Refine
Labor/Institute
Keywords
- Fahrerassistenzsystem (50)
- Neuronales Netz (29)
- Fußgänger (27)
- Mikroelektrode (24)
- Array (17)
- Software Engineering (16)
- Herzmuskelzelle (14)
- Robuste Regelung (14)
- Bildverarbeitung (13)
- Biosensor (9)
Year of publication
Document Type
Objective. Spike detection from extracellular recordings is a crucial preprocessing step when analyzing neuronal activity. The decision whether a specific part of the signal is a spike or not is important for any kind of other subsequent preprocessing steps, like spike sorting or burst detection in order to reduce the classification of erroneously identified spikes. Many spike detection algorithms have already been suggested, all working reasonably well whenever the signal-to-noise ratio is large enough. When the noise level is high, however, these algorithms have a poor performance. Approach. In this paper we present two new spike detection algorithms. The first is based on a stationary wavelet energy operator and the second is based on the time-frequency representation of spikes. Both algorithms are more reliable than all of the most commonly used methods. Main results. The performance of the algorithms is confirmed by using simulated data, resembling original data recorded from cortical neurons with multielectrode arrays. In order to demonstrate that the performance of the algorithms is not restricted to only one specific set of data, we also verify the performance using a simulated publicly available data set. We show that both proposed algorithms have the best performance under all tested methods, regardless of the signal-to-noise ratio in both data sets. Significance. This contribution will redound to the benefit of electrophysiological investigations of human cells. Especially the spatial and temporal analysis of neural network communications is improved by using the proposed spike detection algorithms.
We present a method to detect starting, stopping and bending in intentions of pedestrians from a moving vehicle based on stereo-vision. The method focuses on urban scenarios where these pedestrian movements are common and may result in critical situations. Pedestrian intentions are determined by means of an image-based motion contour histogram of oriented gradient descriptor. It is based on silhouettes gathered from stereo data and does not require any compensation of appearance changes resulting from the ego-motion of a vehicle. Nevertheless, it covers small movements indicating a pedestrian’s intention. A linear support vector machine with probabilistic estimates is used for classification. We evaluated our method on the publicly available Daimler Pedestrian Path Prediction Benchmark Dataset containing detections of a stateof-the-art pedestrian detector. We detect a pedestrian’s stopping intention from 125 ms to 500 ms before standing still within an accuracy range of 80% to 100%. Bending in is detected from 320 ms to 570 ms after a first visible lateral body movement in the same accuracy range. The intention to cross the road from standing still (starting) is detected 250 ms after the first visible motion and, therefore, within the first step with an accuracy of 100%.