Adaptive Linear Combination-Based Contrast-Preserving Decolourization of Macroscopic Skin Images

The dermatological macro-images are colour images. The majority of the available image segmentation and feature extraction algorithms are designed for grayscale images. Hence, the conversion of dermatological images to grayscale is an important step in their automated analysis. Customized algorithms for decolourizing dermatological images are not available. The existing decolourization algorithms for natural-scene images focus only on the preservation of local contrast. Such algorithms may not ensure good accuracy of lesion segmentation on macroscopic images when intensity-based schemes are adopted. Decolourization algorithms that enable effective preservation of both gradient and intensity information are more desirable. This approach in especially, resulted in a 6–8% average DSC improvement over baselines, a 12% increase in the contrast index, and a 9% increase in the perceptual similarity score.