This article was originally published on www.display-central.com.
SID Display Week is just 12 days away and counting (May 19) offering some exceptional opportunities to expand ones’ knowledge well beyond the pixel depth and luminosity of a particular display and delve into new emerging topics that promise to shape the face of displays for years to come. Case in point, the Sunday Short courses, a four-hour deep dive with topics this year ranging from OLED fundamentals, a TFT backplane review, and fundamentals of Touch Technology.
But one topic that caught our eye was the Fundamentals of 3D Computer Vision, and Applications to Interactive Displays, given by Intel’s Dr Achin Bhowmik Director of Perceptual computing for the chip giant Intel. His talk gets into the nuts and bolts, the how’s and why’s of “natural human computer interaction” that promise to expand the scope with “intuitive interfaces immersive applications and new user experiences,” and calling it “…potentially the next big opportunity for the display and consumer electronics ecosystem,” Bhowmik said. Truly interactive displays need human-like senses, and perceptual interfaces, Bhowmik writes. This is evidenced by real-time 3D computer vision technologies.
From Upcoming SID Short Course on 3D Visual Perception given by Dr. Achin Bhowmik, Director of Perceptual Computing at Intel According to Bhowmik, we need to add 3D vision capability for life-like interactions. He sees the disciplines of computer graphics and computer vision at two ends of the same continuum. Computer Graphics produces image data from 3D models while Computer Vision “often” produces 3D models from image data, with Augmented reality somewhere in-between combining the two disciplines.
The course explores key elements of vision including visual perception and the human visual system (seeing vs. perceiving). Bhowmik then delves into Image formation and capture including both 2D and 3D techniques that look at four cases including single and stereo view plus 3D imaging with structured light and time of flight. He next turns to the algorithms dealing with inference and recognition (the math) and leading-edge techniques that include the importance of edge detection and why. Bhowmik shows the calculus that helps identify the edge by displaying the image as an “image intensity function” and characterizing (finding) the edge as the place of rapid change along the horizontal scan line (first derivative) citing the work of John Carry at the MIT A.I. Lab, calling it “…the most widely used edge detector in computer vision today.”
Other techniques covered in detail include image elements like “corners,” “blobs” and “extraction features,” all used in a variety of display vision key application areas like image blending (stitching) and face detection and recognition. The latter two also uses another method called principle component analysis, that breaks down the face into “standardized face ingredients” to create “Eigenfaces” where any human face is a combination of these standard faces. The idea grew out of the Eigenvectors used in early computer vision problems in the late 1980s that uses specific thresholds for recognition.
Needless to say, thanks to synthetic training data and other revolutionary computer vision developments, these 3D computer vision techniques will continue to enable these immersive experiences, and adding other human-like senses beyond sight (eyes) to include ears, voice, and touch will take us further down that same path. To help get there, Intel’s now offering the industry a perceptual computing SDK (software development kit) to explore these technologies in new niche application areas with cool interfaces like speech, facial analysis, hand tracking 3D gestures and object tracking. All on the cusp of our future human/computer interface with the display playing a central role. So rain or shine, Display Week is coming to Vancouver, BC with a kick-off Sunday host of short courses that offer an unparalleled opportunity to dive deep with the experts in some hot display topic areas. Hope you can make it. – Steve Sechrist