Vision is a set of processes that make sense of information received through the visual system. But to have vision also means to be visionary, to have clear, sharp and bright imagination of the future. To imagine what does not yet exist, but might exist someday. And to envision future advances in technology.
The functioning of a machine vision camera can easily be compared to workings of an eye in the human vision system. Both cameras and eyes focus the light from external objects in the field of view onto a light-sensitive medium. In the case of the camera, the medium is an electronic sensor, while in the case of the eye, it is an array of visual receptors. The camera and the eye both work as a transducer, converting light waves into electric signals.
But what we see with our eyes is not simply a translation of the image on the retina. Human visual perception is the ability to interpret information and surroundings from the effects of visible light reaching the eye – our sight is a psychological manifestation of the visual information in our brain’s cortex. And the same is true for cameras. After the eye’s lens and its retina and after the camera’s objective and its sensor, there must be a vision system. Our eyes and cameras are useless without a vision system. The vision system interprets the information from visible light to build a perception of the surrounding world. The quality of the visual system depends on the sharpness, brightness, contrasts and colours of the images and videos. Its quality depends on the quality of vision.
The human visual system accomplishes a number of complex tasks in a single 'box', including the reception of light and the formation of representations, the identification and categorization of visual objects, assessing distances to and between objects and guiding movements in relation to visual objects. And a high-speed smart camera does it, too. All of this in a single box!
Various physiological components involved in the human visual system are the focus of much research in physiology, psychology, cognitive science, neuroscience, and molecular biology.
The major breakthrough in the field of human visual system happened in the 1970s, when David Marr, a British neuroscientist and psychologist, developed a multi-level theory of human vision. The theory described the process of vision on different levels of abstraction. In order to focus on the understanding of specific problems in vision, he identified three levels of analysis that happen in our visual systems. The processing level addresses, at a high level of abstraction, the problems that the visual system must overcome. The algorithmic level attempts to identify the strategy that may be used to solve these problems. The implementation level attempts to explain how these problems are overcome in terms of the actual neural activity necessary.
In 2006, a study at University of Pennsylvania calculated the approximate bandwidth of human retinas to be about 8960 kilobits per second.
In 2007 Qasim Zaidi, professor of vision science from State University of New York and his college-researchers on both sides of the Atlantic discovered that there are two pathways for sight in the retina and peak spectral sensitivity was measured at 481 nm.
All observations on human visual perception and their findings are the main source for the development of machine vision. Special hardware structures and software algorithms provide machines with the capability of interpreting the images coming from a camera’s sensor on the same principals as human vision works. Artificial visual perception as the result of development on this basis is nowadays increasingly used in industrial automation, non-industrial automated processes and even in our everyday life.
The basics of machine vision science as a discipline is a conceptual model describing all of the factors that must be considered when developing a system for creating visual renderings, images, for creating vision system.
Machine vision designers should take as an example the psychophysical processes that take place in human brains as they make sense of information received through the human visual system.
When developing machine vision system, which is a digital vision system, machine vision designers must consider the observables associated with the subjects which will be imaged. These observables generally take the form of emitted or reflected energy, such as electromagnetic energy or mechanical energy.
Once the observables associated with the subject are characterized, machine vision designers can identify and integrate the technologies needed to capture those observables. In digital cameras, these technologies include optics for collecting energy in an appropriate band of the electromagnetic spectrum, and electronic detectors for converting the electromagnetic energy into an electrical signal.
For all digital vision systems, the electrical signals produced by the capture device must be manipulated by an algorithm that elaborates signals in such a way that they can be displayed as an image. This manipulation is called image processing. Different and multiple processors may be involved in image processing for the creation of a digital image, but output can also be only a set of characteristics, or parameters related to the image.