Our goal is to address the problem of mapping application specific requirements on image similarity metrics to the plethora of existing image similarity computations. The work is motivated by the fact that there is a need to choose a similarity metric according to application requirements.
Many conclusions in the biomedical field are based on comparing measured and reference observations. One of the fundamental components of these comparisons is a measure of similarity. Visual inspection and human perception of similarity play a prominent role in biomedical research. However, advances in imaging and the corresponding growth of image data lead to an increasingly high demand for the automation of visual comparisons, and for a transition from expert-applied image similarity to computer-applied image similarity. The overwhelming data volumes represent our major motivation for the automation of visual comparisons and for building reference implementations of measures of image similarity.
Figure 1 illustrates the process of creating similarity metric signatures for a set of images, image descriptors and comparison measures. Given the database of similarity metric signatures, a user can form a query to the database specifying the sensitivity of a suitable image comparison to various position, shape, intensity, and texture changes. The recommender can match the query with the closest signature and retrieve the recommended metric.
Our objective is to provide support for finding similarity metrics that closely match application requirements on image comparisons.
We approached the problem by designing a theoretical and experimental framework for creating sensitivity signatures of similarity metrics. In this paper, we outline the classifications of image similarity metrics found in the literature, the space of application parameters and requirements, derivations of similarity dependencies on application parameters, and experimentally obtained sensitivity signatures of similarity metrics using image simulations. These sensitivity signatures provide a way for users to query a reference database of sensitivity signatures and retrieve a recommendation for an image similarity metric. We illustrate the use of the prototype recommendation system by considering spectral calibration and spatial registration application requirements.
The paper: Peter Bajcsy, Joe Chalfoun, and Mary Brady, “Toward a Recommendation System for Image Similarity Metrics,” Proceedings of the 2nd IASTED International Symposia Imaging and Signal Processing in Health Care and Technology (ISPHT 2012), pp. 94-100, May 14 – 16, 2012 Baltimore, USA; DOI: 10.2316/P.2012.771-014. (download pdf)
Peter Bajcsy, Joe Chalfoun, and Mary Brady, “Toward a Recommendation System for Image Similarity Metrics,” Proceedings of the 2nd IASTED International Symposia Imaging and Signal Processing in Health Care and Technology (ISPHT 2012), pp. 94-100, May 14 – 16, 2012 Baltimore, USA; DOI: 10.2316/P.2012.771-014. (download pdf)