I am plannig to do a sort of gesture or movement controlled device using a webcam and NI machine Vision for my FYP,
The idea is to use the webcam to continuously capture images or video and process them in real time in order to detect predefined movements, and then send signal based on the processing results,
I want some help regarding the video processing part, I have experience working with Image processing using labVIEW for object detection, OCR and quality control, but i dont have a clear idea about how i can apply this on a video.
Can anybody help me with ideas about possible solutions to detect and differentiate different movements using LabVIEW??
Thanks for your help,
Luckily for you, the NI Vision Development Module 2010 has some new handy dandy features that might suit your needs.
This new release apparently includes optical flow (or motion estimation). It can track blobs around your image and give you an idea of movement... may be worth checking into!
Thanks Matt, but I am still using LabVIEW 2009, I should have mentioned that in the detailing and it doesn't contain any similar algorithms.
I am also working on somethng very similar. I am able to
movement detection using LV2009. Thast true that 2009
doesnot have optical flow. I am sure we can co-ordinate
on this project and have a better solution. I am still working
on this product to detect objects orobject movement.My e-mail
Can you guys keep me posted too? I'm definitely interested in this field, and would love to help where I could. There is also definitely a wealth of knowledge on these boards, I know I lean on them every now and again when I'm having trouble with something.
You can reach me at
I was able to detect movement based on a simple algorithm.
I buffered the images and subtracted the old image from the
new image and multplied it with constant gain, and if difference
is greater than threshold we are to able to detect motion. What
algorithm are you using for motion detection.
Also you donot need to have video processing. Use image processing
with minimum delay and it shoudl fairly fast detection, atleast thats what
If backgrounf is always going to be same then you could also get a
constant background image and subtract that from ned acquired image
and you get movement detection. But this has its limitations.
I tried the IMAQ subtract and indeed it can detect motion, what i am trying to do is to find a way to know the type of movement that have occured..
I think I'm going to train the program on the predefined moves and then when i subtract the images in the buffer from each other I'll try to compare the result to the results I've already trained using maximum likelihood or something like that..
I posted the question to see if there are better techniques that do not require much training..
Hopefully you'll find this useful in your project. If you have better ideas keep me posted...
Nice to meet ur guy. I am now doing the color detection on fingers. And i successful to detect the color on my fingers but i unable to make it always allocate the color while i move my hand. Anyone of u have any idea about it? Seem Falcon's FYP thing is almost same with my 1. Would u mind to share it with me. I apologize for my lousy English. If possible let me knew as soon as possible.
Thank You And Best Regard
QY Goh (^.^)