2018-01-17|

A team of postgraduate students from the University of Macau (UM) won a first prize at the 14th National Postgraduate Mathematic Contest in Modelling. This year’s contest attracted 11,834 teams formed by 35,502 postgraduate students from mainland China, the United States, the United Kingdom, and Singapore. A team from UM also won a second prize at last year’s contest.

The UM team was formed by PhD student Li Tianjiu, Guo Li, and Feng Qiying. Led by Associate Professor Chen Long from the Department of Computer and Information Science, the team chose to tackle the problem of foreground object extraction based on surveillance videos.

As an integral part of a smart city, vision surveillance systems are widely used in public facilities such as schools, airports, and hospitals. Every day, massive amounts of data are generated by these surveillance systems, which poses a challenging task for analysers. For example, foreground objects like humans are of the greatest interest to analysers; so the ability to extract the foreground objects automatically will be a great benefit. The UM team studied this problem in the contest. By applying the matrix factorisation methods, the team developed a good solution by separating the background and foreground objects in a video. This solution produced excellent results with the test data provided by the contest organiser and won the praise of the judges.

澳門大學科技學院博士研究生李天鈞、郭莉和馮綺穎組隊,參與“第十四屆全國研究生數學建模競賽",與來自國內、美國、英國、新加坡等大學共11,834支隊伍、35,502名研究生對壘,最終獲得全國一等獎,繼上屆獲全國二等獎再創佳績。

是次澳大參賽隊伍由澳大科技學院副教授陳龍指導,在大賽中挑戰“基於監控視頻的前景目標提取"問題。在智慧城市系統中,監控系統是不可或缺的部分。該系統被廣泛佈置在校園、醫院、機場等公共設施,因而每日產生了海量的監控視頻資料。對於視頻資料的分析人員來說,最感興趣的是視頻中的前景目標,如人、車等物體,但由於背景的不穩定性和複雜性,自動提取視頻中的前景目標是個困難的問題。澳大參賽隊伍利用矩陣分解方法,設計了將視頻的背景和前景目標自動分離的解決方案,提供的測試資料取得了極佳的表現,獲得了評審青睞。

全國研究生數學建模競賽是內地目前層次最高、規模最大、影響最廣泛的全國性研究生學科競賽。近年受到國內外大學的廣泛關注,參賽學生逐年增加,賽題難度越來越大,獲獎難度亦增大。

The winning students and Prof Chen Long (1st from left)
得獎學生與陳龍(左一)