Browsing by Author "Lin, C-T."

Now showing 1 - 3 of 3

A bio-inspired computer fovea model based on hexagonal-type cellular neural networks
(2006)
Huang, C-H.
;
Koeppl, H.
;
Lin, C-T.
;
電控工程研究所
;
Institute of Electrical and Control Engineering
In this work we propose a novel computer fovea model based on hexagonal-type Cellular Neural Networks (hCNN). The hCNN represents a new image processing architecture that is motivated by the overwhelming evidence for hexagonal image processing in biological systems. The necessary new coupling templates and basic hCNN image operators are introduced. The fovea model includes the biological mechanisms of the photoreceptors, the horizontal cells, the ganglions, the bipolar cells, and their cooperation. Thus the model describes the signal processing from the optical stimulation at retina to the output of the ganglion cells. Different building blocks of the model turned out to be useful for practical image enhancement algorithms. Two such applications are considered in this work, namely the image sharpness improvement and the color constancy algorithm.
EEG correlates of spatial orientation in the human retrosplenial complex
(2015-10-15)
Lin, C-T.
;
Chiu, T-C.
;
Gramann, K.
;
資訊工程學系
;
腦科學研究中心
Studies on spatial navigation reliably demonstrate that the retrosplenial complex (RSC) plays a pivotal role for allocentric spatial information processing by transforming egocentric and allocentric spatial information into the respective other spatial reference frame (SRF). While more and more imaging studies investigate the role of the RSC in spatial tasks, high temporal resolution measures such as electroencephalography (EEG) are missing. To investigate the function of the RSC in spatial navigation with high temporal resolution we used EEG to analyze spectral perturbations during navigation based on allocentric and egocentric SRF. Participants performed a path integration task in a clearly structured virtual environment providing allothetic information. Continuous EEG recordings were decomposed by independent component analysis (ICA) with subsequent source reconstruction of independent time source series using equivalent dipole modeling. Time-frequency transformation was used to investigate reference frame-specific orientation processes during navigation as compared to a control condition with identical visual input but no orientation task. Our results demonstrate that navigation based on an egocentric reference frame recruited a network including the parietal, motor, and occipital cortices with dominant perturbations in the alpha band and theta modulation in frontal cortex. Allocentric navigation was accompanied by performance-related desynchronization of the 8-13 Hz frequency band and synchronization in the 12-14 Hz band in the RSC. The results support the claim that the retrosplenial complex is central to translating egocentric spatial information into allocentric reference frames. Modulations in different frequencies with different time courses in the RSC further provide first evidence of two distinct neural processes reflecting translation of spatial information based on distinct reference frames and the computation of heading changes. (C) 2015 Elsevier Inc. All rights reserved.
REAL-TIME VISION-BASED VEHICLE DETECTION AND TRACKING ON A MOVING VEHICLE FOR NIGHTTIME DRIVER ASSISTANCE
(2009)
Chen, Y. -L.
;
Wu, B-F.
;
Lin, C-T.
;
Fan, C-J.
;
Hsieh, C. -M.
This study presents an effective method for detecting and tracking vehicles in front, of a camera-assisted car during nighttime driving. The proposed method identifies vehicles based on detecting and locating vehicle headlights and taillights by using the techniques of image segmentation and pattern analysis. First, to effectively extract bright. objects of interest., a fast bright object segmentation process based on automatic multilevel histogram thresholding is applied on the nighttime road-scene images. This automatic multilevel thresholding approach can provide robustness and adaptability for the detection system to be operated well under various illumination conditions at night. The extracted bright objects are processed by a spatial Clustering and tracking procedure by locating and analyzing the spatial and temporal features of vehicle light patterns, and estimating the distance between the detected vehicles and the camera-assisted car. Experimental results demonstrate the feasibility and effectiveness of the proposed method for detecting and tracking vehicles at night.