Algorithm of moving target detection based on multiply connected domain mergence

Some surveillance videos have the drawbacks of low light, low resolution and low contrast and it is difficult to obtain the better detect results using the traditional target detection algorithm. Aiming at this problem, an algorithm of multiply connected domain mergence based on the features of spatial distribution has been presented. The Vibe background modeling algorithm is utilized to achieve the primary segmentation and then the spatial relationships of the connected domains are analyzed and the different connected domains which belong to one target are merged. Experimental results show that the holes and splits produced by the Vibe moving target detecting algorithm have been corrected and the more accurate detect results have be obtained.