A method of recognition of maritime objects based on FLIR (forward looking infra-red) sensor images using dynamic time warping

Abstract

This paper presents a method of recognition of maritime objects based on their images made by infrared sensors (FLIR – forward looking infra-red) using the time series comparison DTW method (DTW - Dynamic Time Warping). The DTW method allows to find the smallest distance between two time series when the run of time one of the series has been deformed (stretched or compressed). In the presented classifier of maritime objects images the DTW method is used to compare the combined horizontal and vertical brightness histograms for a recognized object and pattern objects. The DTW method allows to compare the histograms of objects whose FLIR images were taken at different angles. To determine the silhouette of a maritime object the Otsu segmentation algorithm is used in this paper. The paper describes the Otsu threshold method, the method of comparing time series DTW and the method of constructing combined histograms of maritime objects silhouettes. The final part of the paper presents the results of research on the developed method of maritime objects classification using a set of FLIR images registered in the Baltic Sea.