A necessity exists for the protection of people and their property against the unauthorized use of photoelectric devices. The retroreflection of optical systems can be utilized to detect and recognize photoelectric devices. The reduction of false alarms and processing speed are two crucial issues, particularly for interior applications involving complex scenes. Here, we propose a local texture criterion that determines the local maximum and local continuity of the reflections. A comprehensive evaluation method that combines a modified shape criterion and the local texture criterion to improve the recognition probability is also presented. A searching strategy for scanning and locating the target candidates with reduced time cost was used in laboratory experiments performed at operating distances of 2.2, 3.2, and 4.2 m with the presence of disturbing objects. The results demonstrated the practicability of the proposed method, its superior recognition probability over that of single-criterion methods, and its enhanced average processing speed in comparison with existing methods.