Combined Motion (Direction) and Appearance (Color) Anomaly Detection in Videos |
Author(s): |
| Harshu Dutt Parashar , Department of Electronics and Communication Engineering, Y.M.C.A. University ; Neetu Gupta, Department of Electronics and Communication Engineering, Y.M.C.A. University |
Keywords: |
| Combined Motion (Direction), Appearance (Color) Anomaly Detection |
Abstract |
|
Video surveillance plays a significant role in both academia as well as industry. Video anomaly detection is an important part of intelligent video surveillance, as it can automatically detect the anomalous events in the monitoring scene and produce alarms that assist the security officers to handle the unexpected events. This paper introduces a new method to refine the detection of anomalies in videos. Indeed, a new approach is established which combines motion estimation and color detection. This approach is able to identify moving objects from stationary ones using optical flow technique and then detect only red colored cars in a scene. An experimental study is performed over a video dataset. The obtained results show that the new method can successfully combine the two algorithms and thus produce better and accurate results. |
Other Details |
|
Paper ID: IJSRDV5I60251 Published in: Volume : 5, Issue : 6 Publication Date: 01/09/2017 Page(s): 831-835 |
Article Preview |
|
|
|
|
