Abstract. In this work we present an efficient approach for content-based retrieval of medical images using the histogram of brightness. The main restriction of histograms is their ambiguity to separate images. That is, histograms express the global brightness distribution of images, but are not invariant to transformations on brightness and contrast performed on the images. In this work we will present an extension of the metric histograms, which overcomes such deficiencies, making the similarity comparison of images much more robust.
[ project home page | contact authors ]
Examples of medical images and their histograms:
 |
| Figure 1. a) Original medical image, b) The same one with brightness increased and c) with auto-levels (contrast increased) |