Avaliação de estratégias para cálculo do contraste local e seus efeitos na estabilidade de pontos de interesse sob mudanças de iluminação

Abstract

Invariance to changes in scene illumination is a desirable feature in computer vision, but its accomplishment is still a challenging task even for modern vision systems. This work presents a novel way to detect keypoints in scale-space by applying the luminance ratio concept for local contrast calculation, first introduced by Wallach’s Lightness Constancy Theory. The hypothesis is that an operator based on luminance ratios may detect more stable keypoints with respect to illumination changes than the usual difference-based operators. In order to test this hypothesis, the SIFT detector was adapted to the proposed approach with minimum implementation efforts. Experimental results show that the proposed approach by luminance ratios has indeed provided more stable keypoint detection than the original SIFT detector approach based on luminance differences, specially in lower illumination conditions. Existing applications using the SIFT detector can be easily adapted to use the luminance ratio concept for improved performance in environments with uncontrolled lighting conditions.