Improving Visual Matching

Many visual matching algorithms can be described in terms of the features and the inter-feature distance or metric. The most commonly used metric is the sum of squared di erences (SSD), which is valid from a maximum likelihood perspective when the real noise distribution is Gaussian. Based on real noise distributions measured from international test sets, we have found experimentally that the Gaussian noise distribution assumption is often invalid. This implies that other metrics, which have distributions closer to the real noise distribution, should be used. In this paper we considered two di erent visual matching applications: content-based retrieval in image databases and stereo matching. Towards broadening the results, we also implemented several sophisticated algorithms from the research literature. In each algorithm we compared the e cacy of the SSD metric, the SAD (sum of the absolute di erences) metric, the Cauchy metric, and the Kullback relative information. Furthermore, in ...

• Type: Conference
• Year: 2000
• Where: CVPR
• Authors: Michael S. Lew, Nicu Sebe, Thomas S. Huang
• Keywords: Computer Vision, CVPR 2000, Gaussian Noise Distribution, Maximum Likelihood Perspective, Real Noise Distribution, Real Noise Distributions, Visual Matching Algorithms