Contour-based object detection as dominant set computation

• Published in: Pattern recognition Vol. 45; no. 5; pp. 1927 - 1936
• Main Authors: Yang, Xingwei, Liu, Hairong, Latecki, Longin Jan
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.05.2012; Elsevier
• Subjects: Applied sciences; Consistency; Detection, estimation, filtering, equalization, prediction; Dominant sets; Exact sciences and technology; Fragments; Graphs; Information, signal and communications theory; Matching; Object detection; Optimization; Pattern recognition; Shape; Shape similarity; Signal and communications theory; Signal processing; Signal, noise; Similarity; Telecommunications and information theory; Shape similarity; Object detection; Dominant sets; Pattern recognition; Shape detection; Edge detection; Target detection; Pattern matching
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2011.11.010

Contour-based object detection can be formulated as a matching problem between model contour parts and image edge fragments. We propose a novel solution by treating this problem as the problem of finding dominant sets in weighted graphs. The nodes of the graph are pairs composed of model contour parts and image edge fragments, and the weights between nodes are based on shape similarity. Because of high consistency between correct correspondences, the correct matching corresponds to a dominant set of the graph. Consequently, when a dominant set is determined, it provides a selection of correct correspondences. As the proposed method is able to get all the dominant sets, we can detect multiple objects in an image in one pass. Moreover, since our approach is purely based on shape, we also determine an optimal scale of target object without a common enumeration of all possible scales. Both theoretic analysis and extensive experimental evaluation illustrate the benefits of our approach. ► We solve the shape based object detection by matching image and model segments. ► We construct a graph. Each node is a pair of model and image segments. ► The sub-dense graph (dominant sets) corresponds to a detection of objects. ► This algorithm is robust to missing edges of objects and clutter background. ► The method is able to detect multiple objects without any post-processing steps.