Abstract: IEEE Content-based cartoon image retrieval systems face challenges due to intra-class variability, shape invariance, and scalability. This paper proposes a novel framework combining Scale-Invariant Feature Transform (SIFT) and Histogram of Oriented Gradients (HOG) for feature extraction, enhanced by Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) for dimensionality reduction. A Kd-tree indexing mechanism ensures efficient retrieval. Experiments on a custom dataset of 600 images (30 classes) demonstrate that fused SIFT+HOG achieves 84% higher precision than standalone methods. The system addresses pose variation, background clutter, and scalability, making it suitable for animation studios, advertising, and educational tools.

Keywords: Cartoon retrieval, feature fusion, SIFT, HOG, dimensionality reduction, Kd-tree indexing

| DOI: 10.17148/IARJSET.2025.125233