科研成果

General object recognition and image understanding is recognized as a dramatic goal for computer vision and multimedia retrieval. In spite of the great efforts devoted in the last two decades, it still remains an open problem. In this paper, we propose a selective attention-driven model for general image understanding, named GORIUM (general object recognition and image understanding model). The key idea of our model is to discover recurring visual objects by selective attention modeling and pairwise local invariant features matching on a large image set in an unsupervised manner. Towards this end, it can be formulated as a four-layer bottomup model, i.e., salient region detection, object segmentation, automatic object discovering and visual dictionary construction. By exploiting multi-task learning methods to model visual saliency simultaneously with the bottom-up and top-down factors, the lowest layer can effectively detect salient objects in an image. The second layer exploits a simple yet effective learning approach to generate two complementary maps from several raw saliency maps, which then can be utilized to segment the salient objects precisely from a complex scene. For the third layer, we have also implemented an unsupervised approach to automatically discover general objects from large image set by pairwise matching with local invariant features. Afterwards, visual dictionary construction can be implemented by using many state-of-the-art algorithms and tools available nowadays.

Studies have demonstrated that cis-pinonic acid (CPA) is an important product from the oxidation of pinenes with ozone. CPA has been measured on aerosols and is used as an aging indicator for secondary organic aerosols (SOA). CPA levels and formation in urban aerosols and its annual variability, however, are still poorly understood. Here, we present monthly CPA average concentrations on aerosols in Toronto, Ontario, Canada based on a two-year-period: 2000-2001. They displayed a seasonal pattern associated with temperature and ozone (O(3)) plus nitrogen dioxide (NO(2)) reflecting the influence these have on emissions of pinenes from forests and their atmospheric oxidation, respectively. However, in Toronto some months with higher CPA concentrations, especially in the winter, were inconsistent with the seasonality of temperature or/and O(3) + NO(2) levels. Instead these deviations were associated with increases in wood burning tracers such as dehydroabietic acid (DHAA) and sugars. Similar features were observed during a two-week-period comparing day and nighttime CPA concentrations in the Lower Fraser Valley (LFV) of British Columbia, Canada, in that the CPA concentrations clearly varied diurnally with temperature and O(3) + NO(2) on some days, but also showed a significant correspondence with variations in the wood burning tracer concentrations, such as levoglucosan. These findings demonstrate that CPA formation is strongly impacted by wood burning activity. Crown Copyright (C) 2011 Published by Elsevier Ltd. All rights reserved.

The production of secondary organic aerosol (SOA) by the dark ozonolysis of gas-phase beta-caryophyllene was studied. The experiments were conducted in a continuous-flow environmental chamber for organic particle mass concentrations of 0.5 to 30 mu g m(-3) and with ozone in excess, thereby allowing the study of second-generation particle-phase products under atmospherically relevant conditions. The particle-phase products were characterized by an ultra-performance liquid chromatograph equipped with an electrospray ionization time-of-flight mass spectrometer (UPLC-ESI-ToF-MS). Fragmentation mass spectra were used for the structural elucidation of each product, and the structures were confirmed as consistent with the accurate m/z values of the parent ions. In total, fifteen products were identified. Of these, three are reported for the first time. The structures showed that 9 out of 15 particle-phase products were second generation, including all three of the new products. The relative abundance of the second-generation products was approximately 90% by mass among the 15 observed products. The O:C and H:C elemental ratios of the 15 products ranged from 0.13 to 0.50 and from 1.43 to 1.60, respectively. Fourteen of the products contained 3 to 5 oxygen atoms. A singular product, which was one of the three newly identified ones, had 7 oxygen atoms, including 1 carboxylic group, 2 carbonyl groups, and 3 hydroxyl groups. It was identified as 2, 3-dihydroxy-4-[2-(4-hydroxy-3-oxobutyl)3, 3-dimethylcyclobutyl]-4-oxobutanoic acid (C14H22O7). The estimated saturation vapor pressure of this product is 3.3x10(-13) Pa, making this product a candidate contributor to new particle formation in the atmosphere.