A panoramic stitching vision performance improvement technique for Minimally Invasive Surgery

Abstract

Minimally Invasive Surgery (MIS) is a current prevalent technique for surgical operations. Compared with traditional surgical methods, MIS can reduce the post-surgical recovery time, as well as the costs and pain patients endure as a result of surgery. It is therefore of interest to both doctors and patients. The major problem of MIS is there being a narrow field of vision. We have developed and validated a MIS Panoramic Endoscope (MISPE) [4] to provide doctors with broad fields of view. MISPE features a combination of image overlapping and image stitching. The panoramic image apparatus has two side-by-side endoscopic lenses that provide wide-angle inputs for image stitching. MISPE can provide doctors with panoramic images, so that doctors can easily discriminate the organ's positions between surgical operations. The two issues of concern of MISPE are the quality and speed of image stitching. However, the image stitching results of MISPE cannot be obtained by using the conventional SIFT[1], SURF[2], or ORB[3] algorithms. Therefore, this paper proposes few novel techniques to improve the image stitching performance of MISPE. Experimental results show that MISPE can enhance the image size by up to 150%. We also obtained 2X performance improvement in comparison with the conventional techniques.