Online recognition of handwritten mathematical expressions with support for matrices

We present an online system for recognizing handwritten mathematical matrices in the context of an interactive computational tool called MathPaper. Automatic segmentation and recognition of multiple expressions are supported based on a spacing algorithm that leverages recognized symbol identities, sizes, and relative locations of individual symbols. Matrices with ellipses can be recognized and instantiated with non-ellipsis elements. Both well- and non-well-formed matrices can also be recognized. Matrix elements can be any general mathematical expressions including imbedded matrices. Our recognizer also addresses the poor column alignment problem of handwritten matrices, and allows for slight horizontal overlaps between elements in neighboring columns and different rows.

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Chuanjun Li, Robert C. Zeleznik, Timothy Miller, Joseph J. LaViola Jr.
• Keywords: Computer Vision, Handwritten Mathematical Matrices, Handwritten Matrices, ICPR 2008, Imbedded Matrices