Video fingerprinting using AED of CGO based signatures

Video fingerprinting is the process of extracting features from a given video. These fingerprints, are used to uniquely identify a video from other videos. The goal of a video fingerprinting system is to decide whether two videos have the same contents or not, by measuring distance between fingerprints extracted from these videos. This paper proposes, a simple and fast video fingerprints matching method. The proposed method is based on matching of centroids of gradient orientation using Absolute Euclidean Distance (AED). The Absolute Euclidean Distance, is chosen for matching because of its low calculation complexity and more accurate performance as compared to other matching measures. Experimental results show that the proposed method performs fairly well for videos affected by various types of video processing techniques and distortions such as compression, resizing, blurring, cropping, rotation etc. as compared to existing SED method.