AR models and bidimensional discrete moments applied to texture modelling and recognition

摘要

The concept of texture is one of the most important approaches used to describe and identify objects or regions of an image with somewhat ordered aspect. Texture recognition and classification are often tackled by resort to co-occurrence or autoregressive (AR) models with a large number of parameters. This paper presents a new method for texture recognition and classification based on characterization by the two-dimensional (2D) discrete moments of the impulse response of their AR models. Four parameters are enough to give a good representation of a deterministic or stochastic texture and to recognize it.