[Solution] An image generation apparatus (100) generates an output image from input images of three cameras. An input image part corresponding to an area, in which the image capture ranges of the cameras overlap each other, forms a grid pattern in the output image. A grid, which is obtained by partitioning the grid pattern by use of a circle delineated by a wave the wave source of which is one of the cameras (2B) and also by use of a circle delineated by a wave the wave source of which is one of the cameras (2R), is defined as unit pattern. The values of the pixels forming a unit pattern are obtained by weighted-averaging the pixel values of a first input image of the camera (2B) and the pixel values of a second input image of the camera (2R). For the pixels forming one of a unit pattern through which an antinodal line formed by the two waves passes or a unit pattern through which a nodal line formed by the two waves passes, the average value of the weights (wB) of the values of the pixels of the first input image is equal to or greater than the average value of the weights (wR) of the values of the pixels of the second input image. For the pixels forming the other, the average value of the weights (wB) is equal to or less than the average value of the weights (wR).