Multi-image Matching Without Similarity Measurement

This paper deals with automatic reconstruction of a digital multi-ray photogrammetric network, which comprises multiple nodes, each node indicating one object point and its corresponding multiple-imaging rays. In the automatic reconstruction, a novel multi image matching (MIM) technique without similarity measurement is proposed. The technique includes consecutive consideration of all images as reference images. In the proposed strategy, matching accuracy is enhanced by sub-pixel interest point extraction and multi-ray traingulation.