Application of graph segmentation method in thermal camera object detection

The paper presents the application of graph-based segmentation algorithm in image object detection. The input images are taken from thermal camera for night surveillance application. The graph-based algorithm was selected due to its low complexity, which allows us to process each image with the complexity of O(n.log(n)) where n is the number of pixels, and due to the fact that thermal images contains smaller number of regions of colors. With the detected regions, some additional measures are used to filter out the artifacts to correctly detect the object in the images. The numerical results have proved the high quality of the proposed solutions.