Indoor place recognition based on set operation of keypoints databases

In this paper, we proposed a new algorithm based on independent keypoints databases for indoor place recognition. In analogy with set operation, a new kind of operations for keypoints sets are defined to describe the process of independent keypoints database establishment and place classification. To obtain the databases, keypoints are firstly extracted from sample images whose class are known, and then they are processed by set operation. A serious of experiments are conducted to test the influence from the parameters of the algorithm on the effect of the algorithm based on the INDECS database and verified that the algorithm is not sensitive to parameters and changing environment. Then an integrated experiment was developed to test the performance of this algorithm, and verified that it is a robust algorithm with high precision.