Application of scale-space descriptors for the reliable detection of keypoints for image registration in optical...

Data acquired using Optical Mapping (OM) studies are affected by motion artifacts due to the inherent contraction of the heart. Those artifacts can be reduced by registering the images obtained by the OM system or by the combination of approaches like physical restraint of the heart or ratiometry with image registration. Due to the lack of high contrast features most registration methods are not suitable for this application. This paper is focused on the utilization of scale space theory and local descriptors to enhance the detection of local features in OM images and to describe the movement of keypoints. This information can be used to determine a suitable set of transformations to perform the registration process.