Upload
prashant-agrawal
View
176
Download
1
Embed Size (px)
Citation preview
Image reconstruction its individual pieces
By : Prashant Agrawal Hitesh Meghani
Input Specifications:The images pieces may be
With Straight Edges(Rectangular)
Non-uniformly Cut
Rotated by Multiples of 90 degrees
Basic Steps :1)Extraction of the individual images2)Selection of the base image
3)Formation of the boundary array
Basic Steps :
4)Search for a matching piece
5)Addition of matched piece
6)Repeat Iteratively
Step 1: Image Extraction
CROP the image to fit into a rectangular bounding box. Blacken the remaining part.
Step 2: Selection of Base
Select the base image – to add other images on it.
Step 3:Formation of the boundary arrayBoundary Array consists of
Distance of the boundary pixel from the side of the bounding box.
The red, green and blue color coordinates
Step 3:Formation of the boundary array
Step 3:Formation of the boundary array
Step 4 :Search for Matching piece
Extraction of Boundaries of two pieces
Sum is not constant
Sum =356+-1Now Check for RGB values
MATCHING POINTS
Step 4 :Search for Matching piece
Step 4:Search for Matching piece Right boundary of base image is
comparedwith: 1)Left boundary of first piece
2)Next boundary of the same piece.3)Next piece is taken.
Next boundary of the base piece is repeated until match is found.
Step 5: Addition of Matched Piece
Piece is rotated if required.The size of the base image is
increased to accommodate both the images.
The image is added on the basis of match points.
Step 5: Addition of Matched Piece
Step 6: Iterative RepetitionThe added
image is made the base image
Identify the image to
add to base
Add the image to
base
Demo :On C ( Open CV)On MATLAB
Implementation And ComparisonsOn MATLAB- User friendly
programming but takes larger time.On C- Very quick execution
ShortcomingsImages with uniform backgroundImage piece cut entirely from
another.
ApplicationsForensic sciencesPattern matching
Thank You
Questions ????