The Depth Estimation Based On Monocular Night Vision Image

The color night vision technology is for a monochrome color image or video image processing technology. Using the high resolution and high sensitivity features of human-eye color vision to achieve color night vision image, we can enhance the understanding of scene, high-lighting the target, it's good for detecting and identificating the target faster and more accurately at the hiding, camouflage and confuse military settings, and it also can reduce the fatigue of observers. The previous colorization for night-vision technology did not take the relationship between scene depth and the color expression of the image into account, depriving the colorized night vision image of cubist vision. During the process of endowing gray image with color, if color contrast and Saturation changes in perspective science are applied to differentiate and express the depth of the scene, the colorful night vision image will be ultimately generated with stereoscopic effect. This paper conducts the exploratory study in theory on the relation between color expression of image and scene depth.The paper which applys the estimation of the depth in the field of the color night vision, introduce an infrared image color algorithm which based on the monocular depth estimation. To the field of vechicle night vision applications, for the problem of estimating the depth from the night vision infrared image, according to the characteristics of infrared images, we use maximum between-cluster variance and the horizon detection methods to get a depth map of an infrared image. Then according the relationship between saturation, hue, and the depth in perspective, we transfer colors to the infrared image and obtain a color infrared image with three-dimensional visual perception.