Structural similarity-based ranking of stereo algorithms for dynamic adaptation in real-time robot navigation

Real-time navigation of mobile robot in an unconstrained environment is a challenging area of research and it has several applications such as autonomous systems, military applications, security systems, etc. State-of-the-art stereo vision-based robot navigation system uses a specific stereo matching algorithm during complete navigation, which fails to handle radiometric variations such as change in illumination and exposure. The available illumination and exposure invariant stereo algorithms are computationally expensive and may not be suitable for real-time applications. This research ranked the existing correlation-based stereo algorithms for different real-time conditions by analysing its quality of the disparity using range structural similarity index measure. The proposed evaluation ranking table can be used to generate better disparity at each instance during the robot navigation.