Computer vision differs from human vision in several ways. Humans can focus on basic features of objects and identify them even under different conditions, such as low light or slight changes in color and shape. In contrast, computer vision focuses on details rather than basic features, relying on a set of rules to detect objects. This can lead to mistakes when slight environmental or object modifications occur, such as changes in lighting. Another key difference is the ability to recognize objects in three dimensions. Humans can perceive depth, while computer vision typically operates in two dimensions, meaning that slight tilting of objects can cause confusion.
Various techniques can help computers see and track objects more effectively, similar to how humans do. One such technique is frame differentiation, which is useful for detecting motion. This is done by comparing consecutive frames, where differences in pixel color indicate movement. Another technique is background subtraction, where the computer is provided with a reference image of the background. When an object is introduced into the scene, the computer detects pixels that differ from the background and identifies the object. A third method is comparing pixels with a threshold value, which is especially useful when there are significant light differences between the background and the object. Object tracking can also be achieved by tracking the brightest pixel in a video frame. Each pixel’s brightness is compared to the brightest encountered, and its location is stored. This technique can be adapted to track the darkest pixel or multiple objects of different colors.
In interactive art, the complexity of implementing certain ideas limits artistic expression, as only a few people have the expertise to implement such designs. However, with ongoing advancements making computer vision techniques more accessible and easier to use, art will increasingly benefit from these technologies.