Real time digital night vision using nonlinear contrast enhancement

This thesis describes a nonlinear contrast enhancement technique to implement night vision in digital video. It is based on the global histogram equalization algorithm. First, the effectiveness of global histogram equalization is examined for images taken in low illumination environments in terms of Peak signal to noise ratio (PSNR) and visual inspection of images. Our analysis establishes the existence of an optimum intensity for which histogram equalization yields the best results in terms of output image quality in the context of night vision. Based on this observation, an incremental approach to histogram equalization is developed which gives better results than the conventional approach in terms of PSNR. This algorithm is also applied to implementing video surveillance in poorly illuminated environments to achieve real time night vision. This involves the application of histogram equalization to digital video frames with data transmission and buffering over a computer network.