Edge foci interest points

Abstract

In this paper, we describe an interest point detector using edge foci. Unlike traditional detectors that compute interest points directly from image intensities, we use normalized intensity edges and their orientations. We hypothesize that detectors based on the presence of oriented edges are more robust to non-linear lighting variations and background clutter than intensity based techniques. Specifically, we detect edge foci, which are points in the image that are roughly equidistant from edges with orientations perpendicular to the point. The scale of the interest point is defined by the distance between the edge foci and the edges. We quantify the performance of our detector using the interest point's repeatability, uniformity of spatial distribution, and the uniqueness of the resulting descriptors. Results are found using traditional datasets and new datasets with challenging non-linear lighting variations and occlusions.

WαSH: Weighted α-Shapes for Local Feature Detection

“…We demonstrate the effectiveness of our WαSH detector by measuring repeatability and matching score on the standard dataset and benchmark of [17], and by carrying out a retrieval experiment on Oxford Buildings [20], following the evaluation protocol that has been introduced in [21] and recently used to evaluate detectors in [26] and [1]. The performance of the WαSH detector depends only on the selection threshold τ .…”

“…Zitnick et al . [26] apply an oriented filter bank to the input image and detect edge foci (EF), i.e. points that are roughly equidistant from edgels with orientations perpendicular to the points.…”

“…4cd). Comparisons are performed against the Hessian-Affine, MSER, SIFT and SURF detectors, plus the recently introduced EF [26] and MFD [1] detectors, follow- ing their experimental framework. Features are extracted by the corresponding publicly available executables, using default parameters.…”

Learning to Assign Orientations to Feature Points

“…Feature points are an essential and ubiquitous tool in computer vision, and extensive research has been conducted on both detectors [3,6,22,25,27,32,48] and descriptors [2,6,22,25,32,38,43,47], including using statistical approaches [33,46]. However, the assignment of a canonical orientation, which is an important common step, has received almost no individual attention, probably since the dominant orientation of SIFT [25] is considered to give good results.…”

“…46272 vertices were obtained using SIFT orientations, and 84087 vertices with our orientations. Edge Foci feature points [48] were used in conjunction with Daisy [38] descriptors for both methods. 1 ther as these descriptors discard rotation sensitive information which can be useful when ideal orientations are given.…”