Vector quantization

Abstract

Data compression is the packing of data, the process of transforming a body data into a smaller representation from which the original or an approximation of the original can be computed at a later time. Most data sources contain redundancies such as non-uniform symbol distribution, pattern repetition, and positional redundancy. A data compression algorithm encodes the data to reduce these redundancies. Vector quantization (VQ) is an efficient data compression technique for speech and images. VQ maps a pattern defined by a discrete vector into a short digital sequence suitable for transmission over a digital channel or storage in a digital medium. Today, with the computing power available, the time spent or data processing has been reduced and the application of vector quantization for real-time coding is once again considered. A study of different techniques of vector quantization will be done. The details of some schemes will be presented.