This paper presents the evaluation results for conventional methods that can be used for vision-based localization. An Active Scope Camera is a very thin snake robot and can be used as a rescue robot for search and rescue missions. Self-position estimation of the Active Scope Camera is important for efficient search. Nevertheless, using sensors for this purpose hinders the movement and maneuvering of the camera through narrow gaps, because sensors are very big and heavy for the Active Scope Camera. Vision-based localization using a fish-eye camera is suitable technique for self-position estimation. However, the images obtained using the Active Scope Camera are not of good quality. The material of objects found in disaster environments and overexposure by light-emitting diodes embedded at the camera tip affects the matching of feature points. In this paper, properties of images of disaster sites obtained using the Active Scope Camera and the accuracy evaluation of vision-based localization are described.