Abstract
Color night vision technology can effectively improve the detection and identification probability. Current color
night vision method based on gray scale modulation fusion, spectrum field fusion, special component fusion and
world famous NRL method, TNO method will bring about serious color distortion, and the observers will be visual
tired after long time observation. Alexander Toet of TNO Human Factors presents a method to fuse multiband night
image a natural day time color appearance, but it need the true color image of the scene to be observed. In this paper
we put forward a color night vision method based on the correlation between natural color image and dual band
night image. Color display is attained through dual-band low light level images and their fusion image. Actual color
image of the similar scene is needed to obtain color night vision image, the actual color image is decomposed to
three gray-scale images of RGB color module, and the short wave LLL image, long wave LLL image and their
fusion image are compared to them through gray-scale spatial correlation method, and the color space mapping
scheme is confirmed by correlation. Gray-scale LLL images and their fusion image are adjusted through the
variation of HSI color space coefficient, and the coefficient matrix is built. Color display coefficient matrix of LLL
night vision system is obtained by multiplying the above coefficient matrix and RGB color space mapping matrix.
Emulation experiments on general scene dual-band color night vision indicate that the color display effect is
approving. This method was experimented on dual channel dual spectrum LLL color night vision experimental
apparatus based on Texas Instruments digital video processing device DM642.
