Indoor environments are structures divided into rooms and corridors, and have doors that are used to move between those sections. Indoor localization is possible by detecting and tracking doors. In this study, a door detection method that detects doors by finding door corners in camera images has been reconsidered and changed to be able to find only the most suitable door candidate in images. Compatibly with the original study, 86.6% true positive success rate is obtained from the method. This method is planned to be used in the future as part of a vision-based indoor localization system.