A top group multiobjective pig foreground detection method in a complex background

The present invention provides a top view of multi-target population of pig foreground detection method in a complex background, pig video sequence group acquired first planar view; then set the "effective area"; Gaussian mixture model-based prediction mechanism proposed foreground detection algorithm to obtain a foreground object; while taking advantage of the color information of the foreground object segmentation to do the maximum entropy threshold value, to obtain another prospect objective; the final results of the two algorithms and mathematical morphology fusion process to obtain the final target prospects. Background of the invention need not initially acquired in advance, for the presence of background light changes, the ground urine stains, stains and fecal matter and other debris interference; stop and go motion pattern an individual pig; foreground foreground object color and diverse other individual pig Complex Background Target Detection. To further explore the pig population of individual identification, behavior analysis laid the foundation.