Modern welding process tends towards auto-welding and asks for efficient quality monitoring measurements. Much work has been done in this area. The application of image processing in welding can help us in controling the process and quality by monitoring the location and status of the dynamic welding molten pool and also make it possible to inspect the flaw of welding without scathing the workpiece. The digital image processing technique using in welding is proved to be superior and very attrachine within and outside China. By using machine vision we can get digital images with direct and rich image information of welding molten pool and plasma, and transfer them into computer cache at the same time. The digital images can be manipulated in real time using efficient image manipulation algorithm to distill the characteristic information. Then we can use the characteristic information as a controlling signal of execute unit and realize the automatization of welding quality controlling. This paper demonstrates the status of application of image processing in welding including the method and principle of image mutograph, the definition of the melting pool characteristics and the technique of digital image processing by referring to a lot of relative papers in order to prospect the technique of digital image and to promote the advance of auto-welding.