Embedded stereo disparity computation

CogniVue is a leading innovator in embedded computer vision, providing both hardware and software solutions that enable its customers to develop high-performance, low-power embedded vision applications. From the success of recent 3D sensing technologies such as the Microsoft Kinect or the Intel Creative camera, the need for innovative low-cost depth sensors is rapidly growing. In the range of 3D sensing technologies that are available stereo vision offers unique value propositions in terms of range, flexibility to lighting conditions, cost and power. 3D vision is required to develop new applications such as more accurate and robust gesture and face recognition, augmented reality gaming and advertising, pedestrian detection and other automotive safety functions, smart TV/Smart STB.The objective of this project is to study the performance of different stereo disparity algorithms in the context of a parallel embedded architecture. The intern will identify a number of approaches and assess their potential for parallelization. Performance will be measured in terms of accuracy, disparity density and speed.

Faculty Supervisor:

Robert Laganiere

Student:

Ramin Azarmehr

Partner:

CogniVue

Discipline:

Interactive arts and technology

Sector:

Digital media

University:

University of Ottawa