BRAND (Binary Robust Appearance and Normals Descriptor) is a robust descriptor that efficiently combines appearance and geometric shape information from RGB-D images, and is largely invariant to rotation and scale transform. This encodes point information as a binary string, so it is suitable for applications that demand low memory consumption and real-time performance. However, since the BRAND algorithm performs an OR operation between appearance and geometric information, there is ambiguity that can cause an erroneous correspondence between different points. Therefore, this study proposes a method to eliminate the ambiguity by modifying the operation algorithm of a BRAND descriptor. Experimental results show that the proposed method has higher precision than the conventional method by better using both texture and geometric information.