Abstract

Content-based image retrieval (CBIR) has become one of the most vital components of modern multimedia systems especially as the digital world continues to generate and store massive volumes of visual data. Among various CBIR approaches the HG (Histogram-Gradient) method has gained significant attention due to its ability to capture both colour distribution and directional texture information from images. Landscape imagery in particular poses unique challenges because of its highly diverse patterns including mountains water bodies forests skies and man-made structures that coexist in complex overlapping forms. The traditional HG method while effective for general image retrieval still struggles with achieving high accuracy in scenarios involving subtle colour variations complex textures variable illumination large intra-class variations and noise-related distortions that commonly occur in real-world landscape photos. This research paper aims to improve the HG method by integrating advan