Revision history [back]

HI All,

I am currently working on motion detection . I have done two different approaches

1. frame Difference using cvAbsdiff(prev,curr,diff) and then drawing the contours on the images and extracting the motion frames based on the contourArea. It is was not good that when small motion contours are not represented properly even when there is some motion.

2. Background subtraction using BackgroundSubtractorMOG2 I'm following http://answers.opencv.org/question/15988/how-to-do-background-substraction/ this approach .

My Issue is : 1.How to Identify the Blobs using OpenCV .I have tried using cvFindContours(Image, storage, contours, Loader.sizeof(CvContour.class),CV_RETR_EXTERNAL, CV_CHAIN_APPROX_NONE);

it was drawing contours only on the parts of the image and not on the entire images.

Should I need to Do this by using HOG or any other classifiers or by any method.

I have even gone through the optical flow approach for motion detection.

Which among the three approaches the motion detection(ImgDiff,optical flow and backgroundsubtraction) can be achieved properly and suggest me with some tutorials to work with them.

Any help is appreciated.

Thanks.

HI All,

I am currently working on motion detection . I have done two different approaches

1. frame Difference using cvAbsdiff(prev,curr,diff) and then drawing the contours on the images and extracting the motion frames based on the contourArea. It is was not good that when small motion contours are not represented properly even when there is some motion.

2. Background subtraction using BackgroundSubtractorMOG2 I'm following http://answers.opencv.org/question/15988/how-to-do-background-substraction/ this approach .

My Issue is : 1.How to Identify the Blobs using OpenCV .I have tried using cvFindContours(Image, storage, contours, Loader.sizeof(CvContour.class),CV_RETR_EXTERNAL, CV_CHAIN_APPROX_NONE);

it was drawing contours only on the parts of the image and not on the entire images.

Should I need to Do this by using HOG or any other classifiers or by any method.

I have even gone through the optical flow approach for motion detection.

Which among the three approaches the motion detection(ImgDiff,optical flow and backgroundsubtraction) can be achieved properly and suggest me with some tutorials to work with them.

Any help is appreciated.

Thanks.