Reduced dimension Vector Quantization encoding method for image compression

Abstract

The codebook and image block compression by Compressive Sampling (CS) in Vector Quantization (VQ) is proposed for image coding. Both the memory storage and the computational complexity in the VQ Encoder could be reduced for resources constrained applications. The deteriorated image produced by only using the first m transformed coefficients for codebook search could be restored and enhanced with a convex optimization program called l1-norm minimization in the decoder. The computational intensive process is shifted from the encoder to the decoder. This feature allows it to be suitable for wireless sensor network applications.