Improved SMVQ Scheme with Edge-Directed Prediction

Abstract

The side-match vector quantization (SMVQ) technique has been widely used in various image compression and hiding techniques. It effectively decreases the bit rates of image with a great compressed quality. However, the prediction is not precise while the variation between the encoded block and its neighboring upper and left blocks is large. Due to the imprecision will cause derailment problem, to solve this problem, this paper proposed an EDPSMVQ scheme which is based on edge-directed prediction (EDP) to precisely predict the side-match pixels avoiding the derailment problem. The experimental results showed that the scheme achieves a high compression ratio with an acceptable image quality.