A mixed reality space image generation method for generating a mixed reality space image formed by superimposing a virtual space image onto a real space image obtained by capturing a real space, includes a first image superimposing step of superimposing a first virtual space image onto a real space image based on an occlusion by a real object, wherein the first virtual space image is an outer appearance of the real object, and obtaining information of location and orientation of the real object. In addition, a second virtual image is set based on the information of location and orientation of the real object, and a second image superimposing step superimposes the second virtual space image onto the superimposed image generated in the first image superimposing step without considering the occlusion, wherein the second virtual space image is an annotation associated with the real object.