This paper compares three robust feature detection methods, they are, Scale Invariant Feature Transform (SIFT), Principal Component Analysis (PCA) -SIFT and Speeded Up Robust Features (SURF). Lowe presented SIFT [1], which was successfully used in recognition, stitching and many other applications because of its robustness. Yan Ke [2] gave a change of SIFT by using PCA to normalize the gradient patch instead of histogram. H. Bay [3] presented a faster method for SURF, which used Fast-Hessian detector. The performance of the three methods is compared for scale changes, rotation , blur, illumination changes and affine transformations, all of which uses repeatability as an evaluation measurement. Additionally, RANSAC is used to reject the inconsistent matches [4]. SIFT presents its stability in most situation except rotation and illumination changes. SURF is the fastest one with good performance as the same as SIFT, PCA-SIFT shows its advantages in rotation, blur and illumination changes.
展开▼
机译：本文比较了三种鲁棒的特征检测方法，分别是尺度不变特征变换（SIFT），主成分分析（PCA）-SIFT和加速鲁棒特征（SURF）。 Lowe提出了SIFT [1]，由于其坚固性，它已成功用于识别，缝合和许多其他应用中。 Yan Ke [2]通过使用PCA而不是直方图归一化梯度贴片来改变SIFT。 H. Bay [3]提出了一种更快的SURF方法，它使用Fast-Hessian检测器。比较了这三种方法在缩放比例变化，旋转，模糊，照明变化和仿射变换方面的性能，所有这些方法均使用可重复性作为评估指标。另外，使用RANSAC拒绝不一致的匹配[4]。除旋转和照明变化外，SIFT在大多数情况下均表现出稳定性。 SURF与SIFT一样，是性能最快的产品，PCA-SIFT在旋转，模糊和照明变化方面显示出优势。
展开▼
