Experimental study on variants of Gaussian mixture model for segmentation
S. K. Katti, Shrinivas D. Desai, Vishwanath P. Baligar, Gururaj N Bhadri · 2024
Gaussian Mixture Models (GMMs) are utilized extensively in many different fields because of their adaptability and capacity to represent complex data distributions. This paper explores the use of Gaussian Mixture Models in image segmentation, offering a strong solution to the difficult task of segmenting different image collections. GMMs are useful for capturing intricate data distributions, and this study thoroughly examines the characteristics and elements of GMMs. The structural details of Gaussian mixture models (GMMs) are explored in detail, emphasizing the flexibility and effectiveness of these models in simulating different application images. It demonstrates how crucial image segmentation is to gaining an extensive understanding of an image&s;s visual characteristics. This study investigates various segmentation strategies that consider the diverse features present in image collections, employing the Expectation-Maximization (EM) and Minorization-Maximization (MM) algorithms. This research goes above and beyond conventional methods to provide a comprehensive understanding of segmentation procedures. The segmentation methodology&s;s methods are thoroughly clarified in the paper, as well as goes into detail about the strategy.