Cloud Storage Optimization for Video Surveillance Applications

The need for automated security devices are increasing at an unexpected rate. Operator controlled video surveillance is tiresome and hence automated video surveillance systems are on-demand. Automatic detection of moving objects can be greatly used in places like forest borders and other isolated places, where this algorithm can be used to save tremendous amounts of data that will subsequently reduce the costs. The method proposed in this paper estimates the motion using KNN algorithm. When motion is detected, the qualitative speed is determined by comparing the Euclidean distance between two successive frames following which an optimal threshold value is set to determine and store only the key frames in order to optimize the storage. The recorded videos with reduce the size are continually stored in the local storage (PC), and are uploaded to the cloud server at midnight and are deleted from the PC. The video recorded by the webcam is simultaneously live streamed to an IP address based web page. When the activity is detected, it triggers an alarm by generating automatic emails to the specified mail ID. The system has been testedin an indoor setting and the size is also reduced.