Color Image Retrieval Schemes Using Index Histograms Based on Various Spatial-domain Vector Quantizers

This paper proposes two new compressed-domain features for color image retrieval based on the YCbCr color space. They are named Multi-Stage Vector Quantization Index Histograms (MSVQIH) and Mean-Removed Vector Quantization Index Histograms (MRVQIH). For each color component, to obtain the MSVQIH features, we extract two MSVQ Index histograms from the two stage VQ index sequences respectively. Similarly, to obtain the MRVQIH features, we extract two MRVQ index histograms from the mean index and residual VQ index sequences respectively. The retrieval simulation results show that, compared with the traditional Spatial-domain Color-Histogram-based (SCH) features and Vector Quantization Index Histograms-based (VQIH) features, the proposed two kinds of features can largely improve the recall and precision performance.