AN EFFICIENT HYBRID FEATURE FOR EDGE-PRESERVING BASED ON BLOCK TRUNCATION CODING AND TREE-STRUCTURED VECTOR QUANTIZATION WITH EDGE ORIENTATION CLASSIFICATION OF BIT-PLANES

摘要

In this paper, we propose a hybrid feature which merges Block Truncation Coding (BTC) and Tree-Structured Vector Quantization (TSVQ) with classification of edge orientation bit-planes. BTC is known for its high quality of reconstructed images, much lower computational complexity and low compression ratios. Vector Quantization (VQ), on the other hand, results in very high compression ratios but with high complexity. To solve these problems, we propose an improved approach for feature extraction based on bit-plane classification according to the Edge Orientation of the Bit-Plane (EOBP) templates and combine it with TSVQ. As different areas of the image have different activities and contrasts, we encode the active blocks containing edges with BTC and EOBP templates, and encode the inactive blocks having low intensity variations between its pixels with the VQ. Different from the original BTC and VQ based feature extraction, we extract the edge orientation information by applying the EOBP classification on the bit- plane image blocks, as a result that we classify image blocks into 28 kinds with different edges. Simulation results show that our proposed method performs better in image compression and image retrieval.