National Aerospace University «Kharkiv Aviation Institute»

Intelligent Methods for Multichannel Image Processing Based on Visual Attention Maps and Image Databases with Distortions, Project №0124U001094

During the execution of the research project, the following scientific and technical tasks will be accomplished:

-  improvement of image compression performance, i.e., either enhanced image quality at a fixed compression ratio or an increased compression ratio at a fixed image quality level;

-  development of an automated procedure for selecting the optimal threshold (ensuring the highest noise suppression efficiency) for filtering by employing more adequate no-reference image quality metrics that account for human visual attention, as well as by training neural networks using newly created, dedicated image databases;

-  formulation of recommendations for the use of pre-filtering and compression to achieve improved performance in object classification and recognition in multichannel images.