Development of recognition methods for night vision applications

Abstract

Video surveillance is essential to all modern security systems that newlineallow us to monitor various objects, including locations, monuments, buildings, newlineand people. The video surveillance system utilizing digital cameras is pervasive newlineand most extensively used for safety and security in everyday life. However, one newlineof the most significant issues with surveillance systems is the ambient lighting newlinechange for nighttime surveillance. This occurs more frequently outdoors, where newlinelighting conditions vary naturally. Occasionally, the environment can be newlinecompletely dark, making nighttime surveillance systems more complex. newlineNight vision allows seeing in low or complete darkness by amplifying newlineavailable light or using infrared technology. It detects and displays objects that are newlinenot visible to the naked eye under the dim light scenario. There are several reasons newlinewhy night vision may be needed: 1. Security, 2. Surveillance, 3. Safety, newline4. Property protection, 5. Military, and 6. Sustain Law and order. The NCRB newline(National Crime Records Bureau) reports that over 60% of burglaries occur newlineat night, most occurring between 6 p.m. and 6 a.m. Overall, night vision is helpful newlinein situations with limited visibility and is necessary for performing a task or newlinemaintaining safety. newlineIn recent years, deep learning has been applied to many industries newlineincluding surveillance systems with breakthrough results compared to legacy newlinesystems. Deep learning in its infancy has shown a lot of promise in improving newlinesome hard, and difficult video surveillance problems. Much more work must be newlinedone to fine-tune the generic deep learning system to learn and detect newlinedomain-specific events unique to Night time surveillance environments. newline newline