Color Spatial Feature Extraction for Image Indexing - A Case Study on the Cell B. E. Processor

The color correlogram has proven its worth as an image descriptor for both comparison and retrieval. The paper presents a case study of developing efficient color correlogram algorithms on the STI Cell Broadband Engine (Cell B. E.), which is a state-of-the-art heterogeneous multicore architecture that consists of a traditional microprocessor (PPE) with eight SIMD co-processing unit (SPE) integrated on-chip. A generic working-partition technique is used in Cell B. E. processor to hide memory access latency, and we apply code SIMDization to efficiently implement pixel level parallelism. The experimental results, conducted on a Cell Blade running at 3.2 GHz, show that the Cell B. E. provides a significant performance speed-up for the color correlogram algorithm in comparison to traditional cache-based microprocessors.