A biosonar measurement tool (BMT) was created to investigate dolphin echolocation search strategies by recording echolocation clicks, returning echoes, and three-dimensional angular motion, velocity, and depth of free-swimming dolphins performing open-water target detections. Trial start and stop times, locations determined from a differential global positioning system (DGPS), and BMT motion and acoustic data were used to produce spatial and acoustic representations of the searches. Two dolphins (LUT, FLP) searched for targets lying on the seafloor of a bay environment while carrying the BMT. LUT searched rapidly (< 10 s), produced few clicks, and varied click-peak frequency (20-120 kHz); FLP searched relatively slowly (tens of seconds) and produced many hundreds of clicks with stereotypical frequency-dependent energy distributions dominating from 30-60 kHz. Dolphins amplified target echo returns by either increasing the click source level or reducing distance to the target but without reducing source level. The distribution of echolocation click-peak frequencies suggested a bias in the dominant frequency components of clicks, possibly due to mechanical constraints of the click generator. Prior training and hearing loss accommodation potentially explain differences in the search strategies of the two dolphins.

No abstract

No abstract

Wavelet Packets (WPs) bases are explored seeking new discriminative features for texture indexing. The task of WP feature design is formulated as a learning decision problem by selecting the filter-bank structure of a basis (within a WPs family) that offers an optimal balance between estimation and approximation errors. To address this problem, a computationally efficient algorithm is adopted that uses the tree-structure of the WPs collection and the Kullback-Leibler divergence as a discrimination criterion. The adaptive nature of the proposed solution is demonstrated in synthetic and real data scenarios. With synthetic data, we demonstrate that the proposed features can identify discriminative bands, which is not possible with standard wavelet decomposition. With data with real textures, we show performance improvements with respect to the conventional Wavelet-based decomposition used under the same conditions and model assumptions.INDEX TERMS Texture indexing, wavelet packets, minimum probability of error, complexity regularization, minimum cost tree pruning.

No abstract