Robust correspondence recognition for computer vision

Abstract

Summary. In this paper we introduce a new robust framework suitable for the task of finding correspondences in computer vision. This task lies in the heart of many problems like stereovision, 3D model reconstruction, image stitching, camera autocalibration, recognition and image retrieval and a host of others. If the problem domain is general enough, the correspondence problem can seldom employ any well-structured prior knowledge. This leads to tasks that have to find maximum cardinality solutions satisfying some weak optimality condition and a set of constraints. To avoid artifacts, robustness is required to cope with decision under occlusion, uncertainty or insufficiency of data and local violations of prior model. The proposed framework is based on a robust modification of graph-theoretic notion known as digraph kernel.

Efficient Sampling of Disparity Space for Fast And Accurate Matching

“…Note that s / ∈ X (s) and that q ∈ X (s) ⇔ s ∈ X (q). The relation q ∈ X (s) represents the relation of occlusion [17]. We say element q ∈ T is a competitor to s ∈ T if q ∈ X (s) and it has better image similarity, i.e.…”

Performance evaluation of building detection and digital surface model extraction algorithms: Outcomes of the PRRS 2008 Algorithm Performance Contest

“…The normalized cross-correlation (MNCC) [16] was used for computing image similarity in a 5×5 neighborhood. This stage was followed by Confidently Stable Matching (CSM) [24] which performed pixel-wise selection from the grown components in a process of their mutual competition. The matching used a modified inhibition zone as described in [4].…”

A Robust Graph-Based Method for The General Correspondence Problem Demonstrated on Image Stitching

“…Proposition 3 (Weak Optimality I [14]). Let D be an interval-oriented graph and let M be a SIS which is also a maximal independent vertex set of D. If there is a sequence of reductions by Rule 1 such that each sink s at which reduction occurs satisfies…”

“…Reduction Rule 1 can be used as an algorithm because of the following proposition: See [14] for a detailed description of an algorithm based on Rule 1. Note that vertices of the largest similarity e(p) tend to be sinks but only if their similarity differ enough from their neighbors.…”

“…This algorithm has been described in [13,14,16] and its new fast variant will be described in Sec. 4.…”