EVALUATION OF DEVELOPED SOFTWARE FOR CALCULATING HISTOGRAM INTERSECTION DISTANCES

Resumo

This paper presents the results obtained using a developed software implementing the histogram intersection metric to compare images of three different objects under different situations. The images were obtained from the website of Amsterdam Library of Object Images and were divided into two sets: changes on viewpoints of the objects and changes on illumination colors. The histogram intersection distances varied from 0.723 to 1.000, when occurs changes in the viewpoints, and from 0.505 to 1.000, when happens changes in illumination colors. These results are expected since the rotation of objects with symmetric shape slightly modifies the histogram distribution. On the other hand, in the case of illumination color changes, bigger variations are predictable since it directly alters the values of pixels.