Video Synchronization and Alignment Using Motion Detection and Contour Filtering

Abstract

The proposed method presents a proficient abandoned functioning of a video synchronization and alignment using motion detection and contour filtering, based on various flat dimensionality frame matching techniques. In the proposed system, motion detection algorithm is used to detect only the motion of the objects and Contour filtering algorithm is used to recognize the objects based on its color. The algorithms are implemented in Java language, which facilitates prototyping using open source library. The application of video alignment includes the detection of objects such as vehicles, which is used in Advance Driver Assistance System (ADAS) and also in video surveillance system for traffic monitoring. The motion detection algorithm is used in CCTV surveillance for detecting terrorist threats. The contour filtering algorithm is implemented in medical examinations. The proposed system is tested on live datasets and obtained a good change detection between the frames. Video synchronization and alignment algorithms had been developed in the earlier years for plain datasets using static cameras. Compared to the algorithms developed in the initial stages, the proposed system provides a better efficiency to speedup the application by optimizing the algorithms, recuperating the data locality and also executing the various modules of the application. The proposed system has obtained a chronological speed up result of 12.39x factor when compared to the existing methods, when processed with the testing dataset with the video resolution of 240 * 320, with 30 frames per second using high definition cameras. The results obtained are further processed to run the embedded CPU applications and GPU processors.