QUALITY INDICATORS FOR REPRODUCING FINE DETAILS OF DIGITAL IMAGES WITH TRESHOLD LIMITING OF SPECTRAL COMPONENTS
MetadataПоказати повний опис матеріалу
The article considers the effect of compression of the information stream on high-definition image quality image parameters on the receiving side. In order to assess objectively the quality of the reconstructed images, two parameters are selected: the peak-to-noise ratio and the ratio of the peak signal-to-noise value at the boundaries of the image objects. The boundaries of objects are proposed to be selected by the gradient method, after that to make a binary matrix whose area will depend on the set threshold of significant signal values. With the use of spectral transformations and the rejection of small spectral components, the compression of the digital stream is carried out with losses. Several quantization matrices of frequency components are considered and it is shown that, in some cases, image textures that have low contrast are distorted or disappeared. Compression coefficients and corresponding signal-to-noise ratios for frequency-dependent quantization matrices are calculated. In this work, we have also resulted in the dependence of the decoded image quality parameters depending on the length of words that define the brightness and color difference signals.