13 research outputs found
Robust corner detector based on corner candidate region
No full textCorner detection is a fundamental step for many computer vision applications to detect the salient image features. Recently, FAST corner detector has been proposed to detect the high repeatable corners with efficient computational time. However, FAST is very sensitive to noise and detects too many unnecessary corners on the noise or texture region. In this paper, we propose a robust corner detector improved from FAST in terms of the localization accuracy and the computational time. First, we construct a gradient map using the Haar-wavelet response by integral image for efficiency. Second, we define a corner candidate region which has large gradient magnitude enough to be corner. Finally, we detect the corner on the corner candidate region by FAST. Experimental results show the proposed method improves localization accuracy measured by the repeatability than standard FAST and the-state-of-art methods. Moreover, the proposed method shows the best computation efficiency. Especially, the proposed method detects the corners more accurately in the image containing many texture regions and corrupted by the Gaussian noise or the impulse noise. © 2013 IEEE
ABFT: Anisotropic binary feature transform based on structure tensor space
No full textLocal feature matching is a fundamental step for many computer vision applications. Recently, binary feature transforms have been popularly proposed to improve the computational efficiency while preserving high matching performance. However, it is sensitive to noise and geometrical distortion such as affine transformation. In this paper, we propose ABFT framework, composed of a noise robust feature detection and affine invariant binary feature description based on a structure tensor space. Experimental results show that ABFT outperforms other state-of-the-art feature transforms in terms of the repeatability, recognition rate, and computational time. © 2013 IEEE
Coherence enhancing diffusion filtering based on connected component analysis structure tensor
No full text
Training improves the capacity of visual working memory when it is adaptive, individualized, and targeted.
No full textThe current study investigated whether training improves the capacity of visual working memory using individualized adaptive training methods. Two groups of participants were trained for two targeted processes, filtering and consolidation. Before and after the training, the participants, including those with no training, performed a lateralized change detection task in which one side of the visual display had to be selected and the other side ignored. Across ten-day training sessions, the participants performed two modified versions of the lateralized change detection task. The number of distractors and duration of the consolidation period were adjusted individually to increase the task difficulty of the filtering and consolidation training, respectively. Results showed that the degree of improvement shown during the training was positively correlated with the increase in memory capacity, and training-induced benefits were most evident for larger set sizes in the filtering training group. These results suggest that visual working memory training is effective, especially when it is adaptive, individualized, and targeted
Spatial statistical analysis of the effects of urban form indicators on road-traffic noise exposure of a city in South Korea
No full textAbstractThe purpose of this study is to present a statistical model which can predict the noise level of road-traffic in urban area. A spatial statistical model which can take into account spatial dependency on geographically neighboring areas is constructed from a noise map of a city in South Korea. A system of 250m×250m grid cells is placed on the city of Cheongju, South Korea, and the noise level and urban form indicators are averaged over each cell. The population-weighted mean of the noise level is subsequently regressed on the average urban form by adopting the spatial autoregressive model (SAR) and the spatial error model (SEM), as well as an ordinary least squares (OLS) model. Direct and indirect impacts are analyzed for a valid interpretation of the spatial statistical models. Factors such as GSI, FSI, traffic volume, traffic speed, road area density, and the fraction of industrial area turn out to have significant impacts on the noise level
Correlation between training gains and threshold improvements across the training sessions.
No full text<p>Tie point 1 (represented by the vertical dotted line) indicates the point at which the threshold of the first session was the same as that of the last session, showing no learning from the training. The error bars indicate the standard error of the mean.</p
