Urban surveillance systems will be more demanding in the future towards smart city to improve the intelligence of cities. Big data analytics and processing for urban surveillance systems become increasingly important research areas because of infinite generation of massive data volumes all over the world.
This thesis focused on solving several challenging big data issues in urban surveillance systems.
First, we proposed several simple yet efficient video data recoding algorithms to be used in urban surveillance systems. The key idea is to record the important video frames when cutting the number of unimportant video frames.
Second, since the DCT based JPEG standard encounters problems such as block artifacts, we proposed a very simple but effective method which results in better quality than widely used filters while consuming much less computer CPU resources.
Third, we designed a novel filter to detect either the vehicle license plates or the vehicles from the images captured by the digital camera imaging sensors. We are the first to design this kind of filter to detect the vehicle/license plate objects.
Fourth, we proposed novel grate filter to identify whether there are objects in these images captured by the cameras. In this way the background images can be updated from time to time when no object is detected.
Finally, we combined image hash with our novel density scan method to solve the problem of retrieving similar duplicate images.