The Research On Detection Of Abandoned Object In Video Surveillance

With the development of society, human own safety problems get more and moreattention. Abandoned package as the main way of terrorist attacks have got increasingconcerns. The original monitoring system has been unable to meet the needs of people.Intelligent video surveillance system is a new type of dangerous behavior testing tools.Abandoned object detection is an important purpose of intelligent video surveillance.Although there are a lot of abandoned object detection algorithms now, these methods arenot very ideal. There are two problems: one is due to improper handling or processing ofinterference is less, the measuring accuracy is not high and often mistakenly identified orleak; another case is too many interference factors were considered, which makes thealgorithm very complicated and difficult to meet the real-time application. In order tosolve above problems, the algorithm is studied and improved in this paper. The main workis as follows:1) This paper further studies and researches the technology of intelligent videosurveillance, analyses the advantages and disadvantages of present method.2) Introduces a kind of abandoned object detection method based on dual foreground,carefully analyses the advantages and disadvantages of this method. On that basis, thispaper proposes a multiple abandoned object detection method which uses recursivemethod.3) This paper designs an abandoned object detection method based on slow Gaussianmodel. The principle of the method is using the Gaussian Mixture Model of low updaterate and low learning rate to judge all foreground objects’ centroid and time individual.Then using recursive method for timers and centroid have order. According to the value ofthe timers to judge abandoned object and start warning device.4) Indoor and outdoor scenes under MATLAB are simulated respectively. Resultsshow that the improved algorithm and new methods of design has better detection effectsand faster processing speed.