Window Intensity Test Detector and Descriptor

In this paper, we present a novel algorithm for feature point detecting and matching. Algorithm is robust to image noise and computationally simple, hence it is more suitable for on-board processing robotics applications. We introduce two repeatability measurements, that can be used in training phase to optimize parameters of the algorithm. Proposed detection algorithm is free from extracting multiple points around the same feature point. We suggest a technique to overcome the effect of scale variation in detection stage, for a compensation of computational simplicity. In matching stage, we directly utilize resulting data from detection, which makes the overall process much faster. We procure a technique for matching point error suppression, for a compensation of computational simplicity. Simulation results provided in the paper illustrate the accuracy and efficiency of the proposed algorithm.