Performance Evaluation of Nature—Inspired Optimization Techniques in Disentangling Text Pattern Overlaps
Sivaramakrishnan Rajaraman, Ganesh Vaidyanathan and Arun Chokkalingam
This paper evaluates the performance of nature-inspired, optimization techniques including Firefly Algorithm (FA), Simulated Annealing (SA), Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) in resolving occlusions in text patterns. These Optimization techniques work individually on a randomly extracted population, from the occluded text strings. FA performs at its best by resolving occlusions even when 80% of a text string is hidden under the other. The brightness of the fireflies is associated with the objective function to be optimized and the algorithm assumes idealized rules for attractiveness and brightness and updates the position of the fireflies to reach a global optimum value. The text patterns are resolved and individually identified after a certain number of iterations. The selection of FA as the best in disentangling text strings is validated with the one-way Analysis of Variance (ANOVA) statistical tool, followed by Tukey’s post-hoc test.
Keywords: Nature-inspired, optimization, occlusion, text patterns, disentangle, objective function, firefly algorithm (FA), simulated annealing (SA), genetic algorithm (GA), particle swarm optimization (PSO), analysis of variance (ANOVA), Tukey’s post-hoc test