Irraivan Elamvazuthi

Multiobjective optimization using particle swarm optimization with non-Gaussian random generators

In engineering optimization, multi-objective (MO) problems are frequently encountered. In this work, a real-world MO problem (resin-bonded sand mould system) is tackled using Particle Swarm Optimization (PSO) in conjunction with the weighted-sum approach. Random generators (stochastic engines) provides sufficient randomness for the algorithm during the search process. The effects of non-Gaussian stochastic engines on the performance of the PSO technique in a MO setting is explored in this work. The stochastic engines operate using the Weibull distribution, Gamma distribution, Gaussian distribution and a chaotic mechanism. The two non-Gaussian distributions are the Weibull and Gamma distributions. The Pareto frontiers obtained were benchmarked using two metrics; the hypervolume indicator (HVI) and the proposed Average Explorative Rate (AER) metric. Detail comparative analysis on the effects of non-Gaussian random generators on the PSO technique is provided.

Musculoskeletal ultrasound image denoising using Daubechies wavelets

Among various existing medical imaging modalities Ultrasound is providing promising future because of its ease availability and use of non-ionizing radiations. In this paper we have attempted to denoise ultrasound image using daubechies wavelet and analyze the results with peak signal to noise ratio and coefficient of correlation as performance measurement index. The different daubechies from 1 to 6 is used on four different ultrasound bone fracture images with three different levels from 1 to 3. The images for visual inspection and PSNR, Coefficient of correlation values are graphically shown for quantitaive analysis of resultant images.