Abstract
This paper performs a comprehensive and comparative evaluation of the stateof the art local features for the task of image based 3D reconstruction. Theevaluated local features cover the recently developed ones by using powerfulmachine learning techniques and the elaborately designed handcrafted features.To obtain a comprehensive evaluation, we choose to include both float typefeatures and binary ones. Meanwhile, two kinds of datasets have been used inthis evaluation. One is a dataset of many different scene types withgroundtruth 3D points, containing images of different scenes captured at fixedpositions, for quantitative performance evaluation of different local featuresin the controlled image capturing situations. The other dataset containsInternet scale image sets of several landmarks with a lot of unrelated images,which is used for qualitative performance evaluation of different localfeatures in the free image collection situations. Our experimental results showthat binary features are competent to reconstruct scenes from controlled imagesequences with only a fraction of processing time compared to use float typefeatures. However, for the case of large scale image set with many distractingimages, float type features show a clear advantage over binary ones.