Control of Distorted Image Points Based on the Mechanism of Identification of a Micro-object with a Cosine Transform

Methodical bases for the identification of images of micro-objects, for example, pollen grains, unicellular microorganisms of medical objects, fingerprints, and others have been developed. The traditional mechanisms of Gaussian, median filtering, fast Fourier transform, wavelet, and shift transformation are investigated. A new approach to solving the identification problem is proposed based on the principles of using information structural components of micro-objects – fractal, morphological, histological, statistical, dynamic, and other characteristics of image points. The traditional identification technology is being modified based on the development of mechanisms based on algorithms for filtering, transformation, segmentation, tracking, detection, and correction of distorted points. The modified algorithms are focused on the recognition and classification of micro-objects and problem solutions taking into account the conditions of a priori insufficiency, parametric uncertainty, and low reliability of information processing. A comparative analysis of the effectiveness of the algorithms was carried out for variants of two–dimensional, and three-dimensional cosine transformations, determination of point correlations, segmentation, construction of an uneven grid, and use of a pyramidal-recursive model. The efficiency of the algorithms was also studied in terms of the image processing labor intensity factor. A software package for visualization, recognition, and classification of images of micro-objects based on the developed functional modules has been implemented and tested using the existing digital technology of computer vision of unicellular microorganisms of medical facilities.