This paper contains overview of following approaches for object recognition in videostream: feature detection, points of interest detection, correlation analysis, machine learning. This overview helped to select method to build the library for videostream analysis. Using this library different systems for videostream analysis can be built. These systems can be used for videosurveillance, biometric identification, indexation and search. The library was tested for different brightness and contrast changes, for different noise levels.
objects recognition, videostream, point of insterest, detector, corelation, machine learning, library for videostream analysis
"Analiz metodiv rozpiznavannia obiektiv v systemakh analizu vizualnoi informatsii" ,
Information Processing Systems,