TESTING SOFTWARE FOR INCREASING THE VISUAL PROPERTIES OF DIGITAL IMAGES

Аннотация

This article discusses the algorithm and the results of the software implementation, which was developed during the creation and research of algorithms to improve the visual properties of digital images. Images from the TID2013 database were used as a test image to test the results of the software. A database of digital images MyImages with different types of distortion from real scenes was created. Evaluation measures were used to validate the quality evaluations of digital images. Pearson's correlation is used for comparison with a reference, and the CLRIQA algorithm is used to evaluate images without a reference. Visual evaluation of the quality of digital images was carried out using social networks. All processed images were estimated on a 5-point scale: 1-satisfied, 2-bad, 3-satisfied, 4-good, 5-very good. All results were validated by indicators. Pearson's correlation coefficient, Spearman's rank correlation and Kendell's correlation coefficient were used to determine the compliance between objective and subjective evaluations. The final testing of the proposed algorithms based on MyImage and TID2013 shows the best result in the range of [0.71-0.83]. This testifies that the objective indicators of image evaluation agrees well with the visual opinions. The developed algorithms for improving the visual quality of digital images are integrated into the AutoImage program. This program received a certificate of state registration of a computer program (rights No. 30080 dated November 8, 2022).