Institution
The Chinese University of Hong Kong
Education•Hong Kong, China•
About: The Chinese University of Hong Kong is a education organization based out in Hong Kong, China. It is known for research contribution in the topics: Population & Cancer. The organization has 43411 authors who have published 93672 publications receiving 3066651 citations.
Topics: Population, Cancer, Poison control, Randomized controlled trial, China
Papers published on a yearly basis
Papers
More filters
••
13 Jun 2010TL;DR: This work proposes a pose-adaptive matching method that uses pose-specific classifiers to deal with different pose combinations of the matching face pair, and finds that a simple normalization mechanism after PCA can further improve the discriminative ability of the descriptor.
Abstract: We present a novel approach to address the representation issue and the matching issue in face recognition (verification). Firstly, our approach encodes the micro-structures of the face by a new learning-based encoding method. Unlike many previous manually designed encoding methods (e.g., LBP or SIFT), we use unsupervised learning techniques to learn an encoder from the training examples, which can automatically achieve very good tradeoff between discriminative power and invariance. Then we apply PCA to get a compact face descriptor. We find that a simple normalization mechanism after PCA can further improve the discriminative ability of the descriptor. The resulting face representation, learning-based (LE) descriptor, is compact, highly discriminative, and easy-to-extract. To handle the large pose variation in real-life scenarios, we propose a pose-adaptive matching method that uses pose-specific classifiers to deal with different pose combinations (e.g., frontal v.s. frontal, frontal v.s. left) of the matching face pair. Our approach is comparable with the state-of-the-art methods on the Labeled Face in Wild (LFW) benchmark (we achieved 84.45% recognition rate), while maintaining excellent compactness, simplicity, and generalization ability across different datasets.
470 citations
••
TL;DR: Experimental results on Concordia University CENPARMI database of handwritten Arabic numerals and Yale face database show that recognition rate is far higher than that of the algorithm adopting single feature or the existing fusion algorithm.
469 citations
••
TL;DR: The results indicate that the effect of urban heat island in Hong Kong is mainly located in three sub-urban areas, namely, Kowloon Island, the northern Hong Kong Island and Hong Kong International Airport.
Abstract: In this paper, the effect of urban heat island is analyzed using the Landsat TM data and ASTER data in 2005 as a case study in Hong Kong. Two algorithms were applied to retrieve the land surface temperature (LST) distribution from the Landsat TM and ASTER data. The spatial pattern of LST in the study area is retrieved to characterize their local effects on urban heat island. In addition, the correlation between LST and the normalized difference vegetation index (NDVI), the normalized difference build-up index (NDBI) is analyzed to explore the impacts of the green land and the build-up land on the urban heat island by calculating the ecological evaluation index of sub-urban areas. The results indicate that the effect of urban heat island in Hong Kong is mainly located in three sub-urban areas, namely, Kowloon Island, the northern Hong Kong Island and Hong Kong International Airport. The correlation between LST and NDVI, NDBI also indicates that the negative correlation of LST and NDVI suggests that the green land can weaken the effect on urban heat island, while the positive correlation between LST and NDBI means that the built-up land can strengthen the effect of urban heat island in our case study. Although satellite data (e.g., Landsat TM and ASTER thermal bands data) can be applied to examine the distribution of urban heat islands in places such as Hong Kong, the method still needs to be refined with in situ measurements of LST in future studies.
469 citations
••
TL;DR: It is reasonable to conclude that intracranial atherosclerosis is the most common vascular lesion in stroke patients worldwide, as the majority of the world’s populations are Asians, Africans or Hispanic, and Caucasians remain the only ethic group with a low frequency of intrac Cranial Atherosclerosis.
Abstract: Stroke is a heterogeneous disease with a plethora of differing stroke mechanisms. Small vessel, large artery atherosclerosis and cardio-embolic stroke are the most common subtypes encountered in clinical practice. Different stroke mechanisms may require different treatment. Anticoagulation for atrial fibrillation is a good example. Many recent research and clinical trials successfully targeted individual stroke subtypes rather than indiscriminately grouping all strokes together. For large artery atherosclerosis, carotid stenosis is the most common vascular lesion found in Caucasians in America and Europe and is extensively studied in terms of epidemiology, pathophysiology and treatment. Unfortunately, relatively little is known about intracranial atherosclerosis until recently; now, modern neuroimaging methods permit noninvasive screening of susceptible patients. Intracranial atherosclerosis affects the middle cerebral artery, intracranial portion of the internal carotid artery, vertebrobasilar artery, posterior, and anterior cerebral arteries. Examples of intracranial stenosis are shown in Figs 1 and 2. Transcranial Doppler ultrasound, CT angiography and MR angiography are now routinely available in clinical practice (1–4). Although digital subtraction angiography remains the gold standard of diagnosis of intracranial atherosclerosis, these noninvasive tests have been validated against clinical outcomes and events (5–7). For decades, it is well described that patients of Asian, African, and Hispanic ancestry were at higher risk of intracranial atherosclerosis (8). More recent studies on consecutive patients confirm this finding and pinpoint the frequency of intracranial atherosclerosis. In Chinese populations, intracranial atherosclerosis accounts for about 33–50% of stroke and 450% of TIA (6, 9, 10). In Thailand, intracranial atherosclerosis was found in 47% of stroke patients (11). In Koreans, 56 3% of stroke patients had intracranial atherosclerosis although the authors used 30% stenosis as cutoff (12). In Singapore, significant intracranial stenosis was found in 47 9% of stroke patients (13). In Japan, the frequency of intracranial atherosclerosis remains high despite increasing frequency of extracranial carotid stenosis (14). In North America, extracranial carotid stenosis remains the most common vascular lesion in Caucasian stroke patients. However, when compared with Caucasians, the relative rate of intracranial atherosclerotic stroke was 5 00 for Hispanics and 5 85 for blacks (15). Based on the widespread observation worldwide that intracranial atherosclerosis is the most common vascular lesions in Asians, Hispanics and Africans, and Caucasians remain the only ethic group with a low frequency of intracranial atherosclerosis. As the majority of the world’s populations are Asians, Africans or Hispanic, it is reasonable to conclude that intracranial atherosclerosis is the most common vascular lesion in stroke patients worldwide. Not only the number of patients with intracranial atherosclerosis is staggering but also patients with intracranial disease are at a high risk of recurrence of up to 25–30% in 2 years after stroke (5, 7, 16), further magnifying the burden of intracranial disease. Unfortunately, there has been
469 citations
••
10 Apr 2017TL;DR: In this article, a sub-pixel motion compensation (SPMC) layer is proposed to fuse multiple frames to reveal image details, which can generate visually and quantitatively high quality results without the need of parameter tuning.
Abstract: Previous CNN-based video super-resolution approaches need to align multiple frames to the reference. In this paper, we show that proper frame alignment and motion compensation is crucial for achieving high quality results. We accordingly propose a “sub-pixel motion compensation” (SPMC) layer in a CNN framework. Analysis and experiments show the suitability of this layer in video SR. The final end-to-end, scalable CNN framework effectively incorporates the SPMC layer and fuses multiple frames to reveal image details. Our implementation can generate visually and quantitatively high-quality results, superior to current state-of-the-arts, without the need of parameter tuning.
469 citations
Authors
Showing all 43993 results
Name | H-index | Papers | Citations |
---|---|---|---|
Michael Marmot | 193 | 1147 | 170338 |
Jing Wang | 184 | 4046 | 202769 |
Jiaguo Yu | 178 | 730 | 113300 |
Yang Yang | 171 | 2644 | 153049 |
Mark Gerstein | 168 | 751 | 149578 |
Gang Chen | 167 | 3372 | 149819 |
Jun Wang | 166 | 1093 | 141621 |
Jean Louis Vincent | 161 | 1667 | 163721 |
Wei Zheng | 151 | 1929 | 120209 |
Rui Zhang | 151 | 2625 | 107917 |
Ben Zhong Tang | 149 | 2007 | 116294 |
Kypros H. Nicolaides | 147 | 1302 | 87091 |
Thomas S. Huang | 146 | 1299 | 101564 |
Galen D. Stucky | 144 | 958 | 101796 |
Joseph J.Y. Sung | 142 | 1240 | 92035 |