Frame differencing for motion detection/moving object in image

Hello Community!

I am doing a frame differency of a camera image where I substract the delayed frame from the current frame with thresholding followed by closing and blob analysis. The image is a binary picture (live image which gets processed by sobel filter in simulink). Blob Analysis config is: number of blobs 2. No variable size but fill value. No image borders. I select the first row of the bbox output and get the centroid of my moving object.

While it detects my movable object I have problem I cannot conclude to solve:

While moving my object it gets detected, but when I put another object into the field of view, which was not in the scene before it gets detected and stops the detection of the dynamic object for this sample. While it being static and the dynamic object is still moving, the detection goes right back to the dynamic one.

Is there a way to ignore new upcoming objects?

Best regards

Answer 1

[vaibhav mishra]

Hi there,

i would suggest to use viola jones object detection framework+ optical flow to track the motion of any object of interest. this will surely give good results.

Answer 2

[Asmita]

frame differencing