IR and Visible Video Fusion for Surveillance

Infrared (IR) imagery offers a promising alternative to visible imagery and is extensively used in military, surveillance and other applications. IR, however, has limitations of not detecting thermal variations after heavy rains when the temperature of the surrounding becomes uniform with the object of interest. Due to this fact, the relevant information from visible and IR videos are combined into one fused video. This paper aims to develop an efficient algorithm in order to fuse videos captured using infrared and visible cameras using different types of transforms and evaluate its performance measures. Tracking of single or multiple objects/people from the fused result has been further undertaken specifically for night vision using Background Subtraction[BS] and Kalman filtering.