Registration and Fusion of EVS and SVS Runway Images for Embedded Systems | Consortium for Embedded Systems | SIU

Abstract: Safely landing an aircraft on a runway in all weather conditions requires fusing information from various sensors. The goal of this project is to develop algorithms to accurately register runways in EVS and SVS images and fuse the registered images so that the information from both images are displayed simultaneously. The registration of runway images will be based on detecting corresponding features in both images. The features will be extracted from the Hough transform of the images and will typically consist of the four corner points of the runway quadrilaterals. Image fusion will be based on combing the discrete wavelet transform (DWT) coefficients of the images. Several combinations rules such as the maximum, average, and a combination of the maximum and averaging of DWT coefficients will be evaluated. The final goal will be to improve the computational efficiency by embedding the algorithms into a multi-core processing architecture.

Problem: This project will focus on developing a sequence of novel algorithms to safely land aircrafts by registering the runways in EVS and SVS images and fusing the registered images so that the information from both images can be displayed optimally.

Rationale / Approach: The precise detection of runways is crucial for safely landing manned and unmanned aircrafts. The challenge is to be able to detect the runway accurately in all possible weather conditions. A novel approach which involves registering and fusing EVS and SVS images is proposed. A successful completion of this project will have a major impact on improving landing safety.

Novelty: The project will focus on problems that have been uniquely identified by Rockwell Collins. It is expected that the techniques developed in this project will not only lead to the development of novel landing heads-up display system but will also be applicable to a vast range of problems involving image registration and image fusion.

Potential Member Company Benefits: Rockwell Collins will define the specific registration and fusion problems of interest to them and SIUC will provide them with the technical background and algorithms to solve the problems.

Deliverables for the proposed year: Image registration and image fusion algorithms.

Milestones for the proposed year: Develop algorithms to:
(a) Register EVS and SVS runways in a sequence of images in varying weather conditions.
(b) Fuse registered EVS and SVS images in the DWT domain.
(c) Evaluate the various registration and fusion algorithms on data provided by Rockwell Collins.