CS297 Proposal

Android de-Shredder App

Vasudha Venkatesh (vasudha.venkatesh@sjsu.edu)

Advisor: Dr. Chris Pollett

Description:

Sensitive documents are usually shredded into strips before discarding them. Shredders are used to cut the pages of a document into thin strips of uniform thickness. Each shredded piece in the collection bin could belong to any of the pages in a document. The task of document reconstruction involves two steps: Identifying the page to which each shred belongs and rearranging the shreds within the page to their original position. The difficulty of the reconstruction process depends on the thickness of the shred and type of cut (horizontal or vertical). Thickness of the shred is directly proportional to the ease of reconstruction. Horizontal cuts are easier to reconstruct because sentences in a page are intact and not broken. Vertical cuts are harder because there is very little information to glean from each shred.

In this project, an Android app will be developed to reconstruct the pages of a shredded document by using photograph of the shreds as input. There will be no prior knowledge of the page to which each shred belongs and the thickness of each shred will conform to the measurements of a standard strip shredder. The type of shredder cut will be vertical. This project is intended to enhance existing work which currently reconstructs upto ten images from a mixed bag of square pieces.

Schedule:

Week 1: Aug. 29 - Sep. 4	First project meeting to discuss logistics of future meetings. Work on an android app to take a picture from camera and pass the image to an activity. Goal of this app should be to learn inter-activity communication in android.
Week 2: Sep. 5 - Sep. 11	Deliverable #1: Working demo of android app for inter-activity communication. Clicking on a button splits the image into chunks.
Week 3: Sep. 12 - Sep. 18	Read and understand the paper on "Solving Multiple Square Jigsaw Puzzles with Missing Pieces" by Paikin & Tal
Week 4: Sep. 19 - Sep. 25	Read and understand existing work for solving multiple square jigsaw puzzles by Zayd Hammoudeh. Begin work on making it to work on different set of images.
Week 5: Sep. 26 - Oct. 2	Deliverable #2: Demo existing code on custom images.
Week 6: Oct. 3 - Oct. 9	Review existing literature on shredded document reconstruction
Week 7: Oct. 10 - Oct. 16	Begin work on converting the Python-based jigsaw solver to Java
Week 8: Oct. 17 - Oct. 23	Begin work on converting the Python-based jigsaw solver to Java-
Week 9: Oct. 24 - Oct. 30	Deliverable #3:Implement Paikin Tal Placer
Week 10: Oct. 31 - Nov. 6	Review existing literature on piece similarity calculation
Week 11: Nov. 7 - Nov. 13	Review existing literature on piece similarity calculation
Week 12: Nov. 14 - Nov. 20	Begin work on hierarchical clustering and segmentation
Week 13: Nov. 21 - Nov. 27	Deliverable #4:Comparison results between image and documents for solver in java using existing metrics
Week 14: Nov. 28 - Dec. 4	Explore LSTM RNN for best matching shred calculation
Week 15: Dec. 5 - Dec. 11	Deliverable #5:Complete CS297 final report

Deliverables:

The full project will be done when CS298 is completed. The following will be done by the end of CS297:

1. Working demo of android app for inter-activity communication. Clicking on a button splits the image into chunks.

2. Demo existing code on custom images.

3. Implement Paikin Tal Placer

4. Comparison results between image and documents for solver in java using existing metrics

5. Complete CS297 Final Report

References:

Z. Hammoudeh, "A Fully Automated Solver For Multiple Square Jigsaw Puzzles Using Hierarchical Clustering," Thesis, 2016.

G. Paikin, A. Tal, "Solving Multiple Square Jigsaw Puzzles with Missing Pieces", IEEE Conference on Computer Vision and Pattern Revognition, 2015.