Tracking Objects Based on Multiple Particle Filters for Multipart Combined Moving Directions Information

Algorithm 3

(MultiPart).

	Step 1: Initialize N-particles set
	Step 2:
	for i = 1 to n do
	Beginfor
	Take patern for part i according to Algorithm 4.
	Train strong classification according to Algorithm 5.
	Endfor
	Step 3:
	while The video is not over do
	Beginwhile
	Get observation photos obs
	for i = 1 to n do
	Beginfor
	Use particle filter to estimate the ith state according to Algorithm 1 or Algorithm 2
	Measure the distance:
	Endfor
	Choose the best part
	Using center part and structure H perform rotation of the remaining n − 1 centers.
	Tranlate n − 1 other particle set to n − 1 has just been centered on .
	Scale the particle sets according to the structure ratio H.
	Translating n − 1 set of other particles toward the part .
	for i = 1 to n do
	Beginfor
	Take a new sample based on the new section rotated from .
	Update strong classification according to Algorithm 6
	Endfor
	Endwhile