Sparse Representation through Multi Sparse Representation through Multi Sparse Representation through Multi Sparse Representation through Multi--Resolution esolution esolution Transform for Image Coding Transform for Image Coding Transform for Image Coding Transform for Image Coding

摘要: Having a compact basis is useful both for compression and designing efficient numerical algorithms. In this paper, new image coding scheme using multi-resolution transform known as Bandelet Transform that provides an optimally images by exploring their directional characteristics proposed. As process results in sparse representation, Zero Vector Pruning applied in-order to extract the non-zero coefficients. Further geometric interpixel redundancies present transformed coefficients are removed. The psycho-visual removed simple Quantization (VQ) process. Finally, Huffman encoder used encode significant proposed method beats standard wavelet based algorithms terms of mean-square-error (MSE) visual quality, especially low-rate regime. A gain bit-rate about 0.81 bpp over achieved yielding similar quality factor.