AbstractThis paper presents a model of a system for invariant object recognition, which consists of five stages. The first stage shifts the object so that the centroid of the object coincides with the center of the image plane. The second stage is an application of the polar-coordinate transforms used to obtain N-dimensional vectors-representations of the input object. In this stage, any rotation of the input object becomes a cyclic shift of the output value of this stage. The third stage employs CT (Certain Transform), a class of shift-invariant transformations to provide invariant representations for cyclically shifted inputs. The next stage normalizes the outputs of the previous stage to obtain scale invariance. The final stage realizes a classification.
Algorithm of traffic signs recognition based on the rapid transform