Quantifying Motion in Video Recordings of Neonatal Seizures by Robust Motion Trackers Based on Block Motion Models

Abstract

This paper introduces a methodology for the development of robust motion trackers for video based on block motion models. According to this methodology, the motion of a site between two successive frames is estimated by minimizing an error function defined in terms of the intensities at these frames. The proposed methodology is used to develop robust motion trackers that rely on fractional block motion models. The motion trackers developed in this paper are utilized to extract motor activity signals from video recordings of neonatal seizures. The experimental results reveal that the proposed motion trackers are more accurate and reliable than existing motion tracking methods relying on pure translation and affine block motion models.

An improved procedure for the extraction of temporal motion strength signals from video recordings of neonatal seizures

“…This would be sensible only if the performance gain offered by optical flow computation is high enough to compensate for its computational requirements, which exceed by far those of the procedure proposed in this paper. The discrimination between focal clonic seizures and random infant movements may also be reinforced by combining motion strength signals with motion trajectory signals extracted from video recordings by motion trackers based on adaptive block matching [33] or block motion models [34,35]. …”

“…Karayiannis, G. Tao / Image and VisionComputing 24 (2006) [27][28][29][30][31][32][33][34][35][36][37][38][39][40] …”

Training Reformulated Radial Basis Function Neural Networks Capable of Identifying Uncertainty in Data Classification

“…After an example is presented to the RBFNN, the weight vectors , and the prototypes , can be updated according to (9) and (10), respectively. Assuming that each of the outputs of the cosine RBFNN represents one of the classes , the reference distances are updated by minimizing the objective function formed by summing over all classes and all RBFs (15) Following the presentation of all examples and the corresponding updates of the output weights and the prototypes, the new estimate of each reference distance , can be obtained by incrementing its current estimate by the amount as (16) where is the learning rate, which can be a fraction of the learning rate used for updating the output weights and the prototypes, , and . In this case, each adaptation cycle involves the update of the output weights and the prototypes following the sequential presentation of all examples included in the training set.…”

Training Reformulated Radial Basis Function Neural Networks Capable of Identifying Uncertainty in the Recognition of Videotaped Neonatal Seizures

“…This problem was dealt with by developing robust motion trackers based on block motion models. Such motion trackers were specifically designed to suppress the effect of noise [10].…”