Mid-project Report
(July 30, 2005)Summary of Work Accomplished:
Implemented edge image symmetry detection. Used both SSD and Chamfer techniques to get the similarity scores.
Adjusted clustering methods. Examined symmetry score data and researched the questions of why group classification based on the scores was sometimes failing, and what techniques of clustering or image comparison could be used.
Implemented an error-handling procedure for failed group detection -- find the nearest group based on the scores.
Wrote code for the mathematical back-end for straightening curvilinear lattices.
Cleaned up overall presentation of code and output
Fixed technical glitches
Tabulated and organized the score data for future reference or possible ideas on clustering and comparison.
Work To Do Ideas (tentative):
A better lattice detection algorithm.
Curvilinear transform displacement vector mesh finding.
Self-verification techniques and bad-image filtering.
Documentation of the m-files in the program for easier future readability.
Applications of the theory to image compression methods for patterns
.