An Accurate and Efficient 3D Reconstruction Method for Wireless Capsule Endoscopy Based on Harris Corner Detection and Fixed Template Matching

Abstract

To overcome the limitation of wireless capsule endoscopy (WCE) in providing only two‑dimensional intestinal images—which constrains accurate diagnosis of intestinal lesions—this study proposes an accurate and efficient three‑dimensional detection method. The approach applies 3D reconstruction to WCE images by computing corresponding points across successive frames and estimating the capsule’s position and orientation. A critical step involves the selection of highly specific feature points and the determination of their correspondences. Accordingly, this paper combines a Harris corner‑based feature extraction algorithm with a fixed matching template algorithm to achieve robust feature point matching in WCE intestinal images, substantially improving both matching efficiency and robustness. Using the obtained feature matching points, an intestinal model is reconstructed via 3D reconstruction. Experimental results demonstrate that the proposed method offers advantages in efficiency and reconstruction quality compared with conventional 3D reconstruction approaches, indicating its potential to enhance WCE‑based lesion localization and diagnostic accuracy.