Vessel Monitoring Systems Based on Aerial Images in Sea Environment

The utilization of aerial image capturing devices such as satellites and drones in the vessel monitoring systems has provided us with powerful tools and challenges at the same time. Some of this challenges would come from large amount of images that needs to be analyzed along side with a typical natural image noise sea environment could impose. To automate the recognition process, image information systems needed to be implemented in such cases. Most of these processes involve recognizing an object in a scene, whether the object possesses similar characteristics to the reference image. There are many methods that can be applied to the schema. One of them is computer vision technique, which enables the systems to automatically identify objects within a given image samples. This work compares the implementation of such method on aerial images for sea environment monitoring system. Two of the most promising methods are SURF and SIFT. They both work similarly as descriptor which involve application of integral image, where storing information inside descriptors enable users to describe images that is needed for matching. In open sea scenario, it is important to be able to recognize objects from higher altitude from any view point, therefore this paper will be focusing more on evaluating descriptor and matching methods toward aerial images through detected key points and detected matching points. This work as well will conclude which of the methods are best to be implemented on analyzing object in sea environment samples based on the evaluation criteria.