Skip to Main Content
Vector quantization is a technique used extensively and successfully to compress digitized data such as speech and images. The design of a vector quantizer is very computationally intensive. Parallel algorithms based on various architectures have been proposed for related applications in clustering. These algorithms are in general not practical for the vector quantization problem because of the magnitude of the parameters involved. We present a parallel SIMD algorithm which can run efficiently on parallel machines of variable sizes. The speedup and efficiency of the algorithms are high across a wide range of input parameters.