Multimedia Image Retrieval Using Novel Modelling of Combined Binary Patterns with Reduced Dimensions

Advancements in computer vision have transformed image retrieval. However, many existing systems struggle to bridge the semantic gap between low-level features and high-level user expectations. This study proposed a new approach that combines texture and colour features to enhance retrieval accuracy and user satisfaction. The proposed method employs the HSV colour model to extract colour features, focusing on dominant features from Hue (H) and Saturation (S) components at different levels of histogram bins, while texture features are derived from the Value (V) component using the Relational Edge Patterns (REPs). These patterns, generated by analyzing binary relationships in all directions between centre pixels and their neighbours for every 5'5 matrix, effectively capture the texture properties of objects in images. After concatenating colour and texture features, Principal Component Analysis (PCA) is applied to reduce dimensionality and retain the most discriminative features. The effectiveness of the proposed method is evaluated on the Corel-10k and 102-flowers datasets, demonstrating superior precision and recall compared to existing methods. Experimental results highlight its ability to form semantically meaningful clusters, bridging the semantic gap and enabling efficient image retrieval.