It is very important for a mobile robot to estimate its current position. Without accurate information about the current position, a successful path planning is impossible to a mobile robot. The perspective image method applied to most vision systems has several flaws in finding objects. It operates at changing the size and the shape of objects and it is not good enough to discriminate objects and surrounding entities, since environment is radiating from the FOE(Focus Of Expansion), while a camera moves. To make up for these weak point in the current method using perspective image, the localization method using the log-polar transform and the HSI color model was examined in this study. The log-polar transform was used to simplify the matching algorithm and the HSI color model to find objects more robustly. The input image was divided into four regions using log-polar transform at vanishing point, then it was quite simpler to find objects for the localization. The position of doors and indoor lights in the log-polar domain helped the localization in short time by the indexing database of the coded input image. This study showed that the log-polar transform and the HSI color model worked effectively in finding a position of an indoor mobile robot.