manavkataria / tangible-images

Images you can Touch and Feel
http://tangible-images.googlecode.com
0 stars 0 forks source link

Exploring: DSP in Image Processing #8

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
What skills/features do you plan to explore/acquire?
See - Application of Fourier Transform in Image Processing and see if you 
find a filter/effect that can be used for tangible images:
http://www.google.co.in/search?sourceid=chrome&ie=UTF-
8&q=Application+of+Fourier+Transform+in+image+processing

If you find 2 good effects, we can declare our project as a "DSP Project" and 
confirm its feasibility.

Original issue reported on code.google.com by manavkataria on 10 Sep 2009 at 1:20

GoogleCodeExporter commented 9 years ago
INTRODUCTION TO FOURIER TRANSFORMS FOR IMAGE PROCESSING - Well illustrated:
http://www.cs.unm.edu/~brayer/vision/fourier.html

+ Proves that Fourier has good applications on images. 
+ How does one relate this to meaningful haptic textures? After applying some 
filters?

Original comment by manavkataria on 10 Sep 2009 at 1:48

GoogleCodeExporter commented 9 years ago
Image Segmentation thru Fourier would establish DSP in IP.

Original comment by manavkataria on 10 Sep 2009 at 2:36

GoogleCodeExporter commented 9 years ago
Examples of DSP in IP for our project.
+ Edge detection masking as Spatial Convolution
+ Depth detection using parallax, correlation and motion estimation
+ Segmentation using Multi-resolution Fourier Tansform  

Jhelum and others: 
Record examples of DSP in IP if you find here.

Original comment by manavkataria on 14 Sep 2009 at 9:16

GoogleCodeExporter commented 9 years ago
== TODO: Team A ==

Jhelum:
+ Understand these algorithms. and 
+ Help the team trying to implement this to Matlab code.

Husain and KL:
+ Attempt to implement high level version of these algorithms
++ Husain - Depth detection
++ KL - Image Segmentation

+ Please note this is the most important task at hand. More important than 
Haptics 
Rendering of Textures. Critical for feasibility. 

Original comment by manavkataria on 14 Sep 2009 at 9:19

GoogleCodeExporter commented 9 years ago
Have already implemented edge detection using prewitt and sobel mask 
convolution.
Program working and showing edges in x- and y- directions. Not able to obtain
gradient coz squaring and  then taking square-root of the two matrices not 
working...
will finish it after mid sem. Also will implement it in frequency domain... 

Original comment by husainma...@gmail.com on 14 Sep 2009 at 2:23

Attachments:

GoogleCodeExporter commented 9 years ago
Thanks for the start Husain! 

I think taking gradient is necessary. The output doesn't look convincing 
otherwise - I 
can't see any edges. At this moment we can't really rely on this implementation.

Original comment by manavkataria on 15 Sep 2009 at 6:36

GoogleCodeExporter commented 9 years ago
Square Root of Matrices: 
http://www.mathworks.com/access/helpdesk/help/techdoc/index.html?
/access/helpdesk/help/techdoc/ref/sqrtm.html

Squaring should be as simple as multiplying a square matrix by itself. 

Let me know,
Manav

Original comment by manavkataria on 15 Sep 2009 at 6:41

GoogleCodeExporter commented 9 years ago
For Archiving - 
"Super Simple Introduction to DSP in Matlab" 
http://blinkdagger.com/category/control-systems/

Original comment by manavkataria on 15 Sep 2009 at 11:16

GoogleCodeExporter commented 9 years ago
Task Done.
Closing.

Original comment by manavkataria on 29 Sep 2009 at 5:20