Abstract:
This project looks at the challenges of creating the automatic system for
reconstruction shredded documents cut vertically (strip) and vertically, horizontally (cross-
cut). The un-shredding problem is of interest in the fields of forensics, investigative sciences,
archaeology and criminal investigation. All stages of the un-shredding pipeline are analysed,
starting from uploaded images of shreds and ending with reconstructed documents. The
current bottlenecks in this pipeline are identified and solutions are proposed.
At the beginning a formal definition of this problem and references to related work
will be given. While there are some approaches published dealing with the reconstruction of
destroyed paper in general, there is barely work done in the field of cross-cut shredding then
giving an introduction to algorithm of ant-colony. The solutions proposed in this project start
with narrowing problem then implement two reconstruction approaches to improve the
results. First approach, Character database, Feature matching algorithm, Row Clustering,
Estimate error evaluation function and Greedy classifier then get the results. To improve the
results that belongs to first approach we implement second approach. Second approach is
reconstructing by Ant colony algorithm by estimate different error evaluation function then
apply the algorithm.