241 r 0.1288 g 0.8444 b

10
RGB to laß
  • // convert RGB to lab (human perceptual space).
  • void rgbtolab (double r, double g, double b,
  • double l, double alpha, double
    beta)
  • double ll
  • double mm
  • double ss
  • double L
  • double M
  • double S
  • rgbtolms (r, g, b, ll, mm, ss)
  • L log10 (ll BLACKOFFSET)
  • M log10 (mm BLACKOFFSET)
  • S log10 (ss BLACKOFFSET)
  • l (L M S) / sqrt (3.0)

11
Colour Transfer Algorithm
  • Swatches
  • Allow more user interaction
  • User can selectively transfer colours between
    corresponding swatches
  • Advantage
  • Prevent wrong parts of the image from corrupting
    the swatch
  • Transfer colour even if two corresponding regions
    vary greatly in colour and luminance

12
Colour Transfer Algorithm
  • Transfer colour
  • Each pixel in the greyscale image must be matched
    to a pixel in the colour image
  • The comparison is made based on luminance value
    and neighbourhood statistics
  • Luminance value l
  • Neighbourhood stats are precomputed over the
    image
  • Standard deviation of the luminance values of the
    pixel neighbourhood

13
Colour Transfer Algorithm
  • Transfer colour
  • The aß values are transferred once the best
    matching pixel is found
  • The original luminance is retained

14
Image Inpainting Algorithm
  • Discrete Equation
  • Rate of change of I(i,n)
  • Estimate Smoothness of Function
  • Use discrete implementation of Laplacian
  • Compute Change of Value
  • Define direction of N (the normal vector)

15
Image Inpainting Algorithm
  • Diffusion Process
  • Interleaved with the image inpainting process
  • Repeated every few steps to ensure the correct
    evolution of the direction field
  • Process is repeated until steady state reached

16
GUI
17
Phases
  • Research
  • Read understand the mathematics behind the
    methods of implementation
  • Develop Algorithm
  • GUI
  • Learn GTK
  • Design Interface
  • Implement the algorithm
  • Test
  • Experiments

18
Questions?
  • ?
Write a Comment
User Comments (0)
About PowerShow.com