Freeman Chain Code Route Length Optimization Using Meta-heuristic Techniques for Handwritten Character Recognition

Chain code is used as representation for an image in form of sequence of directional codes along the border or structure line. Issue arises during its extraction when the line has branches and the sequence must be continuous; no restarting at any junction is allowed. This paper presents a chain codes extraction of Thinned Binary Image (TBI) from upper-case Centre of Excellent for Document Analysis and Recognition (CEDAR) dataset using Meta-heuristic techniques. There are six methods in Meta-heuristic techniques that called Differential Evolution (DE), Particle Swarm Optimization (PSO), Genetic Algorithm (GA) and Ant-Colony Optimization (ACO), Harmony Search Algorithm (HSA) and Simulated Annealing (SA). In the feature extraction, Freeman Chain Code (FCC) was used as data representation that uses 8neighbourhood directions. However, the FCC representation is dependent on the route length and branches of the characters’ node. These six methods are used to find the shortest route that consumed minimum computational time. The experimental result shows that the route length and computation time using DE, PSO, GA, ACO, HSA and SA. Comparing to five other techniques, the results revealed that SA has the shortest chain code length and lowest computational time with 1,856.13 and 0.07 second, respectively.