Multi-view Oblique Aerial Image Sparse Matching

Compared to traditional approximately vertical aerial photogrammetry, multi-view oblique aerial photogrammetry takes photos from multiple directions simultaneously and gets a high degree of overlapping images and side texture so as to effectively compensate for the image occlusion. Multi-view oblique photogrammetry provides a new means for 3D reconstruction and earth observation and becomes an important trend in the survey area. The world's current mainstream multi-view cameras are summarized and characteristics of oblique images are analyzed. Epipolar constraint based on fundamental matrix and exterior orientation elements are compared. The applicability of area-based matching algorithm for oblique images is explored. Experiments show that area-based matching is no longer applicable to multi-view oblique images, but geometric distortion correction or compensation can make area-based matching continue to apply.