opencv - Fastest moving object recognition and tracking in Android -

i working on augment reality game, need recognize , track fast moving object. have tried following image processing libraries,

1. opencv

2. boofcv

3. fastcv

i have tried tld algorithm track object, tracking successful performance needed improved. if object moving faster result takes time, because of processing time taken algorithms. have tried circulant,mean-shift algorithms boofcv.

check these demos :

opentld using fastcv

boofcv demonstration

object tracking in these 2 demos seems calculation takes time.

can go following scenario faster,

1. extract r,g,b matrix of object tracked

2. take camera frames , convert r,g,b matrix , search tracked object matrix in camera frame.

is there better way ??

i suggest using gray scale instead of rgb, done in image processing way computation reduced 1 matrix instead of 3.

if need check color, use rgb when needed not through out whole computation.

tracking fast moving objects difficult. try using camera can take more frames per second although need process more images , suppose on mobile device

also can reduce image size being processed smaller window based on previous object position can estimate next position , limit vecinity , process bits of image. in brief perform optical flow in section of image (use gray scale).