Development of methods, models and information technologies on their basis for improving the quality of pictures of various physical nature

Project author. National Mining University, Department of System Analysis and Control, Dnepropetrovsk (Ukraine).

Project objective. Development of computer software that will provide an increase of resolution capability and reliability of analysis and segmentation procedures of both ordinary and multi-parameter (multi-specter) pictures under the conditions of uncertainty of their formation system and also location and type of a potential interest object.

Project background. At present the pictures are more often used for presentation of information in scientific studies, medicine, ecology, industry. Specific complexity during processing and analysis is presented by low-contrast pictures that are characterized by the presence of “fuzzy” areas, identification and analysis of which presents basic interest (analysis of data of remote sensing, decision making during medical diagnostics and so on).

Since the shape, location (often also presence) and characteristics of the object of interest (anomalies and pathologies) on an aperture of an analyzed picture is a priori considered to be unknown then it is quite hard to separate structural (anatomical) and measuring noises from the useful signal as statistical and spectral characteristics of these noises are also unknown.

At this, the initial pictures conversion is proposed, quality of which is often insufficient because of the imperfection of systems for its gaining, transferring and saving as a shape that is dependent on its type, physical essence and set aim, that is a complex task and often requires development of new approaches and methods. Computer software developed based on new approaches, increases the quantity of analyzed parameters, increasing by this the accuracy and reliability of the analysis, and at the same time, must form one single picture from multidimensional data, simplifying their analysis and interpretation.