A fast stitching method for container images using texture and weighted speed

With the rapid development of oceans economy, a huge number of shipping containers are transported all around the world. To reduce the risk of container damages during the transportation, existing solution relies mainly on human beings to observe the container appearance before or after it enters a dock, which is time-consuming and inaccurate. To solve this problem, one intelligent approach is to develop an automatic container damage detection framework based on computer vision techniques. But how to obtain the panorama images for container damage detection is a challenging issue. In this paper, a real-time container panorama producing system is developed based on container surveillance videos, which is implemented by container image stitching with texture features. When there is no reliable offset, the weighted speed for splicing is used. Experimental results indicate that the developed system could achieve approving results in a real-time manner.