A Very Fast and Robust Method for Refinement of Putative Matches of Features in MIS Images for Robotic-Assisted Surgery

Robotic-assisted minimally invasive surgery (MIS) has a very important place in the landscape of modern surgical practices. Simultaneous localization and mapping (SLAM), 3D visualization, augmented reality, image registration and mosaicking are some of the image processing operations, which are often feature-based, used in robotic-assisted surgery. Feature matching refinement (FMR) is a crucial task in such operations. FMR is more critical, in cases where the percentage of true matches is very low, which is generally the case for MIS images. Since real-time is a requisite of MIS tasks, an FMR scheme must be very fast. In this paper we propose a very fast and accurate FMR scheme. The main idea used in developing the proposed scheme is on deciding the size of a local neighborhood and on devising a mechanism for checking feature topology preservation in the local neighborhood. To assess the effectiveness of the proposed scheme, we compare its performance with that of several state-of-the-art methods on different MIS image datasets, which shows its superiority in terms of both the processing time and performance.