The object is to generate an image in which an overlapped part of a tuned object and a non-tuned object having a contour line drawn therein is naturally represented. Only tuned objects among objects existing in a virtual three-dimensional space are perspective transformed, and image data and depth data of each tuned object is written into a process buffer 163a for each pixel. The contour of each tuned object is detected based on the depth data and normal line data acquired from the perspective transformation process, and contour line data is written into a process buffer 163c together with the depth data. When a non-tuned objects is perspective transformed, in a pixel in which a non-tuned objected is located on the front side of a tuned object, image data and depth data of the process buffer 163a are updated. The contour line data of the process buffer 163c is combined with the image data of the process buffer 163a, and the composed data is written into a frame buffer 112. However, the contour line data of pixels having different depth data is not written.