Learn something new every day
More Info... by email
Blob tracking is a method by which computers can identify and trace the movements of objects within images. A blob is a group of pixels the computer identifies as an object. A person, for example, could be a blob. This tracking method allows the computer to find the blob’s position in successive frames. In effect, the computer sees the motion of the object, an ability which has applications in computerized image analysis and in the development of computer vision.
Before a computer can track blobs, it must define them. The first step in blob tracking is creating a program that detects blobs in an image. The computer finds blobs by grouping pixels with similar light values or color values together. In the real world, every surface has subtle variations, so if the computer selected only one light or color value, a blob might be only a few pixels, which is useless when trying to group images into useful components that represent complete units.
It is the responsibility of the programmer to create a blob detection system that mimics the distinctions between objects that the human eye can see. She does this by writing a tolerance threshold into the program, which tells the computer the maximum difference between values that may occur in a blob. Programmers test and refine blob detection programs until they can identify portions of images correctly.
Next, the programmer must create a way for the computer to track the movement of the blobs. This is difficult because the computer cannot simply scan the next frame for the same blob: the shape and values of a blob can change as it moves around. For example, a person who is sitting down could stand up. The computer must detect the blobs in the new image and make meaningful connections between the seemingly different blobs present in each frame. Programmers create equations that define the relative importance of factors including location, size and color, and the results help it to decide if the blob in the new frame is similar enough to the previous blob to receive the same label.
Blob tracking is a powerful tool, especially when combined with other methods of analyzing images. For example, computers identify areas of high contrast between one pixel and the next to detect edges. Some programs use ridge detection methods to determine the contour of a surface.
These abilities allow computers to conduct sophisticated image analysis. An automated system can scan more images than human reviewers could feasibly look at. Using its toolkit, including blob tracking, it could identify images of interest that require more detailed review and forward those sections to a human.
Another application of blob tracking is in computer vision and robotics. As techniques like this one become more refined, robots will be able to glean meaning from images in a way that approximates image processing in the human brain. Computers could take in information without having it entered manually, gaining more information than ever before. Robots could see the world around them.