Contrast enhancement in wavelet domain for graph-based segmentation in medical imaging

摘要: Despite increased image quality including medical imaging, segmentation continues to represent a major bottle-neck in practical applications due noise and lack of contrast. In this paper, we present new methodology segment low contrast images. There are two stages approach, 1) enhancement stage, that uses stochastic resonance theory applied wavelet domain, is performed by utilizing the data, 2) weighting function proposed for traditional graph-based approaches. Both qualitative quantitative evaluation on publicly available databases imaging modalities reflect potential method.