Tracking Multiple Points Throughout a Video
29 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Hello! I am trying to track multiple points of interest throughout a video and obtain the positions of each point of interest frame by frame in the video.
I would like to be able to manually select each point of interest (reflective markers, see image) and use MatLab to automatically track those markers throughout the entire video. As I am eventually computing distances between markers, I have figured out the "calibration" from pixels to mm, however, I haven't been able to figure out the best way to track the markers and gain marker position frame by frame.
My original idea was to use a color algorithm to track any silver object, however the muted colors of the lab give me concern with tracking that way. Additionally, although captured at 120fps, we are still getting blurred markers that I am concerned wont be picked up with automatic tracking algorithms (see trailing foot in attached image). I also thought about using something like a face detector to track points, however I wasn’t able to get that to work (although it may have been user error!).
In short, does anyone have any ideas on how to 1) manually select and then automatically track multiple objects across a video, frame by frame, 2) visually verify that markers tracked successfully, 3) apply a Kalman filter to estimate points when obstructed and/or 4) how to clean up video post-collection to remove blur (in theory, my understanding was 120fps should not produce blur at these slow velocities)
Any help on any of these topics would be appreciated!
Thanks!
1 Kommentar
Cris LaPierre
am 8 Jan. 2025
These are big topics. This sounds to me like object detection and tracking using Computer Vision.
MathWorks has recently launched a course on Coursera on these topics. It's free to enroll. If you are interested, I'd suggest taking a look at them.
If you need to build you Computer Vision skills, then consider enrolling in the entire Computer Vision for Engineering and Science Specialization.
Antworten (1)
Gayathri
am 9 Jan. 2025
Computer Vision Toolbox provides video tracking algorithms, such as continuously adaptive mean shift (CAMShift) and Kanade-Lucas-Tomasi (KLT). As @Cris LaPierre mentioned object tracking is a vast domain and there are many different ways of doing this. Please refer to the below link, which give an idea about different functions that can be used for object tracking and motion estimation. You can refer to the below link for documentation of the functions, which also include examples for easy understanding.
In this link you can find information of different functions used for object tracking, for example, "configureKalmanFilter" which creates a Kalman filter for object tracking.
In MATLAB R2024b, a new optical flow estimation method using RAFT deep learning algorithm is available.
Please refer to the below link, which also shows an object tracking method using optical flow.
Hope you find this information helpful!
0 Kommentare
Siehe auch
Kategorien
Mehr zu Tracking and Motion Estimation finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!