This paper proposes a method that fuses two kinds of stereo measurements using a fisheye stereo camera. A stereo measurement method can be divided into two types: binocular and motion stereo. Generally, the accuracy of long range depends on the baseline of the camera. In case of the binocular stereo, the direction of the baseline is horizontal. Its size is small due to the convenience of installation conditions. By contrast, in the case of the motion stereo, the direction of the baseline changes depending on the direction of the camera motion. For example, the direction of the baseline is the optical axis when a stereo camera is mounted on a vehicle or a mobile robot that drives in the forward direction. When the speed is high such as in driving, its baseline becomes larger as it becomes the moving distance between frames. Therefore, between the two stereo measurements there are differences in the magnitude of the baseline and the uncertainty of the position in the image. In addition, the area-based approach is used for binocular stereo and the feature-based approach is used for motion stereo at the time of corresponding point search. So, robustness against false matching is different. In this paper, we try to fuse the two stereo measurements in order to realize more accurate range image generation. A method, which restricts the disparity search range using the reference disparity map and a bilateral-like filter, which is a weighted averaging are proposed for fusing the two stereo measurements. The proposed methods are verified by experiments in indoor and outdoor environments.