The projectors that are on sale now have the ability to pre-calibrate the non-white surface for distortion-free image projection. Although it is only possible in monochromatic pastel tones, it is an effort to reduce the distortion of the image that the viewer sees by neutralizing the projection plane color through the user s choice. Image correction studies are being conducted to further correct distortions caused by surface color using a projector-camera system. However, current projector-camera systems have to recalculate the existing linear lighting color model for image correction, because the texture for neutralization changes as the projection plane changes. We propose a method of color correction through artificial neural network, away from existing image correction modeling. Especially, when the camera is embedded in the projector, it is expected that the texture of the projection plane can be neutralized more quickly if the proposed neural network based image projection technology is applied. In order to overcome this problem, we are developing a color correction algorithm that responds sensitively to changes in the projection plane and copes with changes in the projection plane.