HONN approach for automatic model building and 3D object recognition

This work presents a method for automatic model building from multiple images of an object to be recognized. The model contains knowledge which has been computed during the learning phase from a large 2D images of an object. This knowledge is the invariant features including the object itself, and is extracted by a higher-ordered neural network (HONN) structure. In the recognition phase, independent-viewpoint 2D stereo images of the object are taken and the invariant features are extracted from it, and compared to the models stored in the database. Both model and recognition algorithms are tested practically to get a optimal compact model with acceptable recognition rate