A dynamic texture is a linear dynamical system used to model a video sequence. Since
the dynamic texture is a generative probabilistic model, the KL divergence can be used
to compute distances between different dynamic textures. In this note, we derive the
KL divergence between dynamic textures. In Section 2, we start by reviewing the
probability distributions of the dynamic texture model. In Section 3, we derive the KL
divergence between the state spaces of two dynamic textures. In Section 4, we define
the KL divergence between the image spaces of two dynamic textures, and in Section
5, we derive a set of recursive equations for efficiently computing the image space
KL divergence. The recursive equations are computationally efficient and require less
memory storage than the non-recursive counterpart.