The acquisition of complex data of the recorded scene is an important task for video analysis of the processes occurring. The use of IR images allows obtaining data on additional characteristics that are not visible in the optical range. The data obtained by IR sensors can be in the near and far range, which makes it possible to see objects in the dark or to obtain data on their temperature. In the second case, the boundaries of objects are vague and difficult to correlate with the usual optical ranges. To do this a combination of data obtained by a pair of cameras is used. In this paper, we propose using the algorithm for stitching IR images based on data obtained in the optical range. To this end, an approach will be applied that includes parallel analysis of data on: saliency maps; search for boundaries and key points; reduction of bit resolution of images with preservation of borders; image matching; filtering data and restoring the sharpness of object boundaries. As an example, the result of combining data obtained under poor lighting conditions and the results of combining television images will be presented.
|