Authors: Lior Wolf, Liza Potikha, Nachum Dershowitz, Roni Shweka, Yaacov Choueka

Keywords:

• handwritten
• paleography
• fragments
• SIFT
• sparse coding
• dictionaries

Q1: What is the ultimate goal of the authors?

The two main goals are providing tools to bring together fragments of the same page (specifically from the Cairo Genizah) and trying to classify handwriting and dates.

Q2: How is SIFT used?

SIFT is used at various points of a letter to generate descriptors. There are 100,000 descriptors overall before inputting them into k-means. SIFT serves as the main classification technique.

Q3: How did they produce the letter dictionaries?

They produced letter dictionaries using the generated SIFT descriptors and k-means clustering to find representative visual words.

Q4: What is sparse coding and its importance?

Sparse coding is used to code documents (or any other codable thing) with a separating/descriptive code which also shows the similarity between items. An example might be the bag of visual words approach.

Q5: Are there any relevant techniques for our research?

This is relevant to the Divan matching problem. This work might be cited in historical document matching, although it covers techniques that are generally well-known in the field.