A background reconstruction algorithm based on intensity extremum classification

作者：Xiao Mei^*; Zhang Lei; Kou Wenyu; Miao Yonglu; Liu Wei

来源：Advances in Information Sciences and Service Sciences, 2012, 4(13): 337-345.

DOI：10.4156/AISS.vol4.issue13.42

• Background reconstruction
• Cluster-based
• Image data
• Intensity range
• Moving-object detection
• Multi-modal
• Storage spaces

To counter the defects of reconstructing the background intensity with concrete value, which is always in a range of fluctuation, a new background reconstruction algorithm based on intensity extremum classification has been proposed in the paper. The image data is divided into many intensity range classes which are composed of single extreme or double extreme by intensity extremum classification, that is, find the distance between nearest range cluster and new data, and then classified new data into the nearest range cluster or created a new cluster based on the distance and threshold. The appearance weights of each intensity range cluster will be calculated. According to the appearance weight, the intensity range cluster of scene will be chose. The algorithm can correctly build a multi-modal background of the scene. This method can be used in real-time system because it has the characteristics of less storage space and fast speed.

• 出版日期2012