Abstract :
We propose a method for devising approximate multiplication-free algorithms for compressed-domain linear operations on images, e.g., downsampling, translation, filtering, etc. We demonstrate that the approximate algorithms give output images that are perceptually nearly equivalent to those of the exact processing, while the computational complexity is significantly reduced
Keywords :
computational complexity; convolution; data compression; discrete cosine transforms; filtering theory; image coding; image sampling; quantisation (signal); transform coding; JPEG; MPEG; compressed-domain linear operations; computational complexity; convolution; discrete cosine transform; downsampling; exact processing; filtering; image compression; multiplication-free approximate algorithms; output images; quantization; scaling; translation; Computational complexity; Convolution; Discrete cosine transforms; Filtering algorithms; Image coding; Maximum likelihood detection; Nonlinear filters; Quantization; Transform coding; Vectors;
