Method and device for estimating image quality of compressed images and/or video sequences

A device for estimating the quality of images compressed with a transformation based compression method, which uses errors generated by quantization to obtain an image quality indicator, and wherein a reference image is not used, with the transformation based compression method including:

• transforming image data available in a two-dimensional discreet spatial domain into a two-dimensional discreet frequency domain by a transformation function resulting in at least one transformation block comprising at least one transformation coefficient, each transformation block of the same size includes transformation coefficients representing same two-dimensional frequencies;
• scaling of the resulting transformation coefficients;
• and
• quantizing the scaled transformation coefficients using quantization parameters to set a number and/or size of quantization steps,
• wherein said device comprises:
• transformation coefficient counting means that calculates a probability of at least one specific quantized transformation coefficient value for at least all coefficients representing one identical two-dimensional frequency;
• probability density function parameter estimation means that estimates identifying parameters of at least one probability density function of non-quantized coefficient values from the quantized coefficient value probabilities, with a type of probability density function being preset;
• and
• image quality indicator calculation means that calculates an image quality indicator from errors introduced by quantization which are derived from probability density function parameters and quantization parameters.