Research Methodology Applied in Image Processing: A Review

Abstract Image processing has become an indispensable tool across various fields, including medical diagnostics, remote sensing, industrial inspection, and surveillance. The continuous advancement in technology and the growing demand for accurate visual interpretation have led researchers to explore a wide range of methodologies over the years. This review paper focuses on the research methodologies applied in image processing, with particular emphasis on literature published between 2010 and 2016. It presents a comparative analysis of classical image processing techniques—such as thresholding, region-based segmentation, and morphological operations—and modern machine learning approaches, including support vector machines (SVMs), convolutional neural networks (CNNs), and hybrid models. The paper highlights the strengths, limitations, and application domains of these methods and evaluates their performance based on key metrics like accuracy, speed, and robustness. Additionally, it identifies significant research gaps in areas such as real-time processing, multi-modal image fusion, and handling of noisy or low-quality images. A structured gap analysis suggests potential areas for further investigation, including lightweight deep learning models, transfer learning for small datasets, and improved segmentation techniques for complex visual data. This review serves as a valuable reference for researchers, practitioners, and academicians aiming to understand the evolution of image processing methodologies and explore new directions for future research. Keywords - Image processing, Segmentation, Deep learning, Feature extraction, Transfer learning, Pattern recognition