Recognition of maritime objects based on FLIR images using the method of eigenimages

The paper presents the method of recognition of marine objects based on their images taken by infrared sensors (FLIR — forward looking infrared) using the method of eigenimages analysis. This method is based on the principal component analysis (PCA) method that allows to reduce the dimensionality of the problem of image recognition. The paper presents a method of maritime objects recognition based on FLIR images consisting of the following stages: pre-processing of images from the pattern database, calculation of coordinates of images from the pattern database in a new reduced feature space, calculation of coordinates of image cluster centers, pre-processing of a new image, calculation of coordinates of a new image in the reduced feature space, calculation of the distances of the recognized image from all the image cluster centers in the pattern database and making a decision on choosing the most similar maritime object, or stating that there is no image satisfying the similarity condition in the pattern database. The combination of the principal component analysis (PCA) method with the method of eigenimages analysis reduces the dimensionality of the recognition problem. The original dimension is equal to the number of pixels in each image. It must be equal for all the images from pattern database and the image being recognized. The final value of the problem dimension is equal to the number of image clusters in the pattern database. In some situations, the dimensionality of the problem can be reduced due to the low eigenvalues of the matrix of Karhunen-Loeve transformation. The final part of the paper presents the preliminary results of the study of the developed method of classifying maritime objects using a certain set of FLIR images registered in the Baltic Sea.