Disclosed are an endoscopic surgery navigation method and system. The method comprises: reading multi-modal medical image data (101) using any image data in the multi-modal medical image data as a reference image, and using other image data as an image to be registered, so as to perform image full affine matching (102) performing scene reconstruction and hybrid rendering on the image data after the image full affine matching, and obtaining a virtual scene (103) completing a registration between a CT image and a pose of a patient by means of a convex hull fast registration method performing fast calibration by means of a surface point cloud subjected to convex hull optimization tracing an endoscope and surgical tools, and obtaining a pose relationship between the endoscope and surgical tools and the body of the patient (105) according to the obtained pose relationship, obtaining a virtual scene view of the endoscope in the virtual scene (106) and performing Gaussian function decay on an edge of a real-time endoscope image, and fusing same with the virtual scene view of the endoscope, so as to realize layered rendering of the scene (107). The method and system accelerate the rendering speed of an image, and improve the navigation accuracy.La présente invention concerne un procédé et un système de navigation chirurgicale endoscopique. Le procédé comprend : la lecture de données dimagerie médicale multimodale (101) lutilisation de nimporte quelles données dimage dans les données dimagerie médicale multimodale comme image de référence, et lutilisation dautres données dimagerie comme image à enregistrer, afin dexécuter une correspondance affine complète dimagerie (102) lexécution dune reconstruction de scène et dun rendu hybride sur les données dimagerie après la correspondance affine complète dimagerie, et lobtention dune scène virtuelle (103) lachèvement dun enregistrement entre une image de tomodensitométrie et une pose dun patient à laide dun procédé denregistremen
