Dynamic Texture Recognition Using Time-Causal Spatio-Temporal Scale-Space Filters

Abstract

Abstract. This work presents an evaluation of using time-causal scalespace filters as primitives for video analysis. For this purpose, we present a new family of video descriptors based on regional statistics of spatiotemporal scale-space filter responses and evaluate this approach on the problem of dynamic texture recognition. Our approach generalises a previously used method, based on joint histograms of receptive field responses, from the spatial to the spatio-temporal domain. We evaluate one member in this family, constituting a joint binary histogram, on two widely used dynamic texture databases. The experimental evaluation shows competitive performance compared to previous methods for dynamic texture recognition, especially on the more complex DynTex database. These results support the descriptive power of time-causal spatio-temporal scale-space filters as primitives for video analysis.

Dynamic Texture Recognition Using Time-Causal and Time-Recursive Spatio-Temporal Receptive Fields

“…We note that parameter tuning and adding the second-order temporal derivatives of the spatial derivatives, result in improved performance for our new STRF N-jet descriptor compared to the STRF N-jet (previous) descriptor [29]. The new descriptor shows improved accuracy for all the benchmarks.…”

“…The parameters are shown for the UCLA benchmarks in Table 5 and for the DynTex benchmarks in Table 6. For the STRF N-jet, STRF RotInv and RF Spatial descriptors, the parameters have been determined by cross-validation (see Section 6.3) while the STRF N-jet (previous) descriptor is tested with the heuristically chosen parameters used in [29] (and included here for completeness). In the quite frequent case that several sets of parameters give the same classification performance, only one of these parameter settings is reported here.…”

“…The rotationally invariant STRF RotInv descriptor is also a competitive option. The new STRF N-jet achieves better results than STRF N-jet (previous) [29]. All descriptors are binary.…”

“…This descriptor is included for reference and differs from STRF N-jet by lacking the second-order temporal derivatives of the first-and second-order spatial derivatives. This descriptor was evaluated in [29] without parameter tuning and we also here here keep the original parameters. It can be noted that none of these video descriptors makes use of the full spatio-temporal 4-jet.…”

“…(iii) The purely spatial RF Spatial. (iv) The previous spatio-temporal descriptor STRF N-jet (previous) [29].…”