Study On Colorization Of Night Vision Video Based On Color Transfer

Colorization is an image processing technology to color grayscale image or video. Colorization has a wide range of applications in the film and television, medical, military, security and many other industries and scientific fields, and the technology has a lot of potential in the area of image processing. Colorization for night-vision technology was first started in the U.S. MIT laboratory, and the first used in the military field. With the computer hardware and multimedia technology development, colorization technology is also gaining attention, and its usefulness is also receiving more and more value increasingly.According to the processing method to colorization for night vision images, colorization for night vision images can be divided into MIT method, Toet method and the color transfer method. MIT method and Toet method is the classic color for night vision image enhancement technology, while the color transfer method is a new colorization method for night vision technology in recent years. Colorization technology for still image has obtained many research achievements in recent years,, while the research about the colorization technology for dynamic image is rare. This paper focused on the colorization for night-vision video, based on analyzing and summarizing the existing methods , using the different characteristics between night vision image and gray-scale image and video processing technologies, we has developed a practical technology to color night vision video.Firstly, this paper describes night-vision imaging equipment and imaging principle in detail, and thus gives a detailed account about features ofnight-vision images different with common gray-scale images. Since the result is ineffective when the Welsh algorithm is directly applied for night vision image, this paper uses image fusion technology for low-light and infrared images firstly, then colorization the fusion images. Experimental results show the feasibility of this program, and have yielded good experimental results.Secondly, we carried out the actual video capture and processing experiments using the low-light CCD and in-car night vision which we buy ourselves. We design a program to collect infrared and night-vision images of the same scene, and We treat the experimental data for registration, fusion, and colorization.Finally, according to the data collected, using Matlab simulation tools and VC++ development tool, the data obtained were processed according to the actual needs the simulation result shows the superiority and practicability of the proposed algorithm in this paper, and based on the Windows platform, using VC++ development tool to design a simple image colorization software, and was trying to develop a software to color night-vision.