Last Friday, Eric Fossum, a professor at the Thayer School of Engineering, gave a presentation addressing both recent advances in digital imaging research and its ethical implications. Specifically, Fossum considered automating pixel capture and how this technology could serve to reduce privacy and encourage image-centric communications.
While cameras have only been in existence for 200 years and digital cameras just the last 35, mankind has “a long history of capturing and sharing images,” according to Fossum, displaying images from cave drawings to Michelangelo. As a result, image capturing technology has rapidly become integrated into our daily lives, ranging in function from social cell phone photography to medicinal applications in endoscopic procedures and rear-view cameras in modern cars.
While technologies like cell phone cameras and flip-cams have revolutionized our interactions, the world is experiencing a “visual information overload.” People often feel helpless watching live footage of disasters such as the 2011 Japanese tsunami because there is nothing they can do to help in the short term. Another serious concern is loss of privacy, as networked security cameras and facial recognition software become capable of recording and storing a person’s entire life.
Photons, which are particles that make up light, enter a camera at high rates (approximately 1 x 1015 photons/cm2/sec), and are focused onto an image sensor where they are read out. Although this process creates a viewable image, photon shot noise and diffraction limit its quality. Photon shot noise occurs when a large number of photons are accepted by an individual pixel which negatively impacts the image quality and color. Diffraction limits occur when light enters a lens and diffracts to create a “blur zone” which can be up to 250 percent larger than the pixel itself.
Fossum’s research eliminates these problems with what he calls an “active pixel.” Instead of determining the amount of photons to create an image by reading groups of pixels, this technology gives each individual pixel the means to take its own photon measurement and process that reading. This system, combined with another project called “quanta image sensing,” which uses a super–tiny pixel called a “jot” to obtain and average photon distributions over time, can greatly improve digital imaging by eliminating both diffraction limits and photon shot noise.
In the future, digital imaging research could lead to the creation of a 3D television that that would combine both depth and color or the development of even more advanced and miniaturized image capturing devices, among many other technologies.