This article functions as the second step in my project proposal for ScribeX. An architecture draft drawing is provided, as well as a list of MVPs and MDPs.
Architecture
This project will tentatively run in .NET framework. The frontend UI of this program will be a webpage that allows the user to upload an image file. This file, containing handwritten information, will go through a preprocessing engine that performs the following tasks: thresholding (extract ink from paper by converting image to a binary black-and-white one), noise removal, line segmentation and character/word segmentation. The resulting data would be a collection of shapes, which will then be processed by a character recognition engine. Such an engine will be pre-trained before this project goes online. The results could be optionally adjusted based on user’s input history and writing style. In the end, the text data generated will be formatted before returning to user.
Feature list
Tentative feature list at this point is the same as the one posted in a previous article.
MVP:
– Text recognition
– Level of confidence rating
– Extracting graph directly
MDP:
– Multiple files processing
– Special character recognition
– File archive service
– Manual training
– Feedback channel