An Efficient Augmented Reality for Medical Application based on Peicp-Odcnn Spatial Mapping

Abstract

Nowadays, augmented reality (AR) is getting popular as AR can provide visual guidance to users in real environment. It is interactive in nature so many medical fields started to prefer it for provoking better performance in diagnosing as well as detecting disease.But it is not that easy to perform due to the challenges it faces for mapping the virtual and real environment.The proposed framework has developed an efficient AR in a medical application for better performance in detecting and diagnosing a disease. The work mainly concentrates to map the virtual 3D medical image on the real object in varied positions without any preregistration or markers. Initially, 2D to 3D conversion is carried out which gives highly qualified image data without any loss. The conversion is then followed with segmentation which segments the 3d medical image region by referring to the environmental data and avoids the horizontal or vertical artifacts using IMR-CNN. To get the rid of relevant features as well as accurate corner point detection and local point detections of the image the work has developed an RJ-FAST BRISK. Finally, based on the features, the spatial mapping is performed by using PEICP-ODCNN, which reduces the error rate caused by varied poses and also reduces the computation time. Experimental results show that the work obtains a mapping accuracy of 92.24% as compared to the existing state-of-the-art method.