Real-Time Tracking by Double Templates Matching Based on Timed Motion History Image with HSV Feature

Algorithm 1

tMHI-HSV-based DTM tracking.

(1)     Input: Video frames stream 𝑓₁,𝑓₂,…,𝑓_𝑡,…

(2)     Output: Object center position 𝑃_𝑡 and rectangle

     silhouette 𝐶^Best_𝑡.

(3)     for each 𝑓_𝑖, 𝑖=1,2,…,𝑡,…   do

(4)   if  𝑖=1

(5)   Initialize the 𝑓₁ to establish the offline template

  𝑇_Offline, the online template 𝑇_Online and the algorithm

      parameter, such as 𝜎, 𝛿, Δ_MinSize, 𝜃 and so forth.

(6)      else

(7)      Segment the moving object by using tMHI

(8)      Get the candidate object patches 𝐶^𝑗_𝑡 (𝑗=1,2,…,𝑚)

(9)      for each 𝐶^𝑗_𝑡, 𝑗=1,2,…,𝑚  do

(10)      Calculate the HSV color histograms 𝐻^𝑗_𝑡 of 𝐶^𝑗_𝑡.

(11)      Calculate the best candidate object 𝑐_Best and 𝑑_Off-On

(12)      Update the online template, 𝑇_Online ← 𝑐_Best

(13)     Update the offline template, 𝑇_Offline←𝑇_Online if 𝑑_Off-On>𝜃

(14)      Output the center position 𝑃_𝑡 and rectangle

      silhouette 𝐶^Best_𝑡 from the best candidate patch 𝑐_Best

(15)        end for

(16)      end if

(17)   end for