Boosting Octree Color Quantization Descriptor for Content Based Image Retrieval: A Case Study

Color quantization (CQ) plays a crucial role in image processing with reducing the number of colors in an image or of bins in a color histogram. This operation is particularly important in Content-Based Image Retrieval (CBIR) when using color histogram-based descriptors. Among various image processing techniques, the Octree tree data structure stands out as a powerful tool for performing CQ. By manipulating two key parameters, namely the number of colors K and the depth of the tree D, diverse outcomes can be achieved with the Octree. In this study, multiple experiments were conducted to determine the optimal values of K and D that yield the highest precision for the Color Octree Quantization Descriptor (COQD). COQD is a descriptor that utilizes the Octree data structure to quantize images and subsequently extracts features from them based on a color string coding (CSC). The experiments undertaken in this case study demonstrate enhancements in precision.