SEA SURFACE IMAGE SUB-BAND ANALYSIS BASED ON COSINE TRANSFORM

Abstract

Subject of Research. We propose an approach to the research of the frequency properties of the sea surface images sequence (video frames) on the basis of sub-band analysis within the cosine transform. The study of frequency properties makes it possible to solve the problem of object detection on the agitated sea surface. Method. The main theoretical propositions of the sub-band analysis within the cosine transform are given. The paper presents the method of formation of the cosine transform sub-band matrices used in data analysis and method of calculation of the analyzed vector energy parts in different frequency intervals. The sub-band approach to the analysis of the image component energy distribution in the frequency domain is described. Realization of separate components of the sea surface image sequence (video frames) in the vector form is shown. The vectors contain the pixel values with the same coordinates in the image set and give the possibility to calculate the energy parts in specified frequency intervals. Main Results. Video of the sea surface with various agitation levels were used in computing experiments. The distribution analysis of the energy part values of the matching components in the used video records showed that different sea surface agitation corresponds to different distributions of the matching component energy in the frequency domain. It has been found that at low agitation, the energy of the image components is concentrated in the low-frequency region and with the increase of the sea surface agitation, the energy part values of the corresponding image components increase in the high-frequency region. Practical Relevance. The obtained distribution features of the sea surface image component energy parts can be used when solving the detection problems for objects, floating on the sea surface, based on the analysis of their frequency characteristics.