My approach to symbol recognition of Chinese numerals uses techniques well established in digital image processing, including black/white conversion, digital binary morphology, pixel histograms for line counting, Zhang-Suen thinning for skeletonization, and other simple techniques such as median filtering. These techniques are combined to form a feature extraction algorithm. Once features are identified simple table lookups are preformed for number recognition. The table is constructed from known features in each of the non-complex and common Chinese numerals.
This project also implements basic image processing algorithms, including point and neighborhood operations. [More]
This is my implementation of an SQL engine during an undergraduate direct independent study, spring 2003. I used Flex and Bison to define the SQL grammar and parser generation. Inner joins were implemented with conditional predicates. I called my engine Object Oriented because I envisioned the engine storing and manipulating objects: images, sound, graphs, etc. The internal API was extensible to allow for other objects and manipulation algorithms. For example, store and manipulated images files with an image processing algorithm. In one semester's time I was only able to implement the base SQL engine, image viewing, and graph generation based on table data. [More]