Smart video summarization using mealy machine-based trajectory modelling for surveillance applications

Abstract

In this paper, we propose a smart video summarization technique that compiles a synopsis of event(s)-of-interest occurring within a segment of image frames in a video. The proposed solution space consists of extracting appropriate features that represent the dynamics of targets in surveillance environments using their motion trajectories combined with a finite state automaton model for analyzing state changes of such features to detect and localize event(s)-of-interest. We introduce the cumulative moving average (CMA) and the preceding segment average (PSA) statistical metric as features that indicate gradual and sudden changes in the instantaneous velocity of moving targets. In order to support both on-line and off-line summarization, a finite state machine, that is often referred to as Mealy Machine, has been proposed to model the trajectory of a moving target and used for detecting transitions that represents a change from one state to another when initiated by a triggering event or condition. We conduct several systematic experiments on different scenario-specific in-house videos and other publicly available datasets to demonstrate the effectiveness of our proposed approach and benchmark its performance against chosen baseline strategies. The results of our experiments highlight the superiority of our proposed Multimed Tools Appl method in accurately localizing the start and end of event(s)-of-interest in videos within the chosen dataset.

Identification of Trajectory Anomalies on Video Surveillance Systems

“…Simultaneous advancement in multi−object tracking (Walia, & Kapoor, 2016) has presented development in moving object tracking in random and complex environments. Investigators have started concentrating more on detecting abnormality (Dogra, Ahmed, & Bhaskar, 2016) due to adequate tracking results.…”

“…Edge Oriented Histogram (EOH) and Multi-layer Histogram of Optical Flow (MHOF) are two suggested techniques for detecting anomalies that represent appearances and motion, respectively (Cong, Yuan, & Tang, 2013). A different approach makes use of the shift in the temporal pattern by computing Markovian differences from the local pattern while the time scale is modelled globally (Dogra, Ahmed, & Bhaskar, 2016). With the use of the Gaussian Mixture Model (GMM), which uses mean shift to calculate location, speed, and direction, it is possible to identify vehicles in video and determine whether or not an accident will occur (Hui, Yaohua, Lu, & Jiansheng, 2014).…”