The arc in the vicinity of the molten pool in the process of CO2 welding has a complex distribution. Changes of the arc limit the development of seam tracking technology based on vision seriously. Therefore, exploring and studying the image's gray distribution, which is captured by the CCD camera in the vicinity of the molten pool, can provide the basic conditions for the selection of suitable welding current and the irradiated area, thus preparing for getting a clear picture. Firstly, image acquisition system which can be flexibly adjusted is established. Secondly, the numerical model of gray value in the area near the welding pool under different welding currents is built with the toolbox called cftool in Matlab. Experimental results show that: with the increase of welding current, the gray value of the image, which is captured by the CCD camera near the weld pool, is increased slightly; with the increase of distance, the gray value of the image, which is captured by the CCD camera near the weld pool, is reduced. When the distance between torch and the area is less than 10mm, the gray value decreased sharply. When the distance between torch and the area is more than 10mm, the gray value decreased slowly. The fitting results conform to the objective laws. It lays the foundation for the welding seam automatic tracking technology.