Vol. 5 No. 2 (2021)
Articles

Efficient Automatic Segmentation of Multi-Domain Imagery Using Ensemble Feature-Segmenter Pairs with Machine Learning

Published 2021-10-01

Abstract

Automatic image segmentation is an area of research in which pictures are analysed for the best feasible segmentation configuration utilising sophisticated colour, texture, and shape-based characteristics. In terms of computational constants, picture size, enhancement factors, edge thresholds, and so on, these setups vary. Automatic segmentation methods utilise bio-inspired approaches like Genetic Algorithm (GA), particle swarm optimization (PSO), and others to find these constants. When the input picture type changes, these algorithms must be re-trained and re-evaluated. Medical resonance imaging (MRI) and natural pictures, for example, need distinct sets of edge thresholds. As a result, there is no one method that can address the issue of multi-domain automated picture segmentation. To address this flaw, this paper presents a new ensemble-learning-based method for successful segmentation that employs feature-segmenter pairs. When compared to current state-of-the-art algorithms, the suggested method is shown to have a superior peak signal-to-noise ratio (PSNR) and modest latency. While maintaining an optimal probabilistic random index (PRI) and delay performance, the PSNR is enhanced by 10%.